[clang] [llvm] [clang][OpenMP] Improve loop structure for distributed loops (pt 2) (PR #202907)
Robert Imschweiler via cfe-commits
cfe-commits at lists.llvm.org
Wed Jun 10 02:50:34 PDT 2026
https://github.com/ro-i created https://github.com/llvm/llvm-project/pull/202907
This patch complements https://github.com/llvm/llvm-project/pull/201670 for non-reduction loops and wires the existing
`kmp_sched_distr_static_chunk_sched_static_chunkone` to be used by CodeGen for these loops as well.
See the linked PR for history. I open this PR in draft mode for the moment to signal that I'm still doing analysis of different cases and their performance behavior. If anyone has interest in a specific usage pattern of `target teams distribute parallel for` loops, please let me know so that I can include it in my testing.
>From 5de640e05b7418ca4755226cc22d95b6ab2bfe88 Mon Sep 17 00:00:00 2001
From: Robert Imschweiler <robert.imschweiler at amd.com>
Date: Thu, 4 Jun 2026 09:20:39 -0500
Subject: [PATCH] [clang][OpenMP] Improve loop structure for distributed loops
(pt 2)
This patch complements https://github.com/llvm/llvm-project/pull/201670
for non-reduction loops and wires the existing
`kmp_sched_distr_static_chunk_sched_static_chunkone` to be used by
CodeGen for these loops as well.
---
clang/lib/CodeGen/CGStmtOpenMP.cpp | 1 -
.../test/OpenMP/amdgcn_target_device_vla.cpp | 126 +-
.../amdgpu_target_with_aligned_attribute.c | 106 +-
.../metadirective_device_arch_codegen.cpp | 11 +-
clang/test/OpenMP/nvptx_SPMD_codegen.cpp | 5994 +++++++----------
...stribute_parallel_generic_mode_codegen.cpp | 238 +-
..._teams_distribute_parallel_for_codegen.cpp | 1442 +---
...s_distribute_parallel_for_simd_codegen.cpp | 766 +--
...vptx_target_teams_generic_loop_codegen.cpp | 1472 +---
...eams_generic_loop_generic_mode_codegen.cpp | 102 +-
...ams_generic_loop_codegen_as_distribute.cpp | 77 +-
...s_generic_loop_codegen_as_parallel_for.cpp | 480 +-
.../test/offloading/gpupgo/pgo_atomic_teams.c | 6 +-
13 files changed, 3487 insertions(+), 7334 deletions(-)
diff --git a/clang/lib/CodeGen/CGStmtOpenMP.cpp b/clang/lib/CodeGen/CGStmtOpenMP.cpp
index 314f4e14dd1d2..f99b9111fa6f8 100644
--- a/clang/lib/CodeGen/CGStmtOpenMP.cpp
+++ b/clang/lib/CodeGen/CGStmtOpenMP.cpp
@@ -60,7 +60,6 @@ static bool canEmitGPUFusedDistSchedule(const CodeGenModule &CGM,
// need more analysis for maximum profit.
return CGM.getLangOpts().OpenMPIsTargetDevice && CGM.getTriple().isGPU() &&
isOpenMPLoopBoundSharingDirective(DKind) &&
- S.hasClausesOfKind<OMPReductionClause>() &&
!S.getSingleClause<OMPDistScheduleClause>() &&
!S.getSingleClause<OMPScheduleClause>() &&
!S.getSingleClause<OMPOrderedClause>();
diff --git a/clang/test/OpenMP/amdgcn_target_device_vla.cpp b/clang/test/OpenMP/amdgcn_target_device_vla.cpp
index 5064c114c0863..323725f215f79 100644
--- a/clang/test/OpenMP/amdgcn_target_device_vla.cpp
+++ b/clang/test/OpenMP/amdgcn_target_device_vla.cpp
@@ -276,92 +276,32 @@ int main() {
// CHECK-NEXT: store i32 1, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
// CHECK-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
// CHECK-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
-// CHECK-NEXT: call void @__kmpc_distribute_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB2:[0-9]+]] to ptr), i32 [[TMP7]], i32 91, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_COMB_LB_ASCAST]], ptr [[DOTOMP_COMB_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
+// CHECK-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1_ASCAST]], align 4
-// CHECK-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]]
-// CHECK-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK: cond.true:
-// CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1_ASCAST]], align 4
-// CHECK-NEXT: br label [[COND_END:%.*]]
-// CHECK: cond.false:
-// CHECK-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-NEXT: br label [[COND_END]]
-// CHECK: cond.end:
-// CHECK-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ]
-// CHECK-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// CHECK-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV_ASCAST]], align 4
-// CHECK-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK: omp.inner.for.cond:
-// CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4
-// CHECK-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1_ASCAST]], align 4
-// CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], 1
-// CHECK-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP13]], [[ADD]]
-// CHECK-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK: omp.inner.for.body:
-// CHECK-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// CHECK-NEXT: [[TMP16:%.*]] = zext i32 [[TMP15]] to i64
-// CHECK-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64
-// CHECK-NEXT: [[TMP19:%.*]] = load i32, ptr [[M_ADDR_ASCAST]], align 4
-// CHECK-NEXT: store i32 [[TMP19]], ptr addrspace(5) [[M_CASTED]], align 4
-// CHECK-NEXT: [[TMP20:%.*]] = load i64, ptr addrspace(5) [[M_CASTED]], align 8
-// CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 0
-// CHECK-NEXT: [[TMP22:%.*]] = inttoptr i64 [[TMP16]] to ptr
-// CHECK-NEXT: store ptr [[TMP22]], ptr [[TMP21]], align 8
-// CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 1
-// CHECK-NEXT: [[TMP24:%.*]] = inttoptr i64 [[TMP18]] to ptr
-// CHECK-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 8
-// CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 2
-// CHECK-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP20]] to ptr
-// CHECK-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8
-// CHECK-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 3
-// CHECK-NEXT: [[TMP28:%.*]] = inttoptr i64 [[TMP0]] to ptr
-// CHECK-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 8
-// CHECK-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 4
-// CHECK-NEXT: store ptr [[TMP1]], ptr [[TMP29]], align 8
-// CHECK-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// CHECK-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
-// CHECK-NEXT: call void @__kmpc_parallel_60(ptr addrspacecast (ptr addrspace(1) @[[GLOB1]] to ptr), i32 [[TMP31]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo2v_l30_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 5, i32 0)
-// CHECK-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK: omp.inner.for.inc:
-// CHECK-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4
-// CHECK-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
-// CHECK-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP32]], [[TMP33]]
-// CHECK-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV_ASCAST]], align 4
-// CHECK-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// CHECK-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
-// CHECK-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP34]], [[TMP35]]
-// CHECK-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// CHECK-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
-// CHECK-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP36]], [[TMP37]]
-// CHECK-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1_ASCAST]], align 4
-// CHECK-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP38]], [[TMP39]]
-// CHECK-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]]
-// CHECK: cond.true10:
-// CHECK-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1_ASCAST]], align 4
-// CHECK-NEXT: br label [[COND_END12:%.*]]
-// CHECK: cond.false11:
-// CHECK-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-NEXT: br label [[COND_END12]]
-// CHECK: cond.end12:
-// CHECK-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP40]], [[COND_TRUE10]] ], [ [[TMP41]], [[COND_FALSE11]] ]
-// CHECK-NEXT: store i32 [[COND13]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-NEXT: [[TMP42:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// CHECK-NEXT: store i32 [[TMP42]], ptr [[DOTOMP_IV_ASCAST]], align 4
-// CHECK-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK: omp.inner.for.end:
+// CHECK-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
+// CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr [[M_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[TMP10]], ptr addrspace(5) [[M_CASTED]], align 4
+// CHECK-NEXT: [[TMP11:%.*]] = load i64, ptr addrspace(5) [[M_CASTED]], align 8
+// CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 0
+// CHECK-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP7]] to ptr
+// CHECK-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
+// CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 1
+// CHECK-NEXT: [[TMP15:%.*]] = inttoptr i64 [[TMP9]] to ptr
+// CHECK-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8
+// CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 2
+// CHECK-NEXT: [[TMP17:%.*]] = inttoptr i64 [[TMP11]] to ptr
+// CHECK-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8
+// CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 3
+// CHECK-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP0]] to ptr
+// CHECK-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8
+// CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 4
+// CHECK-NEXT: store ptr [[TMP1]], ptr [[TMP20]], align 8
+// CHECK-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
+// CHECK-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
+// CHECK-NEXT: call void @__kmpc_parallel_60(ptr addrspacecast (ptr addrspace(1) @[[GLOB1]] to ptr), i32 [[TMP22]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo2v_l30_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 5, i32 0)
// CHECK-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK: omp.loop.exit:
-// CHECK-NEXT: [[TMP43:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// CHECK-NEXT: [[TMP44:%.*]] = load i32, ptr [[TMP43]], align 4
-// CHECK-NEXT: call void @__kmpc_distribute_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP44]])
// CHECK-NEXT: br label [[OMP_PRECOND_END]]
// CHECK: omp.precond.end:
// CHECK-NEXT: ret void
@@ -448,7 +388,7 @@ int main() {
// CHECK-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
// CHECK-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
-// CHECK-NEXT: call void @__kmpc_for_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB3:[0-9]+]] to ptr), i32 [[TMP9]], i32 33, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
+// CHECK-NEXT: call void @__kmpc_for_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB2:[0-9]+]] to ptr), i32 [[TMP9]], i32 93, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
// CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_LB_ASCAST]], align 4
// CHECK-NEXT: store i32 [[TMP10]], ptr [[DOTOMP_IV_ASCAST]], align 4
// CHECK-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -537,7 +477,7 @@ int main() {
// CHECK: omp.loop.exit:
// CHECK-NEXT: [[TMP34:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP35:%.*]] = load i32, ptr [[TMP34]], align 4
-// CHECK-NEXT: call void @__kmpc_for_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB3]] to ptr), i32 [[TMP35]])
+// CHECK-NEXT: call void @__kmpc_for_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP35]])
// CHECK-NEXT: br label [[OMP_PRECOND_END]]
// CHECK: omp.precond.end:
// CHECK-NEXT: ret void
@@ -648,7 +588,7 @@ int main() {
// CHECK-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
// CHECK-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
-// CHECK-NEXT: call void @__kmpc_distribute_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP7]], i32 92, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
+// CHECK-NEXT: call void @__kmpc_distribute_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB3:[0-9]+]] to ptr), i32 [[TMP7]], i32 92, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
// CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB_ASCAST]], align 4
// CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1_ASCAST]], align 4
// CHECK-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]]
@@ -738,7 +678,7 @@ int main() {
// CHECK: omp.loop.exit:
// CHECK-NEXT: [[TMP38:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP39:%.*]] = load i32, ptr [[TMP38]], align 4
-// CHECK-NEXT: call void @__kmpc_distribute_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP39]])
+// CHECK-NEXT: call void @__kmpc_distribute_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB3]] to ptr), i32 [[TMP39]])
// CHECK-NEXT: br label [[OMP_PRECOND_END]]
// CHECK: omp.precond.end:
// CHECK-NEXT: call void @__kmpc_free_shared(ptr [[N]], i64 4)
@@ -805,7 +745,7 @@ int main() {
// CHECK-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
// CHECK-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK-NEXT: call void @__kmpc_for_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB3]] to ptr), i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
+// CHECK-NEXT: call void @__kmpc_for_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
// CHECK-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK: omp.dispatch.cond:
// CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB_ASCAST]], align 4
@@ -867,7 +807,7 @@ int main() {
// CHECK: omp.dispatch.end:
// CHECK-NEXT: [[TMP26:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP27:%.*]] = load i32, ptr [[TMP26]], align 4
-// CHECK-NEXT: call void @__kmpc_for_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB3]] to ptr), i32 [[TMP27]])
+// CHECK-NEXT: call void @__kmpc_for_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP27]])
// CHECK-NEXT: br label [[OMP_PRECOND_END]]
// CHECK: omp.precond.end:
// CHECK-NEXT: ret void
@@ -1016,7 +956,7 @@ int main() {
// CHECK-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
// CHECK-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK-NEXT: call void @__kmpc_distribute_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
+// CHECK-NEXT: call void @__kmpc_distribute_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB3]] to ptr), i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
// CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB_ASCAST]], align 4
// CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2_ASCAST]], align 4
// CHECK-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
@@ -1105,7 +1045,7 @@ int main() {
// CHECK: omp.loop.exit:
// CHECK-NEXT: [[TMP39:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP40:%.*]] = load i32, ptr [[TMP39]], align 4
-// CHECK-NEXT: call void @__kmpc_distribute_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP40]])
+// CHECK-NEXT: call void @__kmpc_distribute_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB3]] to ptr), i32 [[TMP40]])
// CHECK-NEXT: br label [[OMP_PRECOND_END]]
// CHECK: omp.precond.end:
// CHECK-NEXT: call void @__kmpc_free_shared(ptr [[N1]], i64 4)
@@ -1172,7 +1112,7 @@ int main() {
// CHECK-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
// CHECK-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK-NEXT: call void @__kmpc_for_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB3]] to ptr), i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
+// CHECK-NEXT: call void @__kmpc_for_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
// CHECK-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK: omp.dispatch.cond:
// CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB_ASCAST]], align 4
@@ -1234,7 +1174,7 @@ int main() {
// CHECK: omp.dispatch.end:
// CHECK-NEXT: [[TMP26:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP27:%.*]] = load i32, ptr [[TMP26]], align 4
-// CHECK-NEXT: call void @__kmpc_for_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB3]] to ptr), i32 [[TMP27]])
+// CHECK-NEXT: call void @__kmpc_for_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP27]])
// CHECK-NEXT: br label [[OMP_PRECOND_END]]
// CHECK: omp.precond.end:
// CHECK-NEXT: ret void
diff --git a/clang/test/OpenMP/amdgpu_target_with_aligned_attribute.c b/clang/test/OpenMP/amdgpu_target_with_aligned_attribute.c
index 3f9d2225c7de1..09dcddf3e6c6f 100644
--- a/clang/test/OpenMP/amdgpu_target_with_aligned_attribute.c
+++ b/clang/test/OpenMP/amdgpu_target_with_aligned_attribute.c
@@ -109,90 +109,30 @@ void write_to_aligned_array(int *a, int N) {
// CHECK-AMD-NEXT: store i32 1, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
// CHECK-AMD-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
// CHECK-AMD-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-AMD-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// CHECK-AMD-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
-// CHECK-AMD-NEXT: call void @__kmpc_distribute_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB2:[0-9]+]] to ptr), i32 [[TMP5]], i32 91, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_COMB_LB_ASCAST]], ptr [[DOTOMP_COMB_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-AMD-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
+// CHECK-AMD-NEXT: [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-AMD-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP6]], [[TMP7]]
-// CHECK-AMD-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK-AMD: cond.true:
-// CHECK-AMD-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1_ASCAST]], align 4
-// CHECK-AMD-NEXT: br label [[COND_END:%.*]]
-// CHECK-AMD: cond.false:
-// CHECK-AMD-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-AMD-NEXT: br label [[COND_END]]
-// CHECK-AMD: cond.end:
-// CHECK-AMD-NEXT: [[COND:%.*]] = phi i32 [ [[TMP8]], [[COND_TRUE]] ], [ [[TMP9]], [[COND_FALSE]] ]
-// CHECK-AMD-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// CHECK-AMD-NEXT: store i32 [[TMP10]], ptr [[DOTOMP_IV_ASCAST]], align 4
-// CHECK-AMD-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-AMD: omp.inner.for.cond:
-// CHECK-AMD-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], 1
-// CHECK-AMD-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP11]], [[ADD]]
-// CHECK-AMD-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-AMD: omp.inner.for.body:
-// CHECK-AMD-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[TMP14:%.*]] = zext i32 [[TMP13]] to i64
-// CHECK-AMD-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[TMP16:%.*]] = zext i32 [[TMP15]] to i64
-// CHECK-AMD-NEXT: [[TMP17:%.*]] = load i32, ptr [[N_ADDR_ASCAST]], align 4
-// CHECK-AMD-NEXT: store i32 [[TMP17]], ptr addrspace(5) [[N_CASTED]], align 4
-// CHECK-AMD-NEXT: [[TMP18:%.*]] = load i64, ptr addrspace(5) [[N_CASTED]], align 8
-// CHECK-AMD-NEXT: [[TMP19:%.*]] = load ptr, ptr [[APTR_ADDR_ASCAST]], align 8
-// CHECK-AMD-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 0
-// CHECK-AMD-NEXT: [[TMP21:%.*]] = inttoptr i64 [[TMP14]] to ptr
-// CHECK-AMD-NEXT: store ptr [[TMP21]], ptr [[TMP20]], align 8
-// CHECK-AMD-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 1
-// CHECK-AMD-NEXT: [[TMP23:%.*]] = inttoptr i64 [[TMP16]] to ptr
-// CHECK-AMD-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 8
-// CHECK-AMD-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 2
-// CHECK-AMD-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP18]] to ptr
-// CHECK-AMD-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 8
-// CHECK-AMD-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 3
-// CHECK-AMD-NEXT: store ptr [[TMP19]], ptr [[TMP26]], align 8
-// CHECK-AMD-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// CHECK-AMD-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
-// CHECK-AMD-NEXT: call void @__kmpc_parallel_60(ptr addrspacecast (ptr addrspace(1) @[[GLOB1]] to ptr), i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_write_to_aligned_array_l14_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 4, i32 0)
-// CHECK-AMD-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-AMD: omp.inner.for.inc:
-// CHECK-AMD-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP29]], [[TMP30]]
-// CHECK-AMD-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
-// CHECK-AMD-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
-// CHECK-AMD-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP35]], [[TMP36]]
-// CHECK-AMD-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]]
-// CHECK-AMD: cond.true10:
-// CHECK-AMD-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1_ASCAST]], align 4
-// CHECK-AMD-NEXT: br label [[COND_END12:%.*]]
-// CHECK-AMD: cond.false11:
-// CHECK-AMD-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-AMD-NEXT: br label [[COND_END12]]
-// CHECK-AMD: cond.end12:
-// CHECK-AMD-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP37]], [[COND_TRUE10]] ], [ [[TMP38]], [[COND_FALSE11]] ]
-// CHECK-AMD-NEXT: store i32 [[COND13]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// CHECK-AMD-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// CHECK-AMD-NEXT: store i32 [[TMP39]], ptr [[DOTOMP_IV_ASCAST]], align 4
-// CHECK-AMD-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK-AMD: omp.inner.for.end:
+// CHECK-AMD-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
+// CHECK-AMD-NEXT: [[TMP8:%.*]] = load i32, ptr [[N_ADDR_ASCAST]], align 4
+// CHECK-AMD-NEXT: store i32 [[TMP8]], ptr addrspace(5) [[N_CASTED]], align 4
+// CHECK-AMD-NEXT: [[TMP9:%.*]] = load i64, ptr addrspace(5) [[N_CASTED]], align 8
+// CHECK-AMD-NEXT: [[TMP10:%.*]] = load ptr, ptr [[APTR_ADDR_ASCAST]], align 8
+// CHECK-AMD-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 0
+// CHECK-AMD-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP5]] to ptr
+// CHECK-AMD-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8
+// CHECK-AMD-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 1
+// CHECK-AMD-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP7]] to ptr
+// CHECK-AMD-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8
+// CHECK-AMD-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 2
+// CHECK-AMD-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP9]] to ptr
+// CHECK-AMD-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8
+// CHECK-AMD-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 3
+// CHECK-AMD-NEXT: store ptr [[TMP10]], ptr [[TMP17]], align 8
+// CHECK-AMD-NEXT: [[TMP18:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
+// CHECK-AMD-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP18]], align 4
+// CHECK-AMD-NEXT: call void @__kmpc_parallel_60(ptr addrspacecast (ptr addrspace(1) @[[GLOB1]] to ptr), i32 [[TMP19]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_write_to_aligned_array_l14_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 4, i32 0)
// CHECK-AMD-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-AMD: omp.loop.exit:
-// CHECK-AMD-NEXT: [[TMP40:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// CHECK-AMD-NEXT: [[TMP41:%.*]] = load i32, ptr [[TMP40]], align 4
-// CHECK-AMD-NEXT: call void @__kmpc_distribute_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP41]])
// CHECK-AMD-NEXT: br label [[OMP_PRECOND_END]]
// CHECK-AMD: omp.precond.end:
// CHECK-AMD-NEXT: ret void
@@ -264,7 +204,7 @@ void write_to_aligned_array(int *a, int N) {
// CHECK-AMD-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
// CHECK-AMD-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// CHECK-AMD-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
-// CHECK-AMD-NEXT: call void @__kmpc_for_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB3:[0-9]+]] to ptr), i32 [[TMP7]], i32 33, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
+// CHECK-AMD-NEXT: call void @__kmpc_for_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB2:[0-9]+]] to ptr), i32 [[TMP7]], i32 93, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
// CHECK-AMD-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB_ASCAST]], align 4
// CHECK-AMD-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_IV_ASCAST]], align 4
// CHECK-AMD-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -299,7 +239,7 @@ void write_to_aligned_array(int *a, int N) {
// CHECK-AMD: omp.loop.exit:
// CHECK-AMD-NEXT: [[TMP17:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// CHECK-AMD-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP17]], align 4
-// CHECK-AMD-NEXT: call void @__kmpc_for_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB3]] to ptr), i32 [[TMP18]])
+// CHECK-AMD-NEXT: call void @__kmpc_for_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP18]])
// CHECK-AMD-NEXT: br label [[OMP_PRECOND_END]]
// CHECK-AMD: omp.precond.end:
// CHECK-AMD-NEXT: ret void
diff --git a/clang/test/OpenMP/metadirective_device_arch_codegen.cpp b/clang/test/OpenMP/metadirective_device_arch_codegen.cpp
index 33953b4efa6c2..dd11bc879c3f5 100644
--- a/clang/test/OpenMP/metadirective_device_arch_codegen.cpp
+++ b/clang/test/OpenMP/metadirective_device_arch_codegen.cpp
@@ -53,15 +53,12 @@ int metadirective1() {
// CHECK: ret void
-// CHECK: define internal {{(spir_func )?}}void @[[METADIRECTIVE]]_omp_outlined
-// CHECK: entry:
-// CHECK: call{{.*}} void @__kmpc_distribute_static_init
-// CHECK: omp.loop.exit:
-// CHECK: call{{.*}} void @__kmpc_distribute_static_fini
-
+// On GPU/SPIR targets the combined `teams distribute parallel for` lowering
+// emits a single fused for_static_init (schedule constant 93) in the inner
+// outlined region; no outer distribute_static_init is emitted.
// CHECK: define internal {{(spir_func )?}}void @[[METADIRECTIVE]]_omp_outlined_omp_outlined
// CHECK: entry:
-// CHECK: call{{.*}} void @__kmpc_for_static_init_4
+// CHECK: call{{.*}} void @__kmpc_for_static_init_4{{.*}}, i32 93,
// CHECK: omp.inner.for.body:
// CHECK: store atomic {{.*}} monotonic
// CHECK: omp.loop.exit:
diff --git a/clang/test/OpenMP/nvptx_SPMD_codegen.cpp b/clang/test/OpenMP/nvptx_SPMD_codegen.cpp
index 220fef20e9db7..8f0af742441e3 100644
--- a/clang/test/OpenMP/nvptx_SPMD_codegen.cpp
+++ b/clang/test/OpenMP/nvptx_SPMD_codegen.cpp
@@ -316,8 +316,6 @@ int a;
// CHECK-64-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK-64-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK-64-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [3 x ptr], align 8
-// CHECK-64-NEXT: [[DOTCAPTURE_EXPR__CASTED15:%.*]] = alloca i64, align 8
-// CHECK-64-NEXT: [[CAPTURED_VARS_ADDRS17:%.*]] = alloca [3 x ptr], align 8
// CHECK-64-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK-64-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
@@ -326,150 +324,35 @@ int a;
// CHECK-64-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK-64-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
-// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK-64: cond.true:
-// CHECK-64-NEXT: br label [[COND_END:%.*]]
-// CHECK-64: cond.false:
-// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: br label [[COND_END]]
-// CHECK-64: cond.end:
-// CHECK-64-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
-// CHECK-64-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
-// CHECK-64-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP5]], 0
-// CHECK-64-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
-// CHECK-64: omp_if.then:
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66:![0-9]+]]
-// CHECK-64-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP6]], 10
-// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[LOADEDV2:%.*]] = icmp ne i8 [[TMP11]], 0
-// CHECK-64-NEXT: [[STOREDV:%.*]] = zext i1 [[LOADEDV2]] to i8
-// CHECK-64-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK-64-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP8]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK-64-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK-64-NEXT: [[TMP18:%.*]] = inttoptr i64 [[TMP12]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP18]], ptr [[TMP17]], align 8, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[LOADEDV3:%.*]] = icmp ne i8 [[TMP19]], 0
-// CHECK-64-NEXT: [[TMP20:%.*]] = zext i1 [[LOADEDV3]] to i32
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 [[TMP20]], i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l15_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3, i32 0), !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP23]], [[TMP24]]
-// CHECK-64-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP25]], [[TMP26]]
-// CHECK-64-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP27]], 9
-// CHECK-64-NEXT: br i1 [[CMP6]], label [[COND_TRUE7:%.*]], label [[COND_FALSE8:%.*]]
-// CHECK-64: cond.true7:
-// CHECK-64-NEXT: br label [[COND_END9:%.*]]
-// CHECK-64: cond.false8:
-// CHECK-64-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: br label [[COND_END9]]
-// CHECK-64: cond.end9:
-// CHECK-64-NEXT: [[COND10:%.*]] = phi i32 [ 9, [[COND_TRUE7]] ], [ [[TMP28]], [[COND_FALSE8]] ]
-// CHECK-64-NEXT: store i32 [[COND10]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: store i32 [[TMP29]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP67:![0-9]+]]
-// CHECK-64: omp.inner.for.end:
-// CHECK-64-NEXT: br label [[OMP_IF_END:%.*]]
-// CHECK-64: omp_if.else:
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND11:%.*]]
-// CHECK-64: omp.inner.for.cond11:
-// CHECK-64-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: [[CMP12:%.*]] = icmp slt i32 [[TMP30]], 10
-// CHECK-64-NEXT: br i1 [[CMP12]], label [[OMP_INNER_FOR_BODY13:%.*]], label [[OMP_INNER_FOR_END28:%.*]]
-// CHECK-64: omp.inner.for.body13:
-// CHECK-64-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: [[TMP32:%.*]] = zext i32 [[TMP31]] to i64
-// CHECK-64-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP34:%.*]] = zext i32 [[TMP33]] to i64
-// CHECK-64-NEXT: [[TMP35:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
-// CHECK-64-NEXT: [[LOADEDV14:%.*]] = icmp ne i8 [[TMP35]], 0
-// CHECK-64-NEXT: [[STOREDV16:%.*]] = zext i1 [[LOADEDV14]] to i8
-// CHECK-64-NEXT: store i8 [[STOREDV16]], ptr [[DOTCAPTURE_EXPR__CASTED15]], align 1
-// CHECK-64-NEXT: [[TMP36:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED15]], align 8
-// CHECK-64-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS17]], i64 0, i64 0
-// CHECK-64-NEXT: [[TMP38:%.*]] = inttoptr i64 [[TMP32]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP38]], ptr [[TMP37]], align 8
-// CHECK-64-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS17]], i64 0, i64 1
-// CHECK-64-NEXT: [[TMP40:%.*]] = inttoptr i64 [[TMP34]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP40]], ptr [[TMP39]], align 8
-// CHECK-64-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS17]], i64 0, i64 2
-// CHECK-64-NEXT: [[TMP42:%.*]] = inttoptr i64 [[TMP36]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP42]], ptr [[TMP41]], align 8
-// CHECK-64-NEXT: [[TMP43:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
-// CHECK-64-NEXT: [[LOADEDV18:%.*]] = icmp ne i8 [[TMP43]], 0
-// CHECK-64-NEXT: [[TMP44:%.*]] = zext i1 [[LOADEDV18]] to i32
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 [[TMP44]], i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l15_omp_outlined_omp_outlined1, ptr null, ptr [[CAPTURED_VARS_ADDRS17]], i64 3, i32 0)
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC19:%.*]]
-// CHECK-64: omp.inner.for.inc19:
-// CHECK-64-NEXT: [[TMP45:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: [[TMP46:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-64-NEXT: [[ADD20:%.*]] = add nsw i32 [[TMP45]], [[TMP46]]
-// CHECK-64-NEXT: store i32 [[ADD20]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: [[TMP47:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: [[TMP48:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-64-NEXT: [[ADD21:%.*]] = add nsw i32 [[TMP47]], [[TMP48]]
-// CHECK-64-NEXT: store i32 [[ADD21]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: [[TMP49:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP50:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-64-NEXT: [[ADD22:%.*]] = add nsw i32 [[TMP49]], [[TMP50]]
-// CHECK-64-NEXT: store i32 [[ADD22]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP51:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[CMP23:%.*]] = icmp sgt i32 [[TMP51]], 9
-// CHECK-64-NEXT: br i1 [[CMP23]], label [[COND_TRUE24:%.*]], label [[COND_FALSE25:%.*]]
-// CHECK-64: cond.true24:
-// CHECK-64-NEXT: br label [[COND_END26:%.*]]
-// CHECK-64: cond.false25:
-// CHECK-64-NEXT: [[TMP52:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: br label [[COND_END26]]
-// CHECK-64: cond.end26:
-// CHECK-64-NEXT: [[COND27:%.*]] = phi i32 [ 9, [[COND_TRUE24]] ], [ [[TMP52]], [[COND_FALSE25]] ]
-// CHECK-64-NEXT: store i32 [[COND27]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP53:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: store i32 [[TMP53]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND11]], !llvm.loop [[LOOP70:![0-9]+]]
-// CHECK-64: omp.inner.for.end28:
-// CHECK-64-NEXT: br label [[OMP_IF_END]]
-// CHECK-64: omp_if.end:
+// CHECK-64-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
+// CHECK-64-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP4]], 0
+// CHECK-64-NEXT: [[STOREDV:%.*]] = zext i1 [[LOADEDV]] to i8
+// CHECK-64-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
+// CHECK-64-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK-64-NEXT: [[TMP7:%.*]] = inttoptr i64 [[TMP1]] to ptr
+// CHECK-64-NEXT: store ptr [[TMP7]], ptr [[TMP6]], align 8
+// CHECK-64-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK-64-NEXT: [[TMP9:%.*]] = inttoptr i64 [[TMP3]] to ptr
+// CHECK-64-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 8
+// CHECK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK-64-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP5]] to ptr
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8
+// CHECK-64-NEXT: [[TMP12:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
+// CHECK-64-NEXT: [[LOADEDV1:%.*]] = icmp ne i8 [[TMP12]], 0
+// CHECK-64-NEXT: [[TMP13:%.*]] = zext i1 [[LOADEDV1]] to i32
+// CHECK-64-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP15]], i32 [[TMP13]], i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l15_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3, i32 0)
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
-// CHECK-64-NEXT: [[TMP54:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK-64-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
-// CHECK-64-NEXT: br i1 [[TMP55]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK-64-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
+// CHECK-64-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK-64: .omp.final.then:
// CHECK-64-NEXT: store i32 10, ptr [[I]], align 4
// CHECK-64-NEXT: br label [[DOTOMP_FINAL_DONE]]
@@ -513,143 +396,36 @@ int a;
// CHECK-64: omp_if.then:
// CHECK-64-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
-// CHECK-64-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72:![0-9]+]]
-// CHECK-64-NEXT: [[CONV2:%.*]] = sext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP72]]
-// CHECK-64-NEXT: [[CMP:%.*]] = icmp ule i64 [[CONV2]], [[TMP7]]
-// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
-// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
-// CHECK-64: omp.body.continue:
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP73:![0-9]+]]
-// CHECK-64: omp.inner.for.end:
-// CHECK-64-NEXT: br label [[OMP_IF_END:%.*]]
-// CHECK-64: omp_if.else:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK-64-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP12]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
-// CHECK-64-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
-// CHECK-64-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND4:%.*]]
-// CHECK-64: omp.inner.for.cond4:
-// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: [[CONV5:%.*]] = sext i32 [[TMP14]] to i64
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
-// CHECK-64-NEXT: [[CMP6:%.*]] = icmp ule i64 [[CONV5]], [[TMP15]]
-// CHECK-64-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY7:%.*]], label [[OMP_INNER_FOR_END13:%.*]]
-// CHECK-64: omp.inner.for.body7:
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: [[MUL8:%.*]] = mul nsw i32 [[TMP16]], 1
-// CHECK-64-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL8]]
-// CHECK-64-NEXT: store i32 [[ADD9]], ptr [[I]], align 4
-// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE10:%.*]]
-// CHECK-64: omp.body.continue10:
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC11:%.*]]
-// CHECK-64: omp.inner.for.inc11:
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-64-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP17]], [[TMP18]]
-// CHECK-64-NEXT: store i32 [[ADD12]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND4]], !llvm.loop [[LOOP75:![0-9]+]]
-// CHECK-64: omp.inner.for.end13:
-// CHECK-64-NEXT: br label [[OMP_IF_END]]
-// CHECK-64: omp_if.end:
-// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
-// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP20]])
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK-64-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
-// CHECK-64-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
-// CHECK-64: .omp.final.then:
-// CHECK-64-NEXT: store i32 10, ptr [[I]], align 4
-// CHECK-64-NEXT: br label [[DOTOMP_FINAL_DONE]]
-// CHECK-64: .omp.final.done:
-// CHECK-64-NEXT: ret void
-//
-//
-// CHECK-64-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l15_omp_outlined_omp_outlined1
-// CHECK-64-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
-// CHECK-64-NEXT: entry:
-// CHECK-64-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
-// CHECK-64-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
-// CHECK-64-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
-// CHECK-64-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
-// CHECK-64-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
-// CHECK-64-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
-// CHECK-64-NEXT: [[TMP:%.*]] = alloca i32, align 4
-// CHECK-64-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
-// CHECK-64-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
-// CHECK-64-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
-// CHECK-64-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
-// CHECK-64-NEXT: [[I:%.*]] = alloca i32, align 4
-// CHECK-64-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK-64-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
-// CHECK-64-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
-// CHECK-64-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
-// CHECK-64-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
-// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
-// CHECK-64-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
-// CHECK-64-NEXT: [[TMP0:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
-// CHECK-64-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
-// CHECK-64-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
-// CHECK-64-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
-// CHECK-64-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
-// CHECK-64-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
-// CHECK-64-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK-64-NEXT: [[TMP2:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
-// CHECK-64-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP2]], 0
-// CHECK-64-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
-// CHECK-64: omp_if.then:
-// CHECK-64-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-64-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP76:![0-9]+]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66:![0-9]+]]
// CHECK-64-NEXT: [[CONV2:%.*]] = sext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP76]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP66]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp ule i64 [[CONV2]], [[TMP7]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP76]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP76]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP66]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP76]]
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP76]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP66]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP76]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP77:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP67:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK-64: omp_if.else:
// CHECK-64-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP12]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP12]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-64-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND4:%.*]]
@@ -672,7 +448,7 @@ int a;
// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-64-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP17]], [[TMP18]]
// CHECK-64-NEXT: store i32 [[ADD12]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND4]], !llvm.loop [[LOOP79:![0-9]+]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND4]], !llvm.loop [[LOOP70:![0-9]+]]
// CHECK-64: omp.inner.for.end13:
// CHECK-64-NEXT: br label [[OMP_IF_END]]
// CHECK-64: omp_if.end:
@@ -680,7 +456,7 @@ int a;
// CHECK-64: omp.loop.exit:
// CHECK-64-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP20]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP20]])
// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-64-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -734,7 +510,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -750,53 +526,53 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP80:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72:![0-9]+]]
// CHECK-64-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-64-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-64-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK-64-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP7]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP80]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP72]]
// CHECK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK-64-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP9]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP80]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l18_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP80]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP72]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l18_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP72]]
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-64-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
// CHECK-64-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-64: cond.true5:
// CHECK-64-NEXT: br label [[COND_END7:%.*]]
// CHECK-64: cond.false6:
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-64-NEXT: br label [[COND_END7]]
// CHECK-64: cond.end7:
// CHECK-64-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP81:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP73:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
// CHECK-64-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -837,7 +613,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -853,27 +629,27 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP83:![0-9]+]]
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP83]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP75:![0-9]+]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP75]]
// CHECK-64-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK-64-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP83]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP75]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP83]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP75]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP83]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP75]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP83]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP84:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP75]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP76:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
// CHECK-64-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -927,7 +703,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -943,53 +719,53 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP86:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP78:![0-9]+]]
// CHECK-64-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-64-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-64-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK-64-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP7]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP86]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP78]]
// CHECK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK-64-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP9]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP86]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l21_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP86]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP78]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l21_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP78]]
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-64-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
// CHECK-64-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-64: cond.true5:
// CHECK-64-NEXT: br label [[COND_END7:%.*]]
// CHECK-64: cond.false6:
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-64-NEXT: br label [[COND_END7]]
// CHECK-64: cond.end7:
// CHECK-64-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP87:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP79:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
// CHECK-64-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -1030,34 +806,34 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP89:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP81:![0-9]+]]
// CHECK-64-NEXT: [[CONV2:%.*]] = sext i32 [[TMP5]] to i64
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP89]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP81]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp ule i64 [[CONV2]], [[TMP6]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP89]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP81]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP89]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP81]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP89]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP89]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP81]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP81]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP89]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP90:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP81]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP82:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP10]], 0
// CHECK-64-NEXT: br i1 [[TMP11]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -1111,7 +887,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -1127,53 +903,53 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP92:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84:![0-9]+]]
// CHECK-64-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-64-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-64-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK-64-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP7]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP92]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP84]]
// CHECK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK-64-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP9]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP92]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l24_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP92]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP84]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l24_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP84]]
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-64-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
// CHECK-64-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-64: cond.true5:
// CHECK-64-NEXT: br label [[COND_END7:%.*]]
// CHECK-64: cond.false6:
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-64-NEXT: br label [[COND_END7]]
// CHECK-64: cond.end7:
// CHECK-64-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP93:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP85:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
// CHECK-64-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -1227,23 +1003,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP95:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP95]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP87:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP87]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP95]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP87]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP95]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP87]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP95]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP87]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP95]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP96:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP87]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP88:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -1303,7 +1079,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -1319,53 +1095,53 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP98:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP90:![0-9]+]]
// CHECK-64-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-64-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-64-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK-64-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP7]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP98]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP90]]
// CHECK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK-64-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP9]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP98]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l27_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP98]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP90]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l27_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP90]]
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-64-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
// CHECK-64-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-64: cond.true5:
// CHECK-64-NEXT: br label [[COND_END7:%.*]]
// CHECK-64: cond.false6:
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-64-NEXT: br label [[COND_END7]]
// CHECK-64: cond.end7:
// CHECK-64-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP99:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP91:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
// CHECK-64-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -1419,23 +1195,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP101:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP101]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP93:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP93]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP101]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP93]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP101]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP93]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP101]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP93]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP101]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP102:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP93]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP94:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -1495,7 +1271,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -1511,53 +1287,53 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP104:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP96:![0-9]+]]
// CHECK-64-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-64-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-64-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK-64-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP7]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP104]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP96]]
// CHECK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK-64-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP9]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP104]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l30_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP104]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP96]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l30_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP96]]
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-64-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
// CHECK-64-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-64: cond.true5:
// CHECK-64-NEXT: br label [[COND_END7:%.*]]
// CHECK-64: cond.false6:
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-64-NEXT: br label [[COND_END7]]
// CHECK-64: cond.end7:
// CHECK-64-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP105:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP97:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
// CHECK-64-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -1611,23 +1387,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP107:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP107]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP99:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP99]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP107]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP99]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP107]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP99]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP107]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP99]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP107]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP108:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP99]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP100:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -1687,7 +1463,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -1703,53 +1479,53 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP110:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP102:![0-9]+]]
// CHECK-64-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-64-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-64-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK-64-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP7]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP110]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP102]]
// CHECK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK-64-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP9]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP110]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l33_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP110]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP102]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l33_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP102]]
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-64-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
// CHECK-64-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-64: cond.true5:
// CHECK-64-NEXT: br label [[COND_END7:%.*]]
// CHECK-64: cond.false6:
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-64-NEXT: br label [[COND_END7]]
// CHECK-64: cond.end7:
// CHECK-64-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP111:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP103:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
// CHECK-64-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -1803,23 +1579,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP113:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP113]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP105:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP105]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP113]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP105]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP113]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP105]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP113]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP105]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP113]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP114:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP105]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP106:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -1887,83 +1663,33 @@ int a;
// CHECK-64-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK-64-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
-// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK-64: cond.true:
-// CHECK-64-NEXT: br label [[COND_END:%.*]]
-// CHECK-64: cond.false:
-// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: br label [[COND_END]]
-// CHECK-64: cond.end:
-// CHECK-64-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
-// CHECK-64-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: [[CMP2:%.*]] = icmp slt i32 [[TMP5]], 10
-// CHECK-64-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[A1]], align 4
-// CHECK-64-NEXT: store i32 [[TMP10]], ptr [[A_CASTED]], align 4
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i64, ptr [[A_CASTED]], align 8
-// CHECK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK-64-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP7]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
-// CHECK-64-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK-64-NEXT: [[TMP15:%.*]] = inttoptr i64 [[TMP9]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8
-// CHECK-64-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK-64-NEXT: [[TMP17:%.*]] = inttoptr i64 [[TMP11]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l37_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3, i32 0)
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-64-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
-// CHECK-64-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP24]], 9
-// CHECK-64-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
-// CHECK-64: cond.true6:
-// CHECK-64-NEXT: br label [[COND_END8:%.*]]
-// CHECK-64: cond.false7:
-// CHECK-64-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: br label [[COND_END8]]
-// CHECK-64: cond.end8:
-// CHECK-64-NEXT: [[COND9:%.*]] = phi i32 [ 9, [[COND_TRUE6]] ], [ [[TMP25]], [[COND_FALSE7]] ]
-// CHECK-64-NEXT: store i32 [[COND9]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: store i32 [[TMP26]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK-64: omp.inner.for.end:
+// CHECK-64-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[A1]], align 4
+// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[A_CASTED]], align 4
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i64, ptr [[A_CASTED]], align 8
+// CHECK-64-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK-64-NEXT: [[TMP7:%.*]] = inttoptr i64 [[TMP1]] to ptr
+// CHECK-64-NEXT: store ptr [[TMP7]], ptr [[TMP6]], align 8
+// CHECK-64-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK-64-NEXT: [[TMP9:%.*]] = inttoptr i64 [[TMP3]] to ptr
+// CHECK-64-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 8
+// CHECK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK-64-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP5]] to ptr
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8
+// CHECK-64-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK-64-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP13]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l37_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3, i32 0)
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
-// CHECK-64-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK-64-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
-// CHECK-64-NEXT: br i1 [[TMP28]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK-64-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
+// CHECK-64-NEXT: br i1 [[TMP15]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
// CHECK-64: .omp.lastprivate.then:
-// CHECK-64-NEXT: [[TMP29:%.*]] = load i32, ptr [[A1]], align 4
-// CHECK-64-NEXT: store i32 [[TMP29]], ptr [[A_ADDR]], align 4
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[A1]], align 4
+// CHECK-64-NEXT: store i32 [[TMP16]], ptr [[A_ADDR]], align 4
// CHECK-64-NEXT: br label [[DOTOMP_LASTPRIVATE_DONE]]
// CHECK-64: .omp.lastprivate.done:
// CHECK-64-NEXT: call void @__kmpc_free_shared(ptr [[A1]], i64 4)
@@ -2003,7 +1729,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -2032,7 +1758,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
// CHECK-64-NEXT: br i1 [[TMP12]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
@@ -2087,7 +1813,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -2149,7 +1875,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -2183,7 +1909,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -2219,7 +1945,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-64-NEXT: ret void
//
//
@@ -2266,7 +1992,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -2328,7 +2054,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -2362,7 +2088,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -2389,7 +2115,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-64-NEXT: ret void
//
//
@@ -2436,7 +2162,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -2498,7 +2224,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -2545,23 +2271,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP108:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP108]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP108]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP108]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP108]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP117:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP108]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP109:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -2614,7 +2340,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -2676,7 +2402,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -2723,23 +2449,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP119:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP119]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP111:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP111]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP119]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP111]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP119]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP111]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP119]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP111]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP119]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP120:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP111]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP112:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -2792,7 +2518,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -2854,7 +2580,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -2901,23 +2627,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP122:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP122]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP114:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP114]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP122]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP114]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP122]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP114]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP122]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP114]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP122]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP123:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP114]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP115:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -2970,7 +2696,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -3032,7 +2758,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -3079,23 +2805,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP125:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP125]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP117:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP117]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP125]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP117]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP125]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP117]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP125]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP117]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP125]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP126:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP117]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP118:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -3147,74 +2873,24 @@ int a;
// CHECK-64-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK-64-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
-// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK-64: cond.true:
-// CHECK-64-NEXT: br label [[COND_END:%.*]]
-// CHECK-64: cond.false:
-// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: br label [[COND_END]]
-// CHECK-64: cond.end:
-// CHECK-64-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
-// CHECK-64-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP128:![0-9]+]]
-// CHECK-64-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
-// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
-// CHECK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK-64-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP7]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK-64-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP9]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l58_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
-// CHECK-64-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK-64: cond.true5:
-// CHECK-64-NEXT: br label [[COND_END7:%.*]]
-// CHECK-64: cond.false6:
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: br label [[COND_END7]]
-// CHECK-64: cond.end7:
-// CHECK-64-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP129:![0-9]+]]
-// CHECK-64: omp.inner.for.end:
+// CHECK-64-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
+// CHECK-64-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK-64-NEXT: [[TMP5:%.*]] = inttoptr i64 [[TMP1]] to ptr
+// CHECK-64-NEXT: store ptr [[TMP5]], ptr [[TMP4]], align 8
+// CHECK-64-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK-64-NEXT: [[TMP7:%.*]] = inttoptr i64 [[TMP3]] to ptr
+// CHECK-64-NEXT: store ptr [[TMP7]], ptr [[TMP6]], align 8
+// CHECK-64-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP9]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l58_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0)
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
-// CHECK-64-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK-64-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
-// CHECK-64-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK-64-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP10]], 0
+// CHECK-64-NEXT: br i1 [[TMP11]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK-64: .omp.final.then:
// CHECK-64-NEXT: store i32 10, ptr [[I]], align 4
// CHECK-64-NEXT: br label [[DOTOMP_FINAL_DONE]]
@@ -3252,34 +2928,34 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP131:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120:![0-9]+]]
// CHECK-64-NEXT: [[CONV2:%.*]] = sext i32 [[TMP5]] to i64
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP131]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP120]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp ule i64 [[CONV2]], [[TMP6]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP131]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP131]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP120]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP131]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP131]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP120]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP131]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP132:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP121:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP10]], 0
// CHECK-64-NEXT: br i1 [[TMP11]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -3335,7 +3011,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -3351,53 +3027,53 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP134:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123:![0-9]+]]
// CHECK-64-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-64-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-64-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK-64-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP7]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP134]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP123]]
// CHECK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK-64-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP9]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP134]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP134]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP123]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP123]]
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-64-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
// CHECK-64-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-64: cond.true5:
// CHECK-64-NEXT: br label [[COND_END7:%.*]]
// CHECK-64: cond.false6:
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-64-NEXT: br label [[COND_END7]]
// CHECK-64: cond.end7:
// CHECK-64-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP135:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP124:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
// CHECK-64-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -3438,7 +3114,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -3454,27 +3130,27 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137:![0-9]+]]
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP126:![0-9]+]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP126]]
// CHECK-64-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK-64-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP126]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP126]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP126]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP138:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP126]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP127:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
// CHECK-64-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -3529,7 +3205,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -3545,33 +3221,33 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP140:![0-9]+]]
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP140]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP129:![0-9]+]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP129]]
// CHECK-64-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP140]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP129]]
// CHECK-64-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP140]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP129]]
// CHECK-64-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK-64-NEXT: [[TMP11:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK-64-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP8]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8, !llvm.access.group [[ACC_GRP140]]
+// CHECK-64-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8, !llvm.access.group [[ACC_GRP129]]
// CHECK-64-NEXT: [[TMP13:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK-64-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP10]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8, !llvm.access.group [[ACC_GRP140]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l73_omp_outlined_omp_outlined, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l73_omp_outlined_omp_outlined_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP140]]
+// CHECK-64-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8, !llvm.access.group [[ACC_GRP129]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l73_omp_outlined_omp_outlined, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l73_omp_outlined_omp_outlined_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP129]]
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP140]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP140]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP129]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP129]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP15]], [[TMP16]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP140]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP141:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP129]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP130:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0
// CHECK-64-NEXT: br i1 [[TMP18]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -3615,34 +3291,34 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP143:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP132:![0-9]+]]
// CHECK-64-NEXT: [[CONV2:%.*]] = sext i32 [[TMP5]] to i64
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP143]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP132]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp ule i64 [[CONV2]], [[TMP6]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP143]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP132]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP143]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP132]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP143]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP143]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP132]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP132]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP143]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP144:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP132]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP133:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP10]], 0
// CHECK-64-NEXT: br i1 [[TMP11]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -3716,7 +3392,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -3732,53 +3408,53 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP146:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP135:![0-9]+]]
// CHECK-64-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-64-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-64-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK-64-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP7]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP146]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP135]]
// CHECK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK-64-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP9]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP146]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l81_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP146]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP135]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l81_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP135]]
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-64-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
// CHECK-64-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-64: cond.true5:
// CHECK-64-NEXT: br label [[COND_END7:%.*]]
// CHECK-64: cond.false6:
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-64-NEXT: br label [[COND_END7]]
// CHECK-64: cond.end7:
// CHECK-64-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP147:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP136:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
// CHECK-64-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -3832,23 +3508,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP149:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP149]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP138:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP138]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP149]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP138]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP149]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP138]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP149]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP138]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP149]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP150:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP138]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP139:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -3908,7 +3584,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -3924,53 +3600,53 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP152:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP141:![0-9]+]]
// CHECK-64-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-64-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-64-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK-64-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP7]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP152]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP141]]
// CHECK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK-64-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP9]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP152]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l85_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP152]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP141]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l85_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP141]]
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-64-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
// CHECK-64-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-64: cond.true5:
// CHECK-64-NEXT: br label [[COND_END7:%.*]]
// CHECK-64: cond.false6:
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-64-NEXT: br label [[COND_END7]]
// CHECK-64: cond.end7:
// CHECK-64-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP153:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP142:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
// CHECK-64-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -4024,23 +3700,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP155:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP155]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP144:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP144]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP155]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP144]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP155]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP144]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP155]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP144]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP155]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP156:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP144]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP145:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -4100,7 +3776,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -4116,53 +3792,53 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP158:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP147:![0-9]+]]
// CHECK-64-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-64-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-64-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK-64-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP7]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP158]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP147]]
// CHECK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK-64-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP9]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP158]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l89_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP158]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP147]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l89_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP147]]
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-64-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
// CHECK-64-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-64: cond.true5:
// CHECK-64-NEXT: br label [[COND_END7:%.*]]
// CHECK-64: cond.false6:
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-64-NEXT: br label [[COND_END7]]
// CHECK-64: cond.end7:
// CHECK-64-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP159:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP148:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
// CHECK-64-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -4216,23 +3892,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP161:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP161]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP150:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP150]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP161]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP150]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP161]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP150]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP161]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP150]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP161]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP162:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP150]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP151:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -4292,7 +3968,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -4308,53 +3984,53 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP164:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP153:![0-9]+]]
// CHECK-64-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-64-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-64-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK-64-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP7]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP164]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP153]]
// CHECK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK-64-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP9]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP164]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l93_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP164]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP153]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l93_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0), !llvm.access.group [[ACC_GRP153]]
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-64-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
// CHECK-64-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-64: cond.true5:
// CHECK-64-NEXT: br label [[COND_END7:%.*]]
// CHECK-64: cond.false6:
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-64-NEXT: br label [[COND_END7]]
// CHECK-64: cond.end7:
// CHECK-64-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP165:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP154:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
// CHECK-64-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -4408,23 +4084,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP167:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP167]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP156:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP156]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP167]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP156]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP167]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP156]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP167]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP156]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP167]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP168:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP156]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP157:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -4482,71 +4158,21 @@ int a;
// CHECK-64-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK-64-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
-// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK-64: cond.true:
-// CHECK-64-NEXT: br label [[COND_END:%.*]]
-// CHECK-64: cond.false:
-// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: br label [[COND_END]]
-// CHECK-64: cond.end:
-// CHECK-64-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
-// CHECK-64-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
-// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
-// CHECK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK-64-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP7]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8
-// CHECK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK-64-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP9]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l97_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0)
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
-// CHECK-64-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK-64: cond.true5:
-// CHECK-64-NEXT: br label [[COND_END7:%.*]]
-// CHECK-64: cond.false6:
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: br label [[COND_END7]]
-// CHECK-64: cond.end7:
-// CHECK-64-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK-64: omp.inner.for.end:
+// CHECK-64-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
+// CHECK-64-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK-64-NEXT: [[TMP5:%.*]] = inttoptr i64 [[TMP1]] to ptr
+// CHECK-64-NEXT: store ptr [[TMP5]], ptr [[TMP4]], align 8
+// CHECK-64-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK-64-NEXT: [[TMP7:%.*]] = inttoptr i64 [[TMP3]] to ptr
+// CHECK-64-NEXT: store ptr [[TMP7]], ptr [[TMP6]], align 8
+// CHECK-64-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP9]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l97_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0)
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -4580,7 +4206,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -4607,7 +4233,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-64-NEXT: ret void
//
//
@@ -4654,7 +4280,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -4716,7 +4342,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -4750,7 +4376,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -4786,7 +4412,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-64-NEXT: ret void
//
//
@@ -4833,7 +4459,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -4895,7 +4521,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -4929,7 +4555,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -4956,7 +4582,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-64-NEXT: ret void
//
//
@@ -5003,7 +4629,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -5065,7 +4691,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -5112,23 +4738,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP170:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP170]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP159:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP159]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP170]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP159]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP170]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP159]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP170]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP159]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP170]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP171:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP159]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP160:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -5181,7 +4807,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -5243,7 +4869,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -5290,23 +4916,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP173:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP173]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP162:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP162]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP173]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP162]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP173]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP162]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP173]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP162]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP173]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP174:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP162]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP163:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -5359,7 +4985,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -5421,7 +5047,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -5468,23 +5094,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP176:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP176]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP165:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP165]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP176]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP165]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP176]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP165]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP176]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP165]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP176]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP177:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP165]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP166:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -5537,7 +5163,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -5599,7 +5225,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -5646,23 +5272,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP179:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP179]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP168:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP168]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP179]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP168]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP179]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP168]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP179]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP168]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP179]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP180:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP168]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP169:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -5713,71 +5339,21 @@ int a;
// CHECK-64-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK-64-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
-// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK-64: cond.true:
-// CHECK-64-NEXT: br label [[COND_END:%.*]]
-// CHECK-64: cond.false:
-// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: br label [[COND_END]]
-// CHECK-64: cond.end:
-// CHECK-64-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
-// CHECK-64-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
-// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
-// CHECK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK-64-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP7]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8
-// CHECK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK-64-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP9]] to ptr
-// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
-// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l125_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0)
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
-// CHECK-64-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK-64: cond.true5:
-// CHECK-64-NEXT: br label [[COND_END7:%.*]]
-// CHECK-64: cond.false6:
-// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: br label [[COND_END7]]
-// CHECK-64: cond.end7:
-// CHECK-64-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK-64: omp.inner.for.end:
+// CHECK-64-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
+// CHECK-64-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK-64-NEXT: [[TMP5:%.*]] = inttoptr i64 [[TMP1]] to ptr
+// CHECK-64-NEXT: store ptr [[TMP5]], ptr [[TMP4]], align 8
+// CHECK-64-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK-64-NEXT: [[TMP7:%.*]] = inttoptr i64 [[TMP3]] to ptr
+// CHECK-64-NEXT: store ptr [[TMP7]], ptr [[TMP6]], align 8
+// CHECK-64-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
+// CHECK-64-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP9]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l125_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2, i32 0)
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -5811,7 +5387,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -5838,7 +5414,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-64-NEXT: ret void
//
//
@@ -5885,7 +5461,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -5947,7 +5523,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -5981,7 +5557,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -6017,7 +5593,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-64-NEXT: ret void
//
//
@@ -6064,7 +5640,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -6126,7 +5702,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -6160,7 +5736,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -6187,7 +5763,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-64-NEXT: ret void
//
//
@@ -6234,7 +5810,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -6296,7 +5872,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -6343,23 +5919,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP182:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP182]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP171:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP171]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP182]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP171]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP182]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP171]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP182]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP171]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP182]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP183:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP171]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP172:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -6412,7 +5988,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -6474,7 +6050,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -6521,23 +6097,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP185:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP185]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP174:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP174]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP185]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP174]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP185]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP174]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP185]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP174]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP185]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP186:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP174]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP175:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -6590,7 +6166,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -6652,7 +6228,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -6699,23 +6275,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP188:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP188]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP177:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP177]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP188]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP177]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP188]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP177]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP188]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP177]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP188]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP189:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP177]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP178:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -6768,7 +6344,7 @@ int a;
// CHECK-64-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -6830,7 +6406,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -6877,23 +6453,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP191:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP191]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP180:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP180]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP191]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP180]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP191]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP180]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP191]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP180]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP191]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP192:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP180]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP181:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -6946,7 +6522,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-64: omp.dispatch.cond:
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -6999,7 +6575,7 @@ int a;
// CHECK-64-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -7041,7 +6617,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -7077,7 +6653,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -7119,7 +6695,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-64: omp.dispatch.cond:
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -7172,7 +6748,7 @@ int a;
// CHECK-64-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -7225,23 +6801,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP194:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP194]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP183:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP183]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP194]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP183]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP194]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP183]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP194]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP183]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP194]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP195:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP183]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP184:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -7300,23 +6876,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP197:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP197]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP186:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP186]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP197]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP186]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP197]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP186]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP197]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP186]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP197]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP198:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP186]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP187:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -7375,23 +6951,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP200:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP200]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP189:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP189]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP200]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP189]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP200]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP189]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP200]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP189]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP200]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP201:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP189]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP190:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -7450,23 +7026,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP203:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP203]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP192:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP192]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP203]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP192]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP203]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP192]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP203]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP192]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP203]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP204:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP192]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP193:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -7519,7 +7095,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-64: omp.dispatch.cond:
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -7542,23 +7118,23 @@ int a;
// CHECK-64: omp.dispatch.body:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP206:![0-9]+]]
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP206]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP195:![0-9]+]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP195]]
// CHECK-64-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK-64-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP206]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP195]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP206]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP195]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP206]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP195]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP206]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP207:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP195]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP196:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -7572,7 +7148,7 @@ int a;
// CHECK-64-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK-64-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -7622,7 +7198,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -7638,27 +7214,27 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP209:![0-9]+]]
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP209]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP198:![0-9]+]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP198]]
// CHECK-64-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP209]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP198]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP209]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP198]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP209]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP198]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP209]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP210:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP198]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP199:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP10:%.*]] = icmp ne i32 [[TMP9]], 0
// CHECK-64-NEXT: br i1 [[TMP10]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -7708,7 +7284,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-64: omp.dispatch.cond:
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -7731,23 +7307,23 @@ int a;
// CHECK-64: omp.dispatch.body:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP212:![0-9]+]]
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP212]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP201:![0-9]+]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP201]]
// CHECK-64-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK-64-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP212]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP201]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP212]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP201]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP212]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP201]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP212]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP213:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP201]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP202:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -7761,7 +7337,7 @@ int a;
// CHECK-64-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK-64-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -7822,23 +7398,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP215:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP215]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP204:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP204]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP215]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP204]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP215]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP204]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP215]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP204]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP215]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP216:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP204]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP205:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -7905,23 +7481,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP218:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP218]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP207:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP207]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP218]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP207]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP218]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP207]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP218]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP207]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP218]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP219:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP207]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP208:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -7988,23 +7564,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP210:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP210]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP210]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP210]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP210]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP222:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP210]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP211:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -8071,23 +7647,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP224:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP224]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP213:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP213]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP224]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP213]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP224]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP213]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP224]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP213]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP224]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP225:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP213]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP214:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -8154,24 +7730,24 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP216:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP216]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP216]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP216]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP216]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
-// CHECK-64-NEXT: call void @__kmpc_dispatch_fini_4(ptr @[[GLOB1]], i32 [[TMP1]]), !llvm.access.group [[ACC_GRP227]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP228:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP216]]
+// CHECK-64-NEXT: call void @__kmpc_dispatch_fini_4(ptr @[[GLOB1]], i32 [[TMP1]]), !llvm.access.group [[ACC_GRP216]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP217:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -8227,7 +7803,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -8243,27 +7819,27 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP230:![0-9]+]]
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP230]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP219:![0-9]+]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP219]]
// CHECK-64-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK-64-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP230]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP219]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP230]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP219]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP230]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP219]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP230]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP231:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP219]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP220:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP10:%.*]] = icmp ne i32 [[TMP9]], 0
// CHECK-64-NEXT: br i1 [[TMP10]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -8313,7 +7889,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-64: omp.dispatch.cond:
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -8336,23 +7912,23 @@ int a;
// CHECK-64: omp.dispatch.body:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP233:![0-9]+]]
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP233]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP222:![0-9]+]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP222]]
// CHECK-64-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK-64-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP233]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP222]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP233]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP222]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP233]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP222]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP233]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP234:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP222]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP223:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -8366,7 +7942,7 @@ int a;
// CHECK-64-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK-64-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -8427,23 +8003,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP236:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP236]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP225:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP225]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP236]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP225]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP236]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP225]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP236]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP225]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP236]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP237:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP225]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP226:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -8510,23 +8086,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP239:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP239]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP228:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP228]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP239]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP228]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP239]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP228]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP239]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP228]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP239]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP240:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP228]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP229:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -8593,23 +8169,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP242:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP242]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP231]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP242]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP242]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP231]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP242]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP242]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP243:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP232:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -8676,23 +8252,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP245:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP245]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP234:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP234]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP245]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP234]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP245]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP234]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP245]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP234]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP245]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP246:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP234]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP235:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -8748,7 +8324,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-64: omp.dispatch.cond:
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -8801,7 +8377,7 @@ int a;
// CHECK-64-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -8843,7 +8419,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-64-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -8879,7 +8455,7 @@ int a;
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -8921,7 +8497,7 @@ int a;
// CHECK-64-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-64-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-64: omp.dispatch.cond:
// CHECK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -8974,7 +8550,7 @@ int a;
// CHECK-64-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
-// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-64-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -9027,23 +8603,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP248:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP248]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP237:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP237]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP248]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP237]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP248]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP237]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP248]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP237]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP248]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP249:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP237]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP238:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -9102,23 +8678,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP251:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP251]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP240:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP240]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP251]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP240]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP251]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP240]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP251]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP240]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP251]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP252:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP240]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP241:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -9177,23 +8753,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP254:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP254]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP243:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP243]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP254]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP243]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP254]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP243]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP254]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP243]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP254]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP255:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP243]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP244:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -9252,23 +8828,23 @@ int a;
// CHECK-64-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-64: omp.inner.for.cond:
-// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP257:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP257]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP246:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP246]]
// CHECK-64-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-64-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-64: omp.inner.for.body:
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP257]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP246]]
// CHECK-64-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP257]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP246]]
// CHECK-64-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-64: omp.body.continue:
// CHECK-64-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-64: omp.inner.for.inc:
-// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP257]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP246]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP257]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP258:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP246]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP247:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -9322,8 +8898,6 @@ int a;
// CHECK-32-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK-32-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK-32-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [3 x ptr], align 4
-// CHECK-32-NEXT: [[DOTCAPTURE_EXPR__CASTED15:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[CAPTURED_VARS_ADDRS17:%.*]] = alloca [3 x ptr], align 4
// CHECK-32-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK-32-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
@@ -9332,249 +8906,33 @@ int a;
// CHECK-32-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
-// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK-32: cond.true:
-// CHECK-32-NEXT: br label [[COND_END:%.*]]
-// CHECK-32: cond.false:
-// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: br label [[COND_END]]
-// CHECK-32: cond.end:
-// CHECK-32-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
-// CHECK-32-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
-// CHECK-32-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP5]], 0
-// CHECK-32-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
-// CHECK-32: omp_if.then:
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66:![0-9]+]]
-// CHECK-32-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP6]], 10
-// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[LOADEDV2:%.*]] = icmp ne i8 [[TMP9]], 0
-// CHECK-32-NEXT: [[STOREDV:%.*]] = zext i1 [[LOADEDV2]] to i8
-// CHECK-32-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK-32-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK-32-NEXT: [[TMP14:%.*]] = inttoptr i32 [[TMP8]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK-32-NEXT: [[TMP16:%.*]] = inttoptr i32 [[TMP10]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[LOADEDV3:%.*]] = icmp ne i8 [[TMP17]], 0
-// CHECK-32-NEXT: [[TMP18:%.*]] = zext i1 [[LOADEDV3]] to i32
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 [[TMP18]], i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l15_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3, i32 0), !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP21]], [[TMP22]]
-// CHECK-32-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP23]], [[TMP24]]
-// CHECK-32-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP25]], 9
-// CHECK-32-NEXT: br i1 [[CMP6]], label [[COND_TRUE7:%.*]], label [[COND_FALSE8:%.*]]
-// CHECK-32: cond.true7:
-// CHECK-32-NEXT: br label [[COND_END9:%.*]]
-// CHECK-32: cond.false8:
-// CHECK-32-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: br label [[COND_END9]]
-// CHECK-32: cond.end9:
-// CHECK-32-NEXT: [[COND10:%.*]] = phi i32 [ 9, [[COND_TRUE7]] ], [ [[TMP26]], [[COND_FALSE8]] ]
-// CHECK-32-NEXT: store i32 [[COND10]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: store i32 [[TMP27]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP67:![0-9]+]]
-// CHECK-32: omp.inner.for.end:
-// CHECK-32-NEXT: br label [[OMP_IF_END:%.*]]
-// CHECK-32: omp_if.else:
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND11:%.*]]
-// CHECK-32: omp.inner.for.cond11:
-// CHECK-32-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: [[CMP12:%.*]] = icmp slt i32 [[TMP28]], 10
-// CHECK-32-NEXT: br i1 [[CMP12]], label [[OMP_INNER_FOR_BODY13:%.*]], label [[OMP_INNER_FOR_END28:%.*]]
-// CHECK-32: omp.inner.for.body13:
-// CHECK-32-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP31:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
-// CHECK-32-NEXT: [[LOADEDV14:%.*]] = icmp ne i8 [[TMP31]], 0
-// CHECK-32-NEXT: [[STOREDV16:%.*]] = zext i1 [[LOADEDV14]] to i8
-// CHECK-32-NEXT: store i8 [[STOREDV16]], ptr [[DOTCAPTURE_EXPR__CASTED15]], align 1
-// CHECK-32-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED15]], align 4
-// CHECK-32-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS17]], i32 0, i32 0
-// CHECK-32-NEXT: [[TMP34:%.*]] = inttoptr i32 [[TMP29]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP34]], ptr [[TMP33]], align 4
-// CHECK-32-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS17]], i32 0, i32 1
-// CHECK-32-NEXT: [[TMP36:%.*]] = inttoptr i32 [[TMP30]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP36]], ptr [[TMP35]], align 4
-// CHECK-32-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS17]], i32 0, i32 2
-// CHECK-32-NEXT: [[TMP38:%.*]] = inttoptr i32 [[TMP32]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP38]], ptr [[TMP37]], align 4
-// CHECK-32-NEXT: [[TMP39:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
-// CHECK-32-NEXT: [[LOADEDV18:%.*]] = icmp ne i8 [[TMP39]], 0
-// CHECK-32-NEXT: [[TMP40:%.*]] = zext i1 [[LOADEDV18]] to i32
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 [[TMP40]], i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l15_omp_outlined_omp_outlined1, ptr null, ptr [[CAPTURED_VARS_ADDRS17]], i32 3, i32 0)
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC19:%.*]]
-// CHECK-32: omp.inner.for.inc19:
-// CHECK-32-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: [[TMP42:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-NEXT: [[ADD20:%.*]] = add nsw i32 [[TMP41]], [[TMP42]]
-// CHECK-32-NEXT: store i32 [[ADD20]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: [[TMP43:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: [[TMP44:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-NEXT: [[ADD21:%.*]] = add nsw i32 [[TMP43]], [[TMP44]]
-// CHECK-32-NEXT: store i32 [[ADD21]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: [[TMP45:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP46:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-NEXT: [[ADD22:%.*]] = add nsw i32 [[TMP45]], [[TMP46]]
-// CHECK-32-NEXT: store i32 [[ADD22]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP47:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[CMP23:%.*]] = icmp sgt i32 [[TMP47]], 9
-// CHECK-32-NEXT: br i1 [[CMP23]], label [[COND_TRUE24:%.*]], label [[COND_FALSE25:%.*]]
-// CHECK-32: cond.true24:
-// CHECK-32-NEXT: br label [[COND_END26:%.*]]
-// CHECK-32: cond.false25:
-// CHECK-32-NEXT: [[TMP48:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: br label [[COND_END26]]
-// CHECK-32: cond.end26:
-// CHECK-32-NEXT: [[COND27:%.*]] = phi i32 [ 9, [[COND_TRUE24]] ], [ [[TMP48]], [[COND_FALSE25]] ]
-// CHECK-32-NEXT: store i32 [[COND27]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP49:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: store i32 [[TMP49]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND11]], !llvm.loop [[LOOP70:![0-9]+]]
-// CHECK-32: omp.inner.for.end28:
-// CHECK-32-NEXT: br label [[OMP_IF_END]]
-// CHECK-32: omp_if.end:
-// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
-// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
-// CHECK-32-NEXT: [[TMP50:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK-32-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0
-// CHECK-32-NEXT: br i1 [[TMP51]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
-// CHECK-32: .omp.final.then:
-// CHECK-32-NEXT: store i32 10, ptr [[I]], align 4
-// CHECK-32-NEXT: br label [[DOTOMP_FINAL_DONE]]
-// CHECK-32: .omp.final.done:
-// CHECK-32-NEXT: ret void
-//
-//
-// CHECK-32-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l15_omp_outlined_omp_outlined
-// CHECK-32-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
-// CHECK-32-NEXT: entry:
-// CHECK-32-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
-// CHECK-32-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
-// CHECK-32-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[TMP:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[I:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
-// CHECK-32-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
-// CHECK-32-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
-// CHECK-32-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
-// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
-// CHECK-32-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
-// CHECK-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
-// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
-// CHECK-32-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_LB]], align 4
-// CHECK-32-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_UB]], align 4
-// CHECK-32-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[TMP2:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
// CHECK-32-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP2]], 0
-// CHECK-32-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
-// CHECK-32: omp_if.then:
-// CHECK-32-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
-// CHECK-32-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72:![0-9]+]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-32-NEXT: [[CMP:%.*]] = icmp ule i32 [[TMP6]], [[TMP7]]
-// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
-// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
-// CHECK-32: omp.body.continue:
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP73:![0-9]+]]
-// CHECK-32: omp.inner.for.end:
-// CHECK-32-NEXT: br label [[OMP_IF_END:%.*]]
-// CHECK-32: omp_if.else:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP12]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
-// CHECK-32-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND2:%.*]]
-// CHECK-32: omp.inner.for.cond2:
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
-// CHECK-32-NEXT: [[CMP3:%.*]] = icmp ule i32 [[TMP14]], [[TMP15]]
-// CHECK-32-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY4:%.*]], label [[OMP_INNER_FOR_END10:%.*]]
-// CHECK-32: omp.inner.for.body4:
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: [[MUL5:%.*]] = mul nsw i32 [[TMP16]], 1
-// CHECK-32-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
-// CHECK-32-NEXT: store i32 [[ADD6]], ptr [[I]], align 4
-// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE7:%.*]]
-// CHECK-32: omp.body.continue7:
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC8:%.*]]
-// CHECK-32: omp.inner.for.inc8:
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP17]], [[TMP18]]
-// CHECK-32-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND2]], !llvm.loop [[LOOP75:![0-9]+]]
-// CHECK-32: omp.inner.for.end10:
-// CHECK-32-NEXT: br label [[OMP_IF_END]]
-// CHECK-32: omp_if.end:
+// CHECK-32-NEXT: [[STOREDV:%.*]] = zext i1 [[LOADEDV]] to i8
+// CHECK-32-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1
+// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK-32-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK-32-NEXT: [[TMP5:%.*]] = inttoptr i32 [[TMP0]] to ptr
+// CHECK-32-NEXT: store ptr [[TMP5]], ptr [[TMP4]], align 4
+// CHECK-32-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK-32-NEXT: [[TMP7:%.*]] = inttoptr i32 [[TMP1]] to ptr
+// CHECK-32-NEXT: store ptr [[TMP7]], ptr [[TMP6]], align 4
+// CHECK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK-32-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP3]] to ptr
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
+// CHECK-32-NEXT: [[LOADEDV1:%.*]] = icmp ne i8 [[TMP10]], 0
+// CHECK-32-NEXT: [[TMP11:%.*]] = zext i1 [[LOADEDV1]] to i32
+// CHECK-32-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP13]], i32 [[TMP11]], i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l15_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3, i32 0)
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP20]])
-// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK-32-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
-// CHECK-32-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK-32-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
+// CHECK-32-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK-32: .omp.final.then:
// CHECK-32-NEXT: store i32 10, ptr [[I]], align 4
// CHECK-32-NEXT: br label [[DOTOMP_FINAL_DONE]]
@@ -9582,7 +8940,7 @@ int a;
// CHECK-32-NEXT: ret void
//
//
-// CHECK-32-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l15_omp_outlined_omp_outlined1
+// CHECK-32-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l15_omp_outlined_omp_outlined
// CHECK-32-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
// CHECK-32-NEXT: entry:
// CHECK-32-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
@@ -9616,35 +8974,35 @@ int a;
// CHECK-32: omp_if.then:
// CHECK-32-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP76:![0-9]+]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP76]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66:![0-9]+]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP66]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp ule i32 [[TMP6]], [[TMP7]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP76]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP76]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP66]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP76]]
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP76]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP66]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP76]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP77:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP67:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK-32: omp_if.else:
// CHECK-32-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP12]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP12]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND2:%.*]]
@@ -9666,7 +9024,7 @@ int a;
// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-32-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP17]], [[TMP18]]
// CHECK-32-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND2]], !llvm.loop [[LOOP79:![0-9]+]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND2]], !llvm.loop [[LOOP70:![0-9]+]]
// CHECK-32: omp.inner.for.end10:
// CHECK-32-NEXT: br label [[OMP_IF_END]]
// CHECK-32: omp_if.end:
@@ -9674,7 +9032,7 @@ int a;
// CHECK-32: omp.loop.exit:
// CHECK-32-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP20]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP20]])
// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -9728,7 +9086,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -9744,51 +9102,51 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP80:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72:![0-9]+]]
// CHECK-32-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l18_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP80]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l18_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP72]]
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-32-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32: cond.true5:
// CHECK-32-NEXT: br label [[COND_END7:%.*]]
// CHECK-32: cond.false6:
-// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-32-NEXT: br label [[COND_END7]]
// CHECK-32: cond.end7:
// CHECK-32-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP81:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP73:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -9827,7 +9185,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -9843,27 +9201,27 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP83:![0-9]+]]
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP83]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP75:![0-9]+]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP75]]
// CHECK-32-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP83]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP75]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP83]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP75]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP83]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP75]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP83]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP84:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP75]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP76:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
// CHECK-32-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -9917,7 +9275,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -9933,51 +9291,51 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP86:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP78:![0-9]+]]
// CHECK-32-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l21_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP86]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l21_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP78]]
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-32-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32: cond.true5:
// CHECK-32-NEXT: br label [[COND_END7:%.*]]
// CHECK-32: cond.false6:
-// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-32-NEXT: br label [[COND_END7]]
// CHECK-32: cond.end7:
// CHECK-32-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP87:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP79:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -10016,33 +9374,33 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP89:![0-9]+]]
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP89]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP81:![0-9]+]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP81]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp ule i32 [[TMP5]], [[TMP6]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP89]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP81]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP89]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP81]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP89]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP89]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP81]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP81]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP89]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP90:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP81]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP82:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP10]], 0
// CHECK-32-NEXT: br i1 [[TMP11]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -10096,7 +9454,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -10112,51 +9470,51 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP92:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84:![0-9]+]]
// CHECK-32-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l24_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP92]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l24_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP84]]
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-32-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32: cond.true5:
// CHECK-32-NEXT: br label [[COND_END7:%.*]]
// CHECK-32: cond.false6:
-// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-32-NEXT: br label [[COND_END7]]
// CHECK-32: cond.end7:
// CHECK-32-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP93:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP85:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -10208,23 +9566,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP95:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP95]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP87:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP87]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP95]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP87]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP95]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP87]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP95]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP87]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP95]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP96:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP87]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP88:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -10284,7 +9642,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -10300,51 +9658,51 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP98:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP90:![0-9]+]]
// CHECK-32-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l27_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP98]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l27_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP90]]
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-32-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32: cond.true5:
// CHECK-32-NEXT: br label [[COND_END7:%.*]]
// CHECK-32: cond.false6:
-// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-32-NEXT: br label [[COND_END7]]
// CHECK-32: cond.end7:
// CHECK-32-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP99:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP91:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -10396,23 +9754,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP101:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP101]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP93:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP93]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP101]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP93]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP101]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP93]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP101]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP93]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP101]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP102:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP93]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP94:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -10472,7 +9830,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -10488,51 +9846,51 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP104:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP96:![0-9]+]]
// CHECK-32-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l30_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP104]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l30_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP96]]
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-32-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32: cond.true5:
// CHECK-32-NEXT: br label [[COND_END7:%.*]]
// CHECK-32: cond.false6:
-// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-32-NEXT: br label [[COND_END7]]
// CHECK-32: cond.end7:
// CHECK-32-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP105:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP97:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -10584,23 +9942,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP107:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP107]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP99:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP99]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP107]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP99]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP107]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP99]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP107]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP99]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP107]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP108:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP99]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP100:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -10660,7 +10018,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -10676,51 +10034,51 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP110:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP102:![0-9]+]]
// CHECK-32-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l33_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP110]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l33_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP102]]
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-32-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32: cond.true5:
// CHECK-32-NEXT: br label [[COND_END7:%.*]]
// CHECK-32: cond.false6:
-// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-32-NEXT: br label [[COND_END7]]
// CHECK-32: cond.end7:
// CHECK-32-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP111:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP103:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -10772,23 +10130,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP113:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP113]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP105:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP105]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP113]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP105]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP113]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP105]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP113]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP105]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP113]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP114:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP105]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP106:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -10832,105 +10190,55 @@ int a;
// CHECK-32-NEXT: ret void
//
//
-// CHECK-32-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l37_omp_outlined
-// CHECK-32-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[A:%.*]]) #[[ATTR1]] {
-// CHECK-32-NEXT: entry:
-// CHECK-32-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
-// CHECK-32-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
-// CHECK-32-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[TMP:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[I:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[A_CASTED:%.*]] = alloca i32, align 4
-// CHECK-32-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [3 x ptr], align 4
-// CHECK-32-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
-// CHECK-32-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4
-// CHECK-32-NEXT: [[A1:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i32 4)
-// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
-// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK-32: cond.true:
-// CHECK-32-NEXT: br label [[COND_END:%.*]]
-// CHECK-32: cond.false:
-// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: br label [[COND_END]]
-// CHECK-32: cond.end:
-// CHECK-32-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
-// CHECK-32-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: [[CMP2:%.*]] = icmp slt i32 [[TMP5]], 10
-// CHECK-32-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[A1]], align 4
-// CHECK-32-NEXT: store i32 [[TMP8]], ptr [[A_CASTED]], align 4
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[A_CASTED]], align 4
-// CHECK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK-32-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4
-// CHECK-32-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK-32-NEXT: [[TMP13:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 4
-// CHECK-32-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK-32-NEXT: [[TMP15:%.*]] = inttoptr i32 [[TMP9]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 4
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l37_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3, i32 0)
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
-// CHECK-32-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP22]], 9
-// CHECK-32-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
-// CHECK-32: cond.true6:
-// CHECK-32-NEXT: br label [[COND_END8:%.*]]
-// CHECK-32: cond.false7:
-// CHECK-32-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: br label [[COND_END8]]
-// CHECK-32: cond.end8:
-// CHECK-32-NEXT: [[COND9:%.*]] = phi i32 [ 9, [[COND_TRUE6]] ], [ [[TMP23]], [[COND_FALSE7]] ]
-// CHECK-32-NEXT: store i32 [[COND9]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: store i32 [[TMP24]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK-32: omp.inner.for.end:
+// CHECK-32-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l37_omp_outlined
+// CHECK-32-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[A:%.*]]) #[[ATTR1]] {
+// CHECK-32-NEXT: entry:
+// CHECK-32-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK-32-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK-32-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
+// CHECK-32-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK-32-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK-32-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK-32-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK-32-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK-32-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK-32-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK-32-NEXT: [[A_CASTED:%.*]] = alloca i32, align 4
+// CHECK-32-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [3 x ptr], align 4
+// CHECK-32-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK-32-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK-32-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4
+// CHECK-32-NEXT: [[A1:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i32 4)
+// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK-32-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK-32-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
+// CHECK-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[A1]], align 4
+// CHECK-32-NEXT: store i32 [[TMP2]], ptr [[A_CASTED]], align 4
+// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[A_CASTED]], align 4
+// CHECK-32-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK-32-NEXT: [[TMP5:%.*]] = inttoptr i32 [[TMP0]] to ptr
+// CHECK-32-NEXT: store ptr [[TMP5]], ptr [[TMP4]], align 4
+// CHECK-32-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK-32-NEXT: [[TMP7:%.*]] = inttoptr i32 [[TMP1]] to ptr
+// CHECK-32-NEXT: store ptr [[TMP7]], ptr [[TMP6]], align 4
+// CHECK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK-32-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP3]] to ptr
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4
+// CHECK-32-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP11]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l37_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3, i32 0)
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
-// CHECK-32-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK-32-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
-// CHECK-32-NEXT: br i1 [[TMP26]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK-32-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
+// CHECK-32-NEXT: br i1 [[TMP13]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
// CHECK-32: .omp.lastprivate.then:
-// CHECK-32-NEXT: [[TMP27:%.*]] = load i32, ptr [[A1]], align 4
-// CHECK-32-NEXT: store i32 [[TMP27]], ptr [[A_ADDR]], align 4
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[A1]], align 4
+// CHECK-32-NEXT: store i32 [[TMP14]], ptr [[A_ADDR]], align 4
// CHECK-32-NEXT: br label [[DOTOMP_LASTPRIVATE_DONE]]
// CHECK-32: .omp.lastprivate.done:
// CHECK-32-NEXT: call void @__kmpc_free_shared(ptr [[A1]], i32 4)
@@ -10968,7 +10276,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -10996,7 +10304,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
// CHECK-32-NEXT: br i1 [[TMP12]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
@@ -11051,7 +10359,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -11111,7 +10419,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -11143,7 +10451,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -11179,7 +10487,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-NEXT: ret void
//
//
@@ -11226,7 +10534,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -11286,7 +10594,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -11318,7 +10626,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -11344,7 +10652,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-NEXT: ret void
//
//
@@ -11391,7 +10699,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -11451,7 +10759,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -11496,23 +10804,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP108:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP108]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP108]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP108]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP108]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP117:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP108]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP109:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -11565,7 +10873,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -11625,7 +10933,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -11670,23 +10978,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP119:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP119]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP111:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP111]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP119]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP111]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP119]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP111]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP119]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP111]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP119]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP120:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP111]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP112:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -11739,7 +11047,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -11799,7 +11107,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -11844,23 +11152,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP122:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP122]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP114:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP114]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP122]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP114]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP122]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP114]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP122]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP114]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP122]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP123:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP114]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP115:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -11913,7 +11221,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -11973,7 +11281,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -12018,23 +11326,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP125:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP125]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP117:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP117]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP125]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP117]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP125]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP117]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP125]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP117]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP125]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP126:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP117]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP118:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -12086,72 +11394,22 @@ int a;
// CHECK-32-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
-// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK-32: cond.true:
-// CHECK-32-NEXT: br label [[COND_END:%.*]]
-// CHECK-32: cond.false:
-// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: br label [[COND_END]]
-// CHECK-32: cond.end:
-// CHECK-32-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
-// CHECK-32-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP128:![0-9]+]]
-// CHECK-32-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
-// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK-32-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK-32-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l58_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
-// CHECK-32-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK-32: cond.true5:
-// CHECK-32-NEXT: br label [[COND_END7:%.*]]
-// CHECK-32: cond.false6:
-// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: br label [[COND_END7]]
-// CHECK-32: cond.end7:
-// CHECK-32-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP129:![0-9]+]]
-// CHECK-32: omp.inner.for.end:
+// CHECK-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK-32-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK-32-NEXT: [[TMP3:%.*]] = inttoptr i32 [[TMP0]] to ptr
+// CHECK-32-NEXT: store ptr [[TMP3]], ptr [[TMP2]], align 4
+// CHECK-32-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK-32-NEXT: [[TMP5:%.*]] = inttoptr i32 [[TMP1]] to ptr
+// CHECK-32-NEXT: store ptr [[TMP5]], ptr [[TMP4]], align 4
+// CHECK-32-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP7]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l58_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0)
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
-// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK-32-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
-// CHECK-32-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK-32-NEXT: [[TMP9:%.*]] = icmp ne i32 [[TMP8]], 0
+// CHECK-32-NEXT: br i1 [[TMP9]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK-32: .omp.final.then:
// CHECK-32-NEXT: store i32 10, ptr [[I]], align 4
// CHECK-32-NEXT: br label [[DOTOMP_FINAL_DONE]]
@@ -12187,33 +11445,33 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP131:![0-9]+]]
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP131]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120:![0-9]+]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP120]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp ule i32 [[TMP5]], [[TMP6]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP131]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP131]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP120]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP131]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP131]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP120]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP131]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP132:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP121:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP10]], 0
// CHECK-32-NEXT: br i1 [[TMP11]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -12269,7 +11527,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -12285,51 +11543,51 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP134:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123:![0-9]+]]
// CHECK-32-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP134]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP123]]
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-32-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32: cond.true5:
// CHECK-32-NEXT: br label [[COND_END7:%.*]]
// CHECK-32: cond.false6:
-// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-32-NEXT: br label [[COND_END7]]
// CHECK-32: cond.end7:
// CHECK-32-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP135:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP124:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -12368,7 +11626,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -12384,27 +11642,27 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137:![0-9]+]]
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP126:![0-9]+]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP126]]
// CHECK-32-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP126]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP126]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP126]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP138:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP126]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP127:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
// CHECK-32-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -12459,7 +11717,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -12475,31 +11733,31 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP140:![0-9]+]]
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP140]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP129:![0-9]+]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP129]]
// CHECK-32-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP140]]
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP140]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP129]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP129]]
// CHECK-32-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-NEXT: [[TMP10:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 4, !llvm.access.group [[ACC_GRP140]]
+// CHECK-32-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 4, !llvm.access.group [[ACC_GRP129]]
// CHECK-32-NEXT: [[TMP11:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP8]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4, !llvm.access.group [[ACC_GRP140]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l73_omp_outlined_omp_outlined, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l73_omp_outlined_omp_outlined_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP140]]
+// CHECK-32-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4, !llvm.access.group [[ACC_GRP129]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l73_omp_outlined_omp_outlined, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l73_omp_outlined_omp_outlined_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP129]]
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP140]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP140]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP129]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP129]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP140]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP141:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP129]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP130:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK-32-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -12541,33 +11799,33 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP143:![0-9]+]]
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP143]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP132:![0-9]+]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP132]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp ule i32 [[TMP5]], [[TMP6]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP143]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP132]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP143]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP132]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP143]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP143]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP132]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP132]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP143]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP144:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP132]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP133:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP10]], 0
// CHECK-32-NEXT: br i1 [[TMP11]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -12641,7 +11899,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -12657,51 +11915,51 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP146:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP135:![0-9]+]]
// CHECK-32-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l81_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP146]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l81_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP135]]
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-32-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32: cond.true5:
// CHECK-32-NEXT: br label [[COND_END7:%.*]]
// CHECK-32: cond.false6:
-// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-32-NEXT: br label [[COND_END7]]
// CHECK-32: cond.end7:
// CHECK-32-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP147:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP136:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -12753,23 +12011,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP149:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP149]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP138:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP138]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP149]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP138]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP149]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP138]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP149]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP138]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP149]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP150:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP138]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP139:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -12829,7 +12087,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -12845,51 +12103,51 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP152:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP141:![0-9]+]]
// CHECK-32-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l85_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP152]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l85_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP141]]
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-32-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32: cond.true5:
// CHECK-32-NEXT: br label [[COND_END7:%.*]]
// CHECK-32: cond.false6:
-// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-32-NEXT: br label [[COND_END7]]
// CHECK-32: cond.end7:
// CHECK-32-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP153:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP142:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -12941,23 +12199,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP155:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP155]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP144:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP144]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP155]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP144]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP155]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP144]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP155]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP144]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP155]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP156:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP144]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP145:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -13017,7 +12275,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -13033,51 +12291,51 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP158:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP147:![0-9]+]]
// CHECK-32-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l89_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP158]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l89_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP147]]
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-32-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32: cond.true5:
// CHECK-32-NEXT: br label [[COND_END7:%.*]]
// CHECK-32: cond.false6:
-// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-32-NEXT: br label [[COND_END7]]
// CHECK-32: cond.end7:
// CHECK-32-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP159:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP148:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -13129,23 +12387,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP161:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP161]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP150:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP150]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP161]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP150]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP161]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP150]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP161]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP150]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP161]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP162:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP150]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP151:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -13205,7 +12463,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -13221,51 +12479,51 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP164:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP153:![0-9]+]]
// CHECK-32-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l93_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP164]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l93_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP153]]
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-32-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32: cond.true5:
// CHECK-32-NEXT: br label [[COND_END7:%.*]]
// CHECK-32: cond.false6:
-// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-32-NEXT: br label [[COND_END7]]
// CHECK-32: cond.end7:
// CHECK-32-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP165:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP154:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -13317,23 +12575,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP167:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP167]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP156:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP156]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP167]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP156]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP167]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP156]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP167]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP156]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP167]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP168:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP156]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP157:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -13391,69 +12649,19 @@ int a;
// CHECK-32-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
-// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK-32: cond.true:
-// CHECK-32-NEXT: br label [[COND_END:%.*]]
-// CHECK-32: cond.false:
-// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: br label [[COND_END]]
-// CHECK-32: cond.end:
-// CHECK-32-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
-// CHECK-32-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
-// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK-32-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4
-// CHECK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK-32-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l97_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0)
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
-// CHECK-32-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK-32: cond.true5:
-// CHECK-32-NEXT: br label [[COND_END7:%.*]]
-// CHECK-32: cond.false6:
-// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: br label [[COND_END7]]
-// CHECK-32: cond.end7:
-// CHECK-32-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK-32: omp.inner.for.end:
+// CHECK-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK-32-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK-32-NEXT: [[TMP3:%.*]] = inttoptr i32 [[TMP0]] to ptr
+// CHECK-32-NEXT: store ptr [[TMP3]], ptr [[TMP2]], align 4
+// CHECK-32-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK-32-NEXT: [[TMP5:%.*]] = inttoptr i32 [[TMP1]] to ptr
+// CHECK-32-NEXT: store ptr [[TMP5]], ptr [[TMP4]], align 4
+// CHECK-32-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP7]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l97_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0)
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -13485,7 +12693,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -13511,7 +12719,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-NEXT: ret void
//
//
@@ -13558,7 +12766,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -13618,7 +12826,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -13650,7 +12858,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -13686,7 +12894,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-NEXT: ret void
//
//
@@ -13733,7 +12941,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -13793,7 +13001,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -13825,7 +13033,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -13851,7 +13059,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-NEXT: ret void
//
//
@@ -13898,7 +13106,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -13958,7 +13166,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -14003,23 +13211,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP170:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP170]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP159:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP159]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP170]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP159]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP170]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP159]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP170]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP159]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP170]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP171:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP159]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP160:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -14072,7 +13280,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -14132,7 +13340,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -14177,23 +13385,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP173:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP173]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP162:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP162]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP173]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP162]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP173]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP162]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP173]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP162]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP173]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP174:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP162]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP163:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -14246,7 +13454,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -14306,7 +13514,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -14351,23 +13559,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP176:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP176]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP165:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP165]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP176]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP165]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP176]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP165]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP176]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP165]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP176]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP177:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP165]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP166:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -14420,7 +13628,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -14480,7 +13688,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -14525,23 +13733,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP179:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP179]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP168:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP168]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP179]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP168]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP179]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP168]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP179]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP168]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP179]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP180:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP168]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP169:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -14592,69 +13800,19 @@ int a;
// CHECK-32-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
-// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK-32: cond.true:
-// CHECK-32-NEXT: br label [[COND_END:%.*]]
-// CHECK-32: cond.false:
-// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: br label [[COND_END]]
-// CHECK-32: cond.end:
-// CHECK-32-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
-// CHECK-32-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
-// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK-32-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4
-// CHECK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK-32-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4
-// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l125_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0)
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
-// CHECK-32-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK-32: cond.true5:
-// CHECK-32-NEXT: br label [[COND_END7:%.*]]
-// CHECK-32: cond.false6:
-// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: br label [[COND_END7]]
-// CHECK-32: cond.end7:
-// CHECK-32-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK-32: omp.inner.for.end:
+// CHECK-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK-32-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK-32-NEXT: [[TMP3:%.*]] = inttoptr i32 [[TMP0]] to ptr
+// CHECK-32-NEXT: store ptr [[TMP3]], ptr [[TMP2]], align 4
+// CHECK-32-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK-32-NEXT: [[TMP5:%.*]] = inttoptr i32 [[TMP1]] to ptr
+// CHECK-32-NEXT: store ptr [[TMP5]], ptr [[TMP4]], align 4
+// CHECK-32-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
+// CHECK-32-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP7]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l125_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0)
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -14686,7 +13844,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -14712,7 +13870,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-NEXT: ret void
//
//
@@ -14759,7 +13917,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -14819,7 +13977,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -14851,7 +14009,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -14887,7 +14045,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-NEXT: ret void
//
//
@@ -14934,7 +14092,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -14994,7 +14152,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -15026,7 +14184,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -15052,7 +14210,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-NEXT: ret void
//
//
@@ -15099,7 +14257,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -15159,7 +14317,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -15204,23 +14362,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP182:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP182]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP171:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP171]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP182]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP171]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP182]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP171]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP182]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP171]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP182]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP183:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP171]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP172:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -15273,7 +14431,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -15333,7 +14491,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -15378,23 +14536,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP185:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP185]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP174:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP174]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP185]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP174]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP185]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP174]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP185]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP174]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP185]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP186:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP174]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP175:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -15447,7 +14605,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -15507,7 +14665,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -15552,23 +14710,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP188:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP188]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP177:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP177]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP188]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP177]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP188]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP177]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP188]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP177]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP188]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP189:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP177]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP178:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -15621,7 +14779,7 @@ int a;
// CHECK-32-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -15681,7 +14839,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -15726,23 +14884,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP191:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP191]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP180:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP180]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP191]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP180]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP191]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP180]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP191]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP180]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP191]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP192:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP180]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP181:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -15795,7 +14953,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-32: omp.dispatch.cond:
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -15848,7 +15006,7 @@ int a;
// CHECK-32-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -15890,7 +15048,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -15926,7 +15084,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -15968,7 +15126,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-32: omp.dispatch.cond:
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -16021,7 +15179,7 @@ int a;
// CHECK-32-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -16074,23 +15232,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP194:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP194]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP183:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP183]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP194]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP183]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP194]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP183]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP194]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP183]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP194]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP195:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP183]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP184:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -16149,23 +15307,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP197:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP197]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP186:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP186]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP197]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP186]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP197]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP186]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP197]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP186]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP197]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP198:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP186]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP187:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -16224,23 +15382,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP200:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP200]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP189:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP189]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP200]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP189]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP200]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP189]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP200]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP189]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP200]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP201:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP189]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP190:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -16299,23 +15457,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP203:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP203]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP192:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP192]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP203]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP192]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP203]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP192]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP203]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP192]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP203]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP204:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP192]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP193:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -16368,7 +15526,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-32: omp.dispatch.cond:
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -16391,23 +15549,23 @@ int a;
// CHECK-32: omp.dispatch.body:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP206:![0-9]+]]
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP206]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP195:![0-9]+]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP195]]
// CHECK-32-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK-32-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP206]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP195]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP206]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP195]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP206]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP195]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP206]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP207:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP195]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP196:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -16421,7 +15579,7 @@ int a;
// CHECK-32-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK-32-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -16471,7 +15629,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -16487,27 +15645,27 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP209:![0-9]+]]
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP209]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP198:![0-9]+]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP198]]
// CHECK-32-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP209]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP198]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP209]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP198]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP209]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP198]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP209]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP210:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP198]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP199:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP10:%.*]] = icmp ne i32 [[TMP9]], 0
// CHECK-32-NEXT: br i1 [[TMP10]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -16557,7 +15715,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-32: omp.dispatch.cond:
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -16580,23 +15738,23 @@ int a;
// CHECK-32: omp.dispatch.body:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP212:![0-9]+]]
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP212]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP201:![0-9]+]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP201]]
// CHECK-32-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK-32-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP212]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP201]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP212]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP201]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP212]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP201]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP212]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP213:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP201]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP202:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -16610,7 +15768,7 @@ int a;
// CHECK-32-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK-32-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -16671,23 +15829,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP215:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP215]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP204:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP204]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP215]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP204]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP215]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP204]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP215]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP204]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP215]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP216:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP204]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP205:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -16754,23 +15912,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP218:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP218]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP207:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP207]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP218]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP207]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP218]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP207]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP218]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP207]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP218]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP219:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP207]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP208:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -16837,23 +15995,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP210:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP210]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP210]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP210]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP210]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP222:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP210]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP211:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -16920,23 +16078,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP224:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP224]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP213:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP213]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP224]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP213]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP224]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP213]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP224]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP213]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP224]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP225:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP213]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP214:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -17003,24 +16161,24 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP216:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP216]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP216]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP216]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP216]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
-// CHECK-32-NEXT: call void @__kmpc_dispatch_fini_4(ptr @[[GLOB1]], i32 [[TMP1]]), !llvm.access.group [[ACC_GRP227]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP228:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP216]]
+// CHECK-32-NEXT: call void @__kmpc_dispatch_fini_4(ptr @[[GLOB1]], i32 [[TMP1]]), !llvm.access.group [[ACC_GRP216]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP217:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -17076,7 +16234,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -17092,27 +16250,27 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP230:![0-9]+]]
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP230]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP219:![0-9]+]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP219]]
// CHECK-32-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK-32-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP230]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP219]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP230]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP219]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP230]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP219]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP230]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP231:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP219]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP220:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP10:%.*]] = icmp ne i32 [[TMP9]], 0
// CHECK-32-NEXT: br i1 [[TMP10]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -17162,7 +16320,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-32: omp.dispatch.cond:
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -17185,23 +16343,23 @@ int a;
// CHECK-32: omp.dispatch.body:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP233:![0-9]+]]
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP233]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP222:![0-9]+]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP222]]
// CHECK-32-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK-32-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP233]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP222]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP233]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP222]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP233]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP222]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP233]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP234:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP222]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP223:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -17215,7 +16373,7 @@ int a;
// CHECK-32-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK-32-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -17276,23 +16434,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP236:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP236]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP225:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP225]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP236]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP225]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP236]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP225]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP236]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP225]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP236]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP237:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP225]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP226:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -17359,23 +16517,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP239:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP239]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP228:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP228]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP239]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP228]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP239]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP228]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP239]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP228]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP239]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP240:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP228]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP229:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -17442,23 +16600,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP242:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP242]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP231]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP242]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP242]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP231]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP242]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP242]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP243:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP232:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -17525,23 +16683,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP245:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP245]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP234:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP234]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP245]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP234]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP245]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP234]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP245]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP234]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP245]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP246:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP234]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP235:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -17597,7 +16755,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-32: omp.dispatch.cond:
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -17650,7 +16808,7 @@ int a;
// CHECK-32-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -17692,7 +16850,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -17728,7 +16886,7 @@ int a;
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -17770,7 +16928,7 @@ int a;
// CHECK-32-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-32: omp.dispatch.cond:
// CHECK-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -17823,7 +16981,7 @@ int a;
// CHECK-32-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
-// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -17876,23 +17034,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP248:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP248]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP237:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP237]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP248]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP237]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP248]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP237]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP248]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP237]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP248]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP249:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP237]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP238:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -17951,23 +17109,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP251:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP251]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP240:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP240]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP251]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP240]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP251]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP240]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP251]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP240]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP251]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP252:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP240]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP241:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -18026,23 +17184,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP254:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP254]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP243:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP243]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP254]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP243]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP254]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP243]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP254]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP243]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP254]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP255:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP243]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP244:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -18101,23 +17259,23 @@ int a;
// CHECK-32-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32: omp.inner.for.cond:
-// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP257:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP257]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP246:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP246]]
// CHECK-32-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32: omp.inner.for.body:
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP257]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP246]]
// CHECK-32-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP257]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP246]]
// CHECK-32-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32: omp.body.continue:
// CHECK-32-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32: omp.inner.for.inc:
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP257]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP246]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP257]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP258:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP246]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP247:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -18171,8 +17329,6 @@ int a;
// CHECK-32-EX-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK-32-EX-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK-32-EX-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [3 x ptr], align 4
-// CHECK-32-EX-NEXT: [[DOTCAPTURE_EXPR__CASTED15:%.*]] = alloca i32, align 4
-// CHECK-32-EX-NEXT: [[CAPTURED_VARS_ADDRS17:%.*]] = alloca [3 x ptr], align 4
// CHECK-32-EX-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
@@ -18181,249 +17337,33 @@ int a;
// CHECK-32-EX-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
-// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK-32-EX: cond.true:
-// CHECK-32-EX-NEXT: br label [[COND_END:%.*]]
-// CHECK-32-EX: cond.false:
-// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: br label [[COND_END]]
-// CHECK-32-EX: cond.end:
-// CHECK-32-EX-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
-// CHECK-32-EX-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP5]], 0
-// CHECK-32-EX-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
-// CHECK-32-EX: omp_if.then:
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66:![0-9]+]]
-// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP6]], 10
-// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[LOADEDV2:%.*]] = icmp ne i8 [[TMP9]], 0
-// CHECK-32-EX-NEXT: [[STOREDV:%.*]] = zext i1 [[LOADEDV2]] to i8
-// CHECK-32-EX-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = inttoptr i32 [[TMP8]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = inttoptr i32 [[TMP10]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[LOADEDV3:%.*]] = icmp ne i8 [[TMP17]], 0
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = zext i1 [[LOADEDV3]] to i32
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 [[TMP18]], i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l15_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3, i32 0), !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP21]], [[TMP22]]
-// CHECK-32-EX-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP23]], [[TMP24]]
-// CHECK-32-EX-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP25]], 9
-// CHECK-32-EX-NEXT: br i1 [[CMP6]], label [[COND_TRUE7:%.*]], label [[COND_FALSE8:%.*]]
-// CHECK-32-EX: cond.true7:
-// CHECK-32-EX-NEXT: br label [[COND_END9:%.*]]
-// CHECK-32-EX: cond.false8:
-// CHECK-32-EX-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: br label [[COND_END9]]
-// CHECK-32-EX: cond.end9:
-// CHECK-32-EX-NEXT: [[COND10:%.*]] = phi i32 [ 9, [[COND_TRUE7]] ], [ [[TMP26]], [[COND_FALSE8]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND10]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: store i32 [[TMP27]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP67:![0-9]+]]
-// CHECK-32-EX: omp.inner.for.end:
-// CHECK-32-EX-NEXT: br label [[OMP_IF_END:%.*]]
-// CHECK-32-EX: omp_if.else:
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND11:%.*]]
-// CHECK-32-EX: omp.inner.for.cond11:
-// CHECK-32-EX-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: [[CMP12:%.*]] = icmp slt i32 [[TMP28]], 10
-// CHECK-32-EX-NEXT: br i1 [[CMP12]], label [[OMP_INNER_FOR_BODY13:%.*]], label [[OMP_INNER_FOR_END28:%.*]]
-// CHECK-32-EX: omp.inner.for.body13:
-// CHECK-32-EX-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP31:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
-// CHECK-32-EX-NEXT: [[LOADEDV14:%.*]] = icmp ne i8 [[TMP31]], 0
-// CHECK-32-EX-NEXT: [[STOREDV16:%.*]] = zext i1 [[LOADEDV14]] to i8
-// CHECK-32-EX-NEXT: store i8 [[STOREDV16]], ptr [[DOTCAPTURE_EXPR__CASTED15]], align 1
-// CHECK-32-EX-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED15]], align 4
-// CHECK-32-EX-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS17]], i32 0, i32 0
-// CHECK-32-EX-NEXT: [[TMP34:%.*]] = inttoptr i32 [[TMP29]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP34]], ptr [[TMP33]], align 4
-// CHECK-32-EX-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS17]], i32 0, i32 1
-// CHECK-32-EX-NEXT: [[TMP36:%.*]] = inttoptr i32 [[TMP30]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP36]], ptr [[TMP35]], align 4
-// CHECK-32-EX-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS17]], i32 0, i32 2
-// CHECK-32-EX-NEXT: [[TMP38:%.*]] = inttoptr i32 [[TMP32]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP38]], ptr [[TMP37]], align 4
-// CHECK-32-EX-NEXT: [[TMP39:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
-// CHECK-32-EX-NEXT: [[LOADEDV18:%.*]] = icmp ne i8 [[TMP39]], 0
-// CHECK-32-EX-NEXT: [[TMP40:%.*]] = zext i1 [[LOADEDV18]] to i32
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 [[TMP40]], i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l15_omp_outlined_omp_outlined1, ptr null, ptr [[CAPTURED_VARS_ADDRS17]], i32 3, i32 0)
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC19:%.*]]
-// CHECK-32-EX: omp.inner.for.inc19:
-// CHECK-32-EX-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: [[TMP42:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-EX-NEXT: [[ADD20:%.*]] = add nsw i32 [[TMP41]], [[TMP42]]
-// CHECK-32-EX-NEXT: store i32 [[ADD20]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: [[TMP43:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: [[TMP44:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-EX-NEXT: [[ADD21:%.*]] = add nsw i32 [[TMP43]], [[TMP44]]
-// CHECK-32-EX-NEXT: store i32 [[ADD21]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: [[TMP45:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP46:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-EX-NEXT: [[ADD22:%.*]] = add nsw i32 [[TMP45]], [[TMP46]]
-// CHECK-32-EX-NEXT: store i32 [[ADD22]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP47:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[CMP23:%.*]] = icmp sgt i32 [[TMP47]], 9
-// CHECK-32-EX-NEXT: br i1 [[CMP23]], label [[COND_TRUE24:%.*]], label [[COND_FALSE25:%.*]]
-// CHECK-32-EX: cond.true24:
-// CHECK-32-EX-NEXT: br label [[COND_END26:%.*]]
-// CHECK-32-EX: cond.false25:
-// CHECK-32-EX-NEXT: [[TMP48:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: br label [[COND_END26]]
-// CHECK-32-EX: cond.end26:
-// CHECK-32-EX-NEXT: [[COND27:%.*]] = phi i32 [ 9, [[COND_TRUE24]] ], [ [[TMP48]], [[COND_FALSE25]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND27]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP49:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: store i32 [[TMP49]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND11]], !llvm.loop [[LOOP70:![0-9]+]]
-// CHECK-32-EX: omp.inner.for.end28:
-// CHECK-32-EX-NEXT: br label [[OMP_IF_END]]
-// CHECK-32-EX: omp_if.end:
-// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
-// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
-// CHECK-32-EX-NEXT: [[TMP50:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK-32-EX-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0
-// CHECK-32-EX-NEXT: br i1 [[TMP51]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
-// CHECK-32-EX: .omp.final.then:
-// CHECK-32-EX-NEXT: store i32 10, ptr [[I]], align 4
-// CHECK-32-EX-NEXT: br label [[DOTOMP_FINAL_DONE]]
-// CHECK-32-EX: .omp.final.done:
-// CHECK-32-EX-NEXT: ret void
-//
-//
-// CHECK-32-EX-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l15_omp_outlined_omp_outlined
-// CHECK-32-EX-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
-// CHECK-32-EX-NEXT: entry:
-// CHECK-32-EX-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
-// CHECK-32-EX-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
-// CHECK-32-EX-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
-// CHECK-32-EX-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
-// CHECK-32-EX-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
-// CHECK-32-EX-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
-// CHECK-32-EX-NEXT: [[TMP:%.*]] = alloca i32, align 4
-// CHECK-32-EX-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
-// CHECK-32-EX-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
-// CHECK-32-EX-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
-// CHECK-32-EX-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
-// CHECK-32-EX-NEXT: [[I:%.*]] = alloca i32, align 4
-// CHECK-32-EX-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-EX-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
-// CHECK-32-EX-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
-// CHECK-32-EX-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
-// CHECK-32-EX-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
-// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
-// CHECK-32-EX-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
-// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
-// CHECK-32-EX-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_LB]], align 4
-// CHECK-32-EX-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_UB]], align 4
-// CHECK-32-EX-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
// CHECK-32-EX-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP2]], 0
-// CHECK-32-EX-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
-// CHECK-32-EX: omp_if.then:
-// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
-// CHECK-32-EX-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp ule i32 [[TMP6]], [[TMP7]]
-// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
-// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
-// CHECK-32-EX: omp.body.continue:
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP73:![0-9]+]]
-// CHECK-32-EX: omp.inner.for.end:
-// CHECK-32-EX-NEXT: br label [[OMP_IF_END:%.*]]
-// CHECK-32-EX: omp_if.else:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP12]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
-// CHECK-32-EX-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND2:%.*]]
-// CHECK-32-EX: omp.inner.for.cond2:
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
-// CHECK-32-EX-NEXT: [[CMP3:%.*]] = icmp ule i32 [[TMP14]], [[TMP15]]
-// CHECK-32-EX-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY4:%.*]], label [[OMP_INNER_FOR_END10:%.*]]
-// CHECK-32-EX: omp.inner.for.body4:
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: [[MUL5:%.*]] = mul nsw i32 [[TMP16]], 1
-// CHECK-32-EX-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
-// CHECK-32-EX-NEXT: store i32 [[ADD6]], ptr [[I]], align 4
-// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE7:%.*]]
-// CHECK-32-EX: omp.body.continue7:
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC8:%.*]]
-// CHECK-32-EX: omp.inner.for.inc8:
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-EX-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP17]], [[TMP18]]
-// CHECK-32-EX-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND2]], !llvm.loop [[LOOP75:![0-9]+]]
-// CHECK-32-EX: omp.inner.for.end10:
-// CHECK-32-EX-NEXT: br label [[OMP_IF_END]]
-// CHECK-32-EX: omp_if.end:
+// CHECK-32-EX-NEXT: [[STOREDV:%.*]] = zext i1 [[LOADEDV]] to i8
+// CHECK-32-EX-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1
+// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = inttoptr i32 [[TMP0]] to ptr
+// CHECK-32-EX-NEXT: store ptr [[TMP5]], ptr [[TMP4]], align 4
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = inttoptr i32 [[TMP1]] to ptr
+// CHECK-32-EX-NEXT: store ptr [[TMP7]], ptr [[TMP6]], align 4
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP3]] to ptr
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
+// CHECK-32-EX-NEXT: [[LOADEDV1:%.*]] = icmp ne i8 [[TMP10]], 0
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = zext i1 [[LOADEDV1]] to i32
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP13]], i32 [[TMP11]], i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l15_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3, i32 0)
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP20]])
-// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK-32-EX-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
-// CHECK-32-EX-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
+// CHECK-32-EX-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK-32-EX: .omp.final.then:
// CHECK-32-EX-NEXT: store i32 10, ptr [[I]], align 4
// CHECK-32-EX-NEXT: br label [[DOTOMP_FINAL_DONE]]
@@ -18431,7 +17371,7 @@ int a;
// CHECK-32-EX-NEXT: ret void
//
//
-// CHECK-32-EX-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l15_omp_outlined_omp_outlined1
+// CHECK-32-EX-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l15_omp_outlined_omp_outlined
// CHECK-32-EX-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] {
// CHECK-32-EX-NEXT: entry:
// CHECK-32-EX-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
@@ -18465,35 +17405,35 @@ int a;
// CHECK-32-EX: omp_if.then:
// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-EX-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP76:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP76]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP66]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp ule i32 [[TMP6]], [[TMP7]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP76]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP76]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP66]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP76]]
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP76]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP66]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP76]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP77:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP66]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP67:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK-32-EX: omp_if.else:
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP12]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP12]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-EX-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND2:%.*]]
@@ -18515,7 +17455,7 @@ int a;
// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-32-EX-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP17]], [[TMP18]]
// CHECK-32-EX-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND2]], !llvm.loop [[LOOP79:![0-9]+]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND2]], !llvm.loop [[LOOP70:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end10:
// CHECK-32-EX-NEXT: br label [[OMP_IF_END]]
// CHECK-32-EX: omp_if.end:
@@ -18523,7 +17463,7 @@ int a;
// CHECK-32-EX: omp.loop.exit:
// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP20]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP20]])
// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -18577,7 +17517,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -18593,51 +17533,51 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP80:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72:![0-9]+]]
// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-32-EX-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-EX-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l18_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP80]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l18_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP72]]
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-32-EX-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32-EX: cond.true5:
// CHECK-32-EX-NEXT: br label [[COND_END7:%.*]]
// CHECK-32-EX: cond.false6:
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
// CHECK-32-EX-NEXT: br label [[COND_END7]]
// CHECK-32-EX: cond.end7:
// CHECK-32-EX-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP80]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP81:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP72]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP73:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -18676,7 +17616,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -18692,27 +17632,27 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP83:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP83]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP75:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP75]]
// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP83]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP75]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP83]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP75]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP83]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP75]]
// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP83]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP84:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP75]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP76:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -18766,7 +17706,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -18782,51 +17722,51 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP86:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP78:![0-9]+]]
// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-32-EX-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-EX-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l21_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP86]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l21_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP78]]
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-32-EX-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32-EX: cond.true5:
// CHECK-32-EX-NEXT: br label [[COND_END7:%.*]]
// CHECK-32-EX: cond.false6:
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
// CHECK-32-EX-NEXT: br label [[COND_END7]]
// CHECK-32-EX: cond.end7:
// CHECK-32-EX-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP86]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP87:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP78]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP79:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -18865,33 +17805,33 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP89:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP89]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP81:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP81]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp ule i32 [[TMP5]], [[TMP6]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP89]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP81]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP89]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP81]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP89]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP89]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP81]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP81]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP89]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP90:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP81]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP82:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP10]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP11]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -18945,7 +17885,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -18961,51 +17901,51 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP92:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84:![0-9]+]]
// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-32-EX-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-EX-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l24_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP92]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l24_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP84]]
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-32-EX-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32-EX: cond.true5:
// CHECK-32-EX-NEXT: br label [[COND_END7:%.*]]
// CHECK-32-EX: cond.false6:
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
// CHECK-32-EX-NEXT: br label [[COND_END7]]
// CHECK-32-EX: cond.end7:
// CHECK-32-EX-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP92]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP93:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP85:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -19057,23 +17997,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP95:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP95]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP87:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP87]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP95]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP87]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP95]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP87]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP95]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP87]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP95]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP96:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP87]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP88:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -19133,7 +18073,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -19149,51 +18089,51 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP98:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP90:![0-9]+]]
// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-32-EX-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-EX-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l27_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP98]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l27_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP90]]
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-32-EX-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32-EX: cond.true5:
// CHECK-32-EX-NEXT: br label [[COND_END7:%.*]]
// CHECK-32-EX: cond.false6:
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
// CHECK-32-EX-NEXT: br label [[COND_END7]]
// CHECK-32-EX: cond.end7:
// CHECK-32-EX-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP98]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP99:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP90]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP91:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -19245,23 +18185,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP101:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP101]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP93:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP93]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP101]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP93]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP101]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP93]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP101]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP93]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP101]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP102:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP93]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP94:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -19321,7 +18261,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -19337,51 +18277,51 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP104:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP96:![0-9]+]]
// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-32-EX-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-EX-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l30_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP104]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l30_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP96]]
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-32-EX-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32-EX: cond.true5:
// CHECK-32-EX-NEXT: br label [[COND_END7:%.*]]
// CHECK-32-EX: cond.false6:
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
// CHECK-32-EX-NEXT: br label [[COND_END7]]
// CHECK-32-EX: cond.end7:
// CHECK-32-EX-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP104]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP105:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP96]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP97:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -19433,23 +18373,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP107:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP107]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP99:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP99]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP107]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP99]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP107]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP99]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP107]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP99]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP107]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP108:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP99]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP100:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -19509,7 +18449,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -19525,51 +18465,51 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP110:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP102:![0-9]+]]
// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-32-EX-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-EX-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l33_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP110]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l33_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP102]]
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-32-EX-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32-EX: cond.true5:
// CHECK-32-EX-NEXT: br label [[COND_END7:%.*]]
// CHECK-32-EX: cond.false6:
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
// CHECK-32-EX-NEXT: br label [[COND_END7]]
// CHECK-32-EX: cond.end7:
// CHECK-32-EX-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP110]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP111:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP102]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP103:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -19621,23 +18561,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP113:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP113]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP105:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP105]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP113]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP105]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP113]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP105]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP113]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP105]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP113]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP114:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP105]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP106:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -19705,81 +18645,31 @@ int a;
// CHECK-32-EX-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
-// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK-32-EX: cond.true:
-// CHECK-32-EX-NEXT: br label [[COND_END:%.*]]
-// CHECK-32-EX: cond.false:
-// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: br label [[COND_END]]
-// CHECK-32-EX: cond.end:
-// CHECK-32-EX-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: [[CMP2:%.*]] = icmp slt i32 [[TMP5]], 10
-// CHECK-32-EX-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[A1]], align 4
-// CHECK-32-EX-NEXT: store i32 [[TMP8]], ptr [[A_CASTED]], align 4
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[A_CASTED]], align 4
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 4
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = inttoptr i32 [[TMP9]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l37_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3, i32 0)
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-EX-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
-// CHECK-32-EX-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP22]], 9
-// CHECK-32-EX-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
-// CHECK-32-EX: cond.true6:
-// CHECK-32-EX-NEXT: br label [[COND_END8:%.*]]
-// CHECK-32-EX: cond.false7:
-// CHECK-32-EX-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: br label [[COND_END8]]
-// CHECK-32-EX: cond.end8:
-// CHECK-32-EX-NEXT: [[COND9:%.*]] = phi i32 [ 9, [[COND_TRUE6]] ], [ [[TMP23]], [[COND_FALSE7]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND9]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: store i32 [[TMP24]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK-32-EX: omp.inner.for.end:
+// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[A1]], align 4
+// CHECK-32-EX-NEXT: store i32 [[TMP2]], ptr [[A_CASTED]], align 4
+// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[A_CASTED]], align 4
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = inttoptr i32 [[TMP0]] to ptr
+// CHECK-32-EX-NEXT: store ptr [[TMP5]], ptr [[TMP4]], align 4
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = inttoptr i32 [[TMP1]] to ptr
+// CHECK-32-EX-NEXT: store ptr [[TMP7]], ptr [[TMP6]], align 4
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP3]] to ptr
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP11]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l37_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3, i32 0)
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
-// CHECK-32-EX-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK-32-EX-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
-// CHECK-32-EX-NEXT: br i1 [[TMP26]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
+// CHECK-32-EX-NEXT: br i1 [[TMP13]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
// CHECK-32-EX: .omp.lastprivate.then:
-// CHECK-32-EX-NEXT: [[TMP27:%.*]] = load i32, ptr [[A1]], align 4
-// CHECK-32-EX-NEXT: store i32 [[TMP27]], ptr [[A_ADDR]], align 4
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[A1]], align 4
+// CHECK-32-EX-NEXT: store i32 [[TMP14]], ptr [[A_ADDR]], align 4
// CHECK-32-EX-NEXT: br label [[DOTOMP_LASTPRIVATE_DONE]]
// CHECK-32-EX: .omp.lastprivate.done:
// CHECK-32-EX-NEXT: call void @__kmpc_free_shared(ptr [[A1]], i32 4)
@@ -19817,7 +18707,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -19845,7 +18735,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP12]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
@@ -19900,7 +18790,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -19960,7 +18850,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -19992,7 +18882,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -20028,7 +18918,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -20075,7 +18965,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -20135,7 +19025,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -20167,7 +19057,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -20193,7 +19083,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -20240,7 +19130,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -20300,7 +19190,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -20345,23 +19235,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP108:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP108]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP108]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP108]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP108]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP117:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP108]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP109:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -20414,7 +19304,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -20474,7 +19364,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -20519,23 +19409,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP119:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP119]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP111:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP111]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP119]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP111]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP119]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP111]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP119]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP111]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP119]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP120:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP111]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP112:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -20588,7 +19478,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -20648,7 +19538,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -20693,23 +19583,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP122:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP122]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP114:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP114]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP122]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP114]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP122]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP114]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP122]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP114]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP122]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP123:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP114]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP115:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -20762,7 +19652,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -20822,7 +19712,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -20867,23 +19757,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP125:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP125]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP117:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP117]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP125]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP117]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP125]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP117]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP125]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP117]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP125]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP126:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP117]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP118:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -20935,72 +19825,22 @@ int a;
// CHECK-32-EX-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
-// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK-32-EX: cond.true:
-// CHECK-32-EX-NEXT: br label [[COND_END:%.*]]
-// CHECK-32-EX: cond.false:
-// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: br label [[COND_END]]
-// CHECK-32-EX: cond.end:
-// CHECK-32-EX-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP128:![0-9]+]]
-// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
-// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l58_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
-// CHECK-32-EX-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK-32-EX: cond.true5:
-// CHECK-32-EX-NEXT: br label [[COND_END7:%.*]]
-// CHECK-32-EX: cond.false6:
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: br label [[COND_END7]]
-// CHECK-32-EX: cond.end7:
-// CHECK-32-EX-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP128]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP129:![0-9]+]]
-// CHECK-32-EX: omp.inner.for.end:
+// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK-32-EX-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK-32-EX-NEXT: [[TMP3:%.*]] = inttoptr i32 [[TMP0]] to ptr
+// CHECK-32-EX-NEXT: store ptr [[TMP3]], ptr [[TMP2]], align 4
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = inttoptr i32 [[TMP1]] to ptr
+// CHECK-32-EX-NEXT: store ptr [[TMP5]], ptr [[TMP4]], align 4
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP7]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l58_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0)
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
-// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK-32-EX-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
-// CHECK-32-EX-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = icmp ne i32 [[TMP8]], 0
+// CHECK-32-EX-NEXT: br i1 [[TMP9]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK-32-EX: .omp.final.then:
// CHECK-32-EX-NEXT: store i32 10, ptr [[I]], align 4
// CHECK-32-EX-NEXT: br label [[DOTOMP_FINAL_DONE]]
@@ -21036,33 +19876,33 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP131:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP131]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP120]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp ule i32 [[TMP5]], [[TMP6]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP131]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP131]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP120]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP131]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP131]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP120]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP131]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP132:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP121:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP10]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP11]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -21118,7 +19958,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -21134,51 +19974,51 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP134:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123:![0-9]+]]
// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-32-EX-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-EX-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP134]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l66_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP123]]
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-32-EX-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32-EX: cond.true5:
// CHECK-32-EX-NEXT: br label [[COND_END7:%.*]]
// CHECK-32-EX: cond.false6:
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
// CHECK-32-EX-NEXT: br label [[COND_END7]]
// CHECK-32-EX: cond.end7:
// CHECK-32-EX-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP134]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP135:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP124:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -21217,7 +20057,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -21233,27 +20073,27 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP126:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP126]]
// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP126]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP126]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP126]]
// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP138:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP126]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP127:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -21308,7 +20148,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -21324,31 +20164,31 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP140:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP140]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP129:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP129]]
// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP140]]
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP140]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP129]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP129]]
// CHECK-32-EX-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 4, !llvm.access.group [[ACC_GRP140]]
+// CHECK-32-EX-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 4, !llvm.access.group [[ACC_GRP129]]
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-EX-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP8]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4, !llvm.access.group [[ACC_GRP140]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l73_omp_outlined_omp_outlined, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l73_omp_outlined_omp_outlined_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP140]]
+// CHECK-32-EX-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4, !llvm.access.group [[ACC_GRP129]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l73_omp_outlined_omp_outlined, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l73_omp_outlined_omp_outlined_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP129]]
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP140]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP140]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP129]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP129]]
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP140]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP141:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP129]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP130:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -21390,33 +20230,33 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP143:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP143]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP132:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP132]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp ule i32 [[TMP5]], [[TMP6]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP143]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP132]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP143]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP132]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP143]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP143]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP132]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP132]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP143]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP144:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP132]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP133:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP10]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP11]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -21490,7 +20330,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -21506,51 +20346,51 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP146:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP135:![0-9]+]]
// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-32-EX-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-EX-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l81_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP146]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l81_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP135]]
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-32-EX-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32-EX: cond.true5:
// CHECK-32-EX-NEXT: br label [[COND_END7:%.*]]
// CHECK-32-EX: cond.false6:
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
// CHECK-32-EX-NEXT: br label [[COND_END7]]
// CHECK-32-EX: cond.end7:
// CHECK-32-EX-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP146]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP147:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP135]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP136:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -21602,23 +20442,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP149:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP149]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP138:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP138]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP149]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP138]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP149]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP138]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP149]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP138]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP149]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP150:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP138]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP139:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -21678,7 +20518,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -21694,51 +20534,51 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP152:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP141:![0-9]+]]
// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-32-EX-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-EX-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l85_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP152]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l85_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP141]]
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-32-EX-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32-EX: cond.true5:
// CHECK-32-EX-NEXT: br label [[COND_END7:%.*]]
// CHECK-32-EX: cond.false6:
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
// CHECK-32-EX-NEXT: br label [[COND_END7]]
// CHECK-32-EX: cond.end7:
// CHECK-32-EX-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP152]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP153:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP141]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP142:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -21790,23 +20630,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP155:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP155]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP144:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP144]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP155]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP144]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP155]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP144]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP155]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP144]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP155]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP156:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP144]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP145:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -21866,7 +20706,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -21882,51 +20722,51 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP158:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP147:![0-9]+]]
// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-32-EX-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-EX-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l89_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP158]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l89_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP147]]
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-32-EX-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32-EX: cond.true5:
// CHECK-32-EX-NEXT: br label [[COND_END7:%.*]]
// CHECK-32-EX: cond.false6:
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
// CHECK-32-EX-NEXT: br label [[COND_END7]]
// CHECK-32-EX: cond.end7:
// CHECK-32-EX-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP158]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP159:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP147]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP148:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -21978,23 +20818,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP161:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP161]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP150:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP150]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP161]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP150]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP161]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP150]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP161]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP150]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP161]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP162:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP150]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP151:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -22054,7 +20894,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -22070,51 +20910,51 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP164:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP153:![0-9]+]]
// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-32-EX-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK-32-EX-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK-32-EX-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l93_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP164]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l93_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0), !llvm.access.group [[ACC_GRP153]]
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-32-EX-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK-32-EX: cond.true5:
// CHECK-32-EX-NEXT: br label [[COND_END7:%.*]]
// CHECK-32-EX: cond.false6:
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
// CHECK-32-EX-NEXT: br label [[COND_END7]]
// CHECK-32-EX: cond.end7:
// CHECK-32-EX-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP164]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP165:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP153]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP154:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -22166,23 +21006,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP167:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP167]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP156:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP156]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP167]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP156]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP167]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP156]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP167]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP156]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP167]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP168:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP156]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP157:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -22240,69 +21080,19 @@ int a;
// CHECK-32-EX-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
-// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK-32-EX: cond.true:
-// CHECK-32-EX-NEXT: br label [[COND_END:%.*]]
-// CHECK-32-EX: cond.false:
-// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: br label [[COND_END]]
-// CHECK-32-EX: cond.end:
-// CHECK-32-EX-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
-// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l97_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0)
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
-// CHECK-32-EX-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK-32-EX: cond.true5:
-// CHECK-32-EX-NEXT: br label [[COND_END7:%.*]]
-// CHECK-32-EX: cond.false6:
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: br label [[COND_END7]]
-// CHECK-32-EX: cond.end7:
-// CHECK-32-EX-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK-32-EX: omp.inner.for.end:
+// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK-32-EX-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK-32-EX-NEXT: [[TMP3:%.*]] = inttoptr i32 [[TMP0]] to ptr
+// CHECK-32-EX-NEXT: store ptr [[TMP3]], ptr [[TMP2]], align 4
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = inttoptr i32 [[TMP1]] to ptr
+// CHECK-32-EX-NEXT: store ptr [[TMP5]], ptr [[TMP4]], align 4
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP7]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l97_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0)
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -22334,7 +21124,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -22360,7 +21150,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -22407,7 +21197,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -22467,7 +21257,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -22499,7 +21289,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -22535,7 +21325,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -22582,7 +21372,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -22642,7 +21432,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -22674,7 +21464,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -22700,7 +21490,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -22747,7 +21537,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -22807,7 +21597,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -22852,23 +21642,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP170:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP170]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP159:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP159]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP170]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP159]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP170]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP159]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP170]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP159]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP170]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP171:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP159]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP160:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -22921,7 +21711,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -22981,7 +21771,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -23026,23 +21816,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP173:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP173]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP162:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP162]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP173]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP162]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP173]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP162]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP173]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP162]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP173]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP174:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP162]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP163:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -23095,7 +21885,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -23155,7 +21945,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -23200,23 +21990,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP176:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP176]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP165:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP165]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP176]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP165]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP176]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP165]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP176]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP165]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP176]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP177:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP165]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP166:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -23269,7 +22059,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -23329,7 +22119,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -23374,23 +22164,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP179:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP179]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP168:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP168]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP179]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP168]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP179]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP168]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP179]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP168]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP179]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP180:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP168]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP169:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -23441,69 +22231,19 @@ int a;
// CHECK-32-EX-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
-// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK-32-EX: cond.true:
-// CHECK-32-EX-NEXT: br label [[COND_END:%.*]]
-// CHECK-32-EX: cond.false:
-// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: br label [[COND_END]]
-// CHECK-32-EX: cond.end:
-// CHECK-32-EX-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP5]], 10
-// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = inttoptr i32 [[TMP6]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l125_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0)
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
-// CHECK-32-EX-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK-32-EX: cond.true5:
-// CHECK-32-EX-NEXT: br label [[COND_END7:%.*]]
-// CHECK-32-EX: cond.false6:
-// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: br label [[COND_END7]]
-// CHECK-32-EX: cond.end7:
-// CHECK-32-EX-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
-// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK-32-EX: omp.inner.for.end:
+// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK-32-EX-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK-32-EX-NEXT: [[TMP3:%.*]] = inttoptr i32 [[TMP0]] to ptr
+// CHECK-32-EX-NEXT: store ptr [[TMP3]], ptr [[TMP2]], align 4
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = inttoptr i32 [[TMP1]] to ptr
+// CHECK-32-EX-NEXT: store ptr [[TMP5]], ptr [[TMP4]], align 4
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP7]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3foov_l125_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2, i32 0)
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -23535,7 +22275,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -23561,7 +22301,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -23608,7 +22348,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -23668,7 +22408,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -23700,7 +22440,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -23736,7 +22476,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -23783,7 +22523,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -23843,7 +22583,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -23875,7 +22615,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -23901,7 +22641,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -23948,7 +22688,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -24008,7 +22748,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -24053,23 +22793,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP182:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP182]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP171:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP171]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP182]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP171]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP182]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP171]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP182]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP171]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP182]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP183:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP171]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP172:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -24122,7 +22862,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -24182,7 +22922,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -24227,23 +22967,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP185:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP185]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP174:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP174]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP185]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP174]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP185]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP174]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP185]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP174]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP185]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP186:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP174]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP175:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -24296,7 +23036,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -24356,7 +23096,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -24401,23 +23141,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP188:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP188]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP177:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP177]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP188]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP177]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP188]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP177]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP188]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP177]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP188]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP189:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP177]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP178:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -24470,7 +23210,7 @@ int a;
// CHECK-32-EX-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -24530,7 +23270,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -24575,23 +23315,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP191:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP191]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP180:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP180]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP191]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP180]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP191]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP180]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP191]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP180]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP191]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP192:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP180]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP181:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -24644,7 +23384,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-32-EX: omp.dispatch.cond:
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -24697,7 +23437,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -24739,7 +23479,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -24775,7 +23515,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -24817,7 +23557,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-32-EX: omp.dispatch.cond:
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -24870,7 +23610,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -24923,23 +23663,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP194:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP194]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP183:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP183]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP194]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP183]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP194]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP183]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP194]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP183]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP194]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP195:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP183]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP184:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -24998,23 +23738,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP197:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP197]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP186:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP186]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP197]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP186]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP197]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP186]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP197]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP186]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP197]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP198:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP186]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP187:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -25073,23 +23813,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP200:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP200]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP189:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP189]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP200]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP189]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP200]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP189]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP200]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP189]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP200]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP201:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP189]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP190:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -25148,23 +23888,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP203:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP203]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP192:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP192]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP203]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP192]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP203]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP192]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP203]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP192]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP203]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP204:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP192]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP193:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -25217,7 +23957,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-32-EX: omp.dispatch.cond:
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -25240,23 +23980,23 @@ int a;
// CHECK-32-EX: omp.dispatch.body:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP206:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP206]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP195:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP195]]
// CHECK-32-EX-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK-32-EX-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP206]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP195]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP206]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP195]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP206]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP195]]
// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP206]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP207:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP195]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP196:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -25270,7 +24010,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -25320,7 +24060,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -25336,27 +24076,27 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP209:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP209]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP198:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP198]]
// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP209]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP198]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP209]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP198]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP209]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP198]]
// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP209]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP210:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP198]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP199:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = icmp ne i32 [[TMP9]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP10]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -25406,7 +24146,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-32-EX: omp.dispatch.cond:
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -25429,23 +24169,23 @@ int a;
// CHECK-32-EX: omp.dispatch.body:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP212:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP212]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP201:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP201]]
// CHECK-32-EX-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK-32-EX-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP212]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP201]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP212]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP201]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP212]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP201]]
// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP212]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP213:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP201]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP202:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -25459,7 +24199,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -25520,23 +24260,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP215:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP215]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP204:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP204]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP215]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP204]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP215]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP204]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP215]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP204]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP215]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP216:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP204]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP205:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -25603,23 +24343,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP218:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP218]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP207:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP207]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP218]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP207]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP218]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP207]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP218]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP207]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP218]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP219:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP207]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP208:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -25686,23 +24426,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP210:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP210]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP210]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP210]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP210]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP222:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP210]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP211:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -25769,23 +24509,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP224:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP224]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP213:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP213]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP224]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP213]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP224]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP213]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP224]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP213]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP224]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP225:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP213]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP214:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -25852,24 +24592,24 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP216:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP216]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP216]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP216]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP216]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
-// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_fini_4(ptr @[[GLOB1]], i32 [[TMP1]]), !llvm.access.group [[ACC_GRP227]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP228:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP216]]
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_fini_4(ptr @[[GLOB1]], i32 [[TMP1]]), !llvm.access.group [[ACC_GRP216]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP217:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -25925,7 +24665,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -25941,27 +24681,27 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP230:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP230]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP219:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP219]]
// CHECK-32-EX-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK-32-EX-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP230]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP219]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP230]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP219]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP230]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP219]]
// CHECK-32-EX-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP230]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP231:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP219]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP220:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP10:%.*]] = icmp ne i32 [[TMP9]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP10]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -26011,7 +24751,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-32-EX: omp.dispatch.cond:
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -26034,23 +24774,23 @@ int a;
// CHECK-32-EX: omp.dispatch.body:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP233:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP233]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP222:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP222]]
// CHECK-32-EX-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK-32-EX-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP233]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP222]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP233]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP222]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP233]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP222]]
// CHECK-32-EX-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP233]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP234:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP222]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP223:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -26064,7 +24804,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK-32-EX-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -26125,23 +24865,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP236:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP236]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP225:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP225]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP236]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP225]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP236]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP225]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP236]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP225]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP236]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP237:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP225]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP226:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -26208,23 +24948,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP239:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP239]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP228:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP228]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP239]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP228]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP239]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP228]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP239]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP228]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP239]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP240:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP228]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP229:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -26291,23 +25031,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP242:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP242]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP231]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP242]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP242]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP231]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP242]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP242]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP243:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP232:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -26374,23 +25114,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP245:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP245]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP234:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP234]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP245]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP234]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP245]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP234]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP245]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP234]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP245]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP246:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP234]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP235:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -26446,7 +25186,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-32-EX: omp.dispatch.cond:
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -26499,7 +25239,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -26541,7 +25281,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 9
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
@@ -26577,7 +25317,7 @@ int a;
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -26619,7 +25359,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-32-EX-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK-32-EX-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK-32-EX: omp.dispatch.cond:
// CHECK-32-EX-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
@@ -26672,7 +25412,7 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
-// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK-32-EX-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -26725,23 +25465,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP248:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP248]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP237:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP237]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP248]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP237]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP248]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP237]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP248]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP237]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP248]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP249:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP237]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP238:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -26800,23 +25540,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP251:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP251]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP240:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP240]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP251]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP240]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP251]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP240]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP251]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP240]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP251]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP252:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP240]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP241:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -26875,23 +25615,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP254:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP254]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP243:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP243]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP254]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP243]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP254]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP243]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP254]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP243]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP254]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP255:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP243]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP244:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -26950,23 +25690,23 @@ int a;
// CHECK-32-EX-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_IV]], align 4
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK-32-EX: omp.inner.for.cond:
-// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP257:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP257]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP246:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP246]]
// CHECK-32-EX-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP4]], [[TMP5]]
// CHECK-32-EX-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK-32-EX: omp.inner.for.body:
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP257]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP246]]
// CHECK-32-EX-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP6]], 1
// CHECK-32-EX-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP257]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP246]]
// CHECK-32-EX-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK-32-EX: omp.body.continue:
// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK-32-EX: omp.inner.for.inc:
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP257]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP246]]
// CHECK-32-EX-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP257]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP258:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP246]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP247:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
diff --git a/clang/test/OpenMP/nvptx_distribute_parallel_generic_mode_codegen.cpp b/clang/test/OpenMP/nvptx_distribute_parallel_generic_mode_codegen.cpp
index 5a59d0ecbebb7..a6ac96aed0885 100644
--- a/clang/test/OpenMP/nvptx_distribute_parallel_generic_mode_codegen.cpp
+++ b/clang/test/OpenMP/nvptx_distribute_parallel_generic_mode_codegen.cpp
@@ -116,97 +116,37 @@ int main(int argc, char **argv) {
// CHECK4-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK4-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[B4]], ptr align 4 [[TMP0]], i64 40, i1 false)
// CHECK4-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK4-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK4-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
-// CHECK4-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK4-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK4-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK4-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK4-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
-// CHECK4-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]]
-// CHECK4-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK4: cond.true:
-// CHECK4-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
-// CHECK4-NEXT: br label [[COND_END:%.*]]
-// CHECK4: cond.false:
-// CHECK4-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK4-NEXT: br label [[COND_END]]
-// CHECK4: cond.end:
-// CHECK4-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
-// CHECK4-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK4-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK4-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4
-// CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK4: omp.inner.for.cond:
-// CHECK4-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK4-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
-// CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], 1
-// CHECK4-NEXT: [[CMP7:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
-// CHECK4-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK4: omp.inner.for.body:
-// CHECK4-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK4-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64
-// CHECK4-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK4-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64
-// CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK4-NEXT: [[TMP22:%.*]] = inttoptr i64 [[TMP18]] to ptr
-// CHECK4-NEXT: store ptr [[TMP22]], ptr [[TMP21]], align 8
-// CHECK4-NEXT: [[TMP23:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK4-NEXT: [[TMP24:%.*]] = inttoptr i64 [[TMP20]] to ptr
-// CHECK4-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 8
-// CHECK4-NEXT: [[TMP25:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK4-NEXT: store ptr [[ARGC_ADDR]], ptr [[TMP25]], align 8
-// CHECK4-NEXT: [[TMP26:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK4-NEXT: store ptr [[TMP2]], ptr [[TMP26]], align 8
-// CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
-// CHECK4-NEXT: store ptr [[B4]], ptr [[TMP27]], align 8
-// CHECK4-NEXT: [[TMP28:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 5
-// CHECK4-NEXT: store ptr [[C1]], ptr [[TMP28]], align 8
-// CHECK4-NEXT: [[TMP29:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 6
-// CHECK4-NEXT: store ptr [[TMP3]], ptr [[TMP29]], align 8
-// CHECK4-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK4-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP30]], 0
-// CHECK4-NEXT: [[TMP31:%.*]] = zext i1 [[TOBOOL]] to i32
-// CHECK4-NEXT: [[TMP32:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK4-NEXT: [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
-// CHECK4-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP33]], i32 [[TMP31]], i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l19_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 7, i32 0)
-// CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK4: omp.inner.for.inc:
-// CHECK4-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK4-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK4-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP34]], [[TMP35]]
-// CHECK4-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4
-// CHECK4-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK4-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK4-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP36]], [[TMP37]]
-// CHECK4-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK4-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK4-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK4-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP38]], [[TMP39]]
-// CHECK4-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK4-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK4-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
-// CHECK4-NEXT: [[CMP11:%.*]] = icmp sgt i32 [[TMP40]], [[TMP41]]
-// CHECK4-NEXT: br i1 [[CMP11]], label [[COND_TRUE12:%.*]], label [[COND_FALSE13:%.*]]
-// CHECK4: cond.true12:
-// CHECK4-NEXT: [[TMP42:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
-// CHECK4-NEXT: br label [[COND_END14:%.*]]
-// CHECK4: cond.false13:
-// CHECK4-NEXT: [[TMP43:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK4-NEXT: br label [[COND_END14]]
-// CHECK4: cond.end14:
-// CHECK4-NEXT: [[COND15:%.*]] = phi i32 [ [[TMP42]], [[COND_TRUE12]] ], [ [[TMP43]], [[COND_FALSE13]] ]
-// CHECK4-NEXT: store i32 [[COND15]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK4-NEXT: [[TMP44:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK4-NEXT: store i32 [[TMP44]], ptr [[DOTOMP_IV]], align 4
-// CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK4: omp.inner.for.end:
+// CHECK4-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
+// CHECK4-NEXT: [[TMP12:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK4-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP9]] to ptr
+// CHECK4-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
+// CHECK4-NEXT: [[TMP14:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK4-NEXT: [[TMP15:%.*]] = inttoptr i64 [[TMP11]] to ptr
+// CHECK4-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8
+// CHECK4-NEXT: [[TMP16:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK4-NEXT: store ptr [[ARGC_ADDR]], ptr [[TMP16]], align 8
+// CHECK4-NEXT: [[TMP17:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK4-NEXT: store ptr [[TMP2]], ptr [[TMP17]], align 8
+// CHECK4-NEXT: [[TMP18:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
+// CHECK4-NEXT: store ptr [[B4]], ptr [[TMP18]], align 8
+// CHECK4-NEXT: [[TMP19:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 5
+// CHECK4-NEXT: store ptr [[C1]], ptr [[TMP19]], align 8
+// CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 6
+// CHECK4-NEXT: store ptr [[TMP3]], ptr [[TMP20]], align 8
+// CHECK4-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK4-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP21]], 0
+// CHECK4-NEXT: [[TMP22:%.*]] = zext i1 [[TOBOOL]] to i32
+// CHECK4-NEXT: [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK4-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
+// CHECK4-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP24]], i32 [[TMP22]], i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l19_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 7, i32 0)
// CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK4: omp.loop.exit:
-// CHECK4-NEXT: [[TMP45:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK4-NEXT: [[TMP46:%.*]] = load i32, ptr [[TMP45]], align 4
-// CHECK4-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP46]])
-// CHECK4-NEXT: [[TMP47:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK4-NEXT: [[TMP48:%.*]] = icmp ne i32 [[TMP47]], 0
-// CHECK4-NEXT: br i1 [[TMP48]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
+// CHECK4-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK4-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
+// CHECK4-NEXT: br i1 [[TMP26]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
// CHECK4: .omp.lastprivate.then:
// CHECK4-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP1]], ptr align 4 [[C1]], i64 40, i1 false)
// CHECK4-NEXT: br label [[DOTOMP_LASTPRIVATE_DONE]]
@@ -281,7 +221,7 @@ int main(int argc, char **argv) {
// CHECK4-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[B4]], ptr align 4 [[TMP2]], i64 40, i1 false)
// CHECK4-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK4-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
-// CHECK4-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP12]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK4-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP12]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK4-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK4-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
// CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -329,7 +269,7 @@ int main(int argc, char **argv) {
// CHECK4: omp.loop.exit:
// CHECK4-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK4-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
-// CHECK4-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP23]])
+// CHECK4-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP23]])
// CHECK4-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK4-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK4-NEXT: br i1 [[TMP25]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
@@ -434,95 +374,35 @@ int main(int argc, char **argv) {
// CHECK5-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[B4]], ptr align 4 [[TMP0]], i32 40, i1 false)
// CHECK5-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK5-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
-// CHECK5-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
-// CHECK5-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]]
-// CHECK5-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK5: cond.true:
-// CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
-// CHECK5-NEXT: br label [[COND_END:%.*]]
-// CHECK5: cond.false:
-// CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK5-NEXT: br label [[COND_END]]
-// CHECK5: cond.end:
-// CHECK5-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
-// CHECK5-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK5-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4
-// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK5: omp.inner.for.cond:
-// CHECK5-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK5-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
-// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], 1
-// CHECK5-NEXT: [[CMP7:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
-// CHECK5-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK5: omp.inner.for.body:
-// CHECK5-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK5-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK5-NEXT: [[TMP19:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK5-NEXT: [[TMP20:%.*]] = inttoptr i32 [[TMP17]] to ptr
-// CHECK5-NEXT: store ptr [[TMP20]], ptr [[TMP19]], align 4
-// CHECK5-NEXT: [[TMP21:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK5-NEXT: [[TMP22:%.*]] = inttoptr i32 [[TMP18]] to ptr
-// CHECK5-NEXT: store ptr [[TMP22]], ptr [[TMP21]], align 4
-// CHECK5-NEXT: [[TMP23:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK5-NEXT: store ptr [[ARGC_ADDR]], ptr [[TMP23]], align 4
-// CHECK5-NEXT: [[TMP24:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
-// CHECK5-NEXT: store ptr [[TMP2]], ptr [[TMP24]], align 4
-// CHECK5-NEXT: [[TMP25:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 4
-// CHECK5-NEXT: store ptr [[B4]], ptr [[TMP25]], align 4
-// CHECK5-NEXT: [[TMP26:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 5
-// CHECK5-NEXT: store ptr [[C1]], ptr [[TMP26]], align 4
-// CHECK5-NEXT: [[TMP27:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 6
-// CHECK5-NEXT: store ptr [[TMP3]], ptr [[TMP27]], align 4
-// CHECK5-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP2]], align 4
-// CHECK5-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP28]], 0
-// CHECK5-NEXT: [[TMP29:%.*]] = zext i1 [[TOBOOL]] to i32
-// CHECK5-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK5-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
-// CHECK5-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP31]], i32 [[TMP29]], i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l19_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 7, i32 0)
-// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK5: omp.inner.for.inc:
-// CHECK5-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK5-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK5-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP32]], [[TMP33]]
-// CHECK5-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4
-// CHECK5-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK5-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK5-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP34]], [[TMP35]]
-// CHECK5-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK5-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK5-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK5-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP36]], [[TMP37]]
-// CHECK5-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK5-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK5-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
-// CHECK5-NEXT: [[CMP11:%.*]] = icmp sgt i32 [[TMP38]], [[TMP39]]
-// CHECK5-NEXT: br i1 [[CMP11]], label [[COND_TRUE12:%.*]], label [[COND_FALSE13:%.*]]
-// CHECK5: cond.true12:
-// CHECK5-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
-// CHECK5-NEXT: br label [[COND_END14:%.*]]
-// CHECK5: cond.false13:
-// CHECK5-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK5-NEXT: br label [[COND_END14]]
-// CHECK5: cond.end14:
-// CHECK5-NEXT: [[COND15:%.*]] = phi i32 [ [[TMP40]], [[COND_TRUE12]] ], [ [[TMP41]], [[COND_FALSE13]] ]
-// CHECK5-NEXT: store i32 [[COND15]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK5-NEXT: [[TMP42:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK5-NEXT: store i32 [[TMP42]], ptr [[DOTOMP_IV]], align 4
-// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK5: omp.inner.for.end:
+// CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK5-NEXT: [[TMP10:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK5-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP8]] to ptr
+// CHECK5-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4
+// CHECK5-NEXT: [[TMP12:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK5-NEXT: [[TMP13:%.*]] = inttoptr i32 [[TMP9]] to ptr
+// CHECK5-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 4
+// CHECK5-NEXT: [[TMP14:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK5-NEXT: store ptr [[ARGC_ADDR]], ptr [[TMP14]], align 4
+// CHECK5-NEXT: [[TMP15:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
+// CHECK5-NEXT: store ptr [[TMP2]], ptr [[TMP15]], align 4
+// CHECK5-NEXT: [[TMP16:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 4
+// CHECK5-NEXT: store ptr [[B4]], ptr [[TMP16]], align 4
+// CHECK5-NEXT: [[TMP17:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 5
+// CHECK5-NEXT: store ptr [[C1]], ptr [[TMP17]], align 4
+// CHECK5-NEXT: [[TMP18:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 6
+// CHECK5-NEXT: store ptr [[TMP3]], ptr [[TMP18]], align 4
+// CHECK5-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK5-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP19]], 0
+// CHECK5-NEXT: [[TMP20:%.*]] = zext i1 [[TOBOOL]] to i32
+// CHECK5-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK5-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
+// CHECK5-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP22]], i32 [[TMP20]], i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l19_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 7, i32 0)
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit:
-// CHECK5-NEXT: [[TMP43:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK5-NEXT: [[TMP44:%.*]] = load i32, ptr [[TMP43]], align 4
-// CHECK5-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP44]])
-// CHECK5-NEXT: [[TMP45:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK5-NEXT: [[TMP46:%.*]] = icmp ne i32 [[TMP45]], 0
-// CHECK5-NEXT: br i1 [[TMP46]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
+// CHECK5-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK5-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
+// CHECK5-NEXT: br i1 [[TMP24]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
// CHECK5: .omp.lastprivate.then:
// CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP1]], ptr align 4 [[C1]], i32 40, i1 false)
// CHECK5-NEXT: br label [[DOTOMP_LASTPRIVATE_DONE]]
@@ -595,7 +475,7 @@ int main(int argc, char **argv) {
// CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[B3]], ptr align 4 [[TMP2]], i32 40, i1 false)
// CHECK5-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
-// CHECK5-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP12]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK5-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP12]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -639,7 +519,7 @@ int main(int argc, char **argv) {
// CHECK5: omp.loop.exit:
// CHECK5-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK5-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
-// CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP23]])
+// CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP23]])
// CHECK5-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK5-NEXT: br i1 [[TMP25]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
diff --git a/clang/test/OpenMP/nvptx_target_teams_distribute_parallel_for_codegen.cpp b/clang/test/OpenMP/nvptx_target_teams_distribute_parallel_for_codegen.cpp
index 94eac3824463c..c06e878496aa9 100644
--- a/clang/test/OpenMP/nvptx_target_teams_distribute_parallel_for_codegen.cpp
+++ b/clang/test/OpenMP/nvptx_target_teams_distribute_parallel_for_codegen.cpp
@@ -457,89 +457,29 @@ int bar(int n){
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP6]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = zext i32 [[TMP5]] to i64
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP7]], [[TMP8]]
-// CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK1: cond.true:
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: br label [[COND_END:%.*]]
-// CHECK1: cond.false:
-// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END]]
-// CHECK1: cond.end:
-// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP9]], [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
-// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], 1
-// CHECK1-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP12]], [[ADD]]
-// CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP15:%.*]] = zext i32 [[TMP14]] to i64
-// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
-// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK1-NEXT: store i32 [[TMP18]], ptr [[N_CASTED]], align 4
-// CHECK1-NEXT: [[TMP19:%.*]] = load i64, ptr [[N_CASTED]], align 8
-// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK1-NEXT: [[TMP21:%.*]] = inttoptr i64 [[TMP15]] to ptr
-// CHECK1-NEXT: store ptr [[TMP21]], ptr [[TMP20]], align 8
-// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK1-NEXT: [[TMP23:%.*]] = inttoptr i64 [[TMP17]] to ptr
-// CHECK1-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 8
-// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK1-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP19]] to ptr
-// CHECK1-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 8
-// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 8
-// CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
-// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l34_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP29]], [[TMP30]]
-// CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
-// CHECK1-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
-// CHECK1-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP35]], [[TMP36]]
-// CHECK1-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]]
-// CHECK1: cond.true10:
-// CHECK1-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: br label [[COND_END12:%.*]]
-// CHECK1: cond.false11:
-// CHECK1-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END12]]
-// CHECK1: cond.end12:
-// CHECK1-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP37]], [[COND_TRUE10]] ], [ [[TMP38]], [[COND_FALSE11]] ]
-// CHECK1-NEXT: store i32 [[COND13]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP39]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP9]], ptr [[N_CASTED]], align 4
+// CHECK1-NEXT: [[TMP10:%.*]] = load i64, ptr [[N_CASTED]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK1-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP6]] to ptr
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK1-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP8]] to ptr
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK1-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr
+// CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP17]], align 8
+// CHECK1-NEXT: [[TMP18:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP18]], align 4
+// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP19]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l34_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
-// CHECK1-NEXT: [[TMP40:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP41:%.*]] = load i32, ptr [[TMP40]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP41]])
// CHECK1-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1: omp.precond.end:
// CHECK1-NEXT: ret void
@@ -596,7 +536,7 @@ int bar(int n){
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -686,73 +626,23 @@ int bar(int n){
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
-// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK1: cond.true:
-// CHECK1-NEXT: br label [[COND_END:%.*]]
-// CHECK1: cond.false:
-// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END]]
-// CHECK1: cond.end:
-// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
-// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP6]], 10
-// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
-// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK1-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP8]] to ptr
-// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8
-// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK1-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP10]] to ptr
-// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8
-// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 8
-// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l39_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3, i32 0)
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK1-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
-// CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP22]], 9
-// CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK1: cond.true5:
-// CHECK1-NEXT: br label [[COND_END7:%.*]]
-// CHECK1: cond.false6:
-// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END7]]
-// CHECK1: cond.end7:
-// CHECK1-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP23]], [[COND_FALSE6]] ]
-// CHECK1-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP24]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK1-NEXT: [[TMP6:%.*]] = inttoptr i64 [[TMP2]] to ptr
+// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP5]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK1-NEXT: [[TMP8:%.*]] = inttoptr i64 [[TMP4]] to ptr
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
+// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP11]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l39_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3, i32 0)
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
-// CHECK1-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP2]])
// CHECK1-NEXT: ret void
//
//
@@ -789,7 +679,7 @@ int bar(int n){
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -885,79 +775,29 @@ int bar(int n){
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99
-// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK1: cond.true:
-// CHECK1-NEXT: br label [[COND_END:%.*]]
-// CHECK1: cond.false:
-// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END]]
-// CHECK1: cond.end:
-// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
-// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[CMP2:%.*]] = icmp slt i32 [[TMP6]], 100
-// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
-// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[F_ADDR]], align 4
-// CHECK1-NEXT: store i32 [[TMP11]], ptr [[F_CASTED]], align 4
-// CHECK1-NEXT: [[TMP12:%.*]] = load i64, ptr [[F_CASTED]], align 8
-// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK1-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP8]] to ptr
-// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8
-// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK1-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr
-// CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8
-// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP17]], align 8
-// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK1-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP12]] to ptr
-// CHECK1-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8
-// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l44_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
-// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
-// CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP24]], [[TMP25]]
-// CHECK1-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP26]], 99
-// CHECK1-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
-// CHECK1: cond.true6:
-// CHECK1-NEXT: br label [[COND_END8:%.*]]
-// CHECK1: cond.false7:
-// CHECK1-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END8]]
-// CHECK1: cond.end8:
-// CHECK1-NEXT: [[COND9:%.*]] = phi i32 [ 99, [[COND_TRUE6]] ], [ [[TMP27]], [[COND_FALSE7]] ]
-// CHECK1-NEXT: store i32 [[COND9]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP28]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[F_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP5]], ptr [[F_CASTED]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[F_CASTED]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK1-NEXT: [[TMP8:%.*]] = inttoptr i64 [[TMP2]] to ptr
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK1-NEXT: [[TMP10:%.*]] = inttoptr i64 [[TMP4]] to ptr
+// CHECK1-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK1-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP6]] to ptr
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
+// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP15]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l44_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
-// CHECK1-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP2]])
// CHECK1-NEXT: ret void
//
//
@@ -999,7 +839,7 @@ int bar(int n){
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -1144,87 +984,27 @@ int bar(int n){
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK1-NEXT: [[CONV11:%.*]] = zext i32 [[NVPTX_NUM_THREADS]] to i64
-// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_init_8(ptr @[[GLOB2]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 [[CONV11]])
-// CHECK1-NEXT: [[TMP10:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK1-NEXT: [[CMP12:%.*]] = icmp sgt i64 [[TMP10]], [[TMP11]]
-// CHECK1-NEXT: br i1 [[CMP12]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK1: cond.true:
-// CHECK1-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK1-NEXT: br label [[COND_END:%.*]]
-// CHECK1: cond.false:
-// CHECK1-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: br label [[COND_END]]
-// CHECK1: cond.end:
-// CHECK1-NEXT: [[COND:%.*]] = phi i64 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
-// CHECK1-NEXT: store i64 [[COND]], ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK1-NEXT: store i64 [[TMP14]], ptr [[DOTOMP_IV]], align 8
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
-// CHECK1-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK1-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP16]], 1
-// CHECK1-NEXT: [[CMP13:%.*]] = icmp slt i64 [[TMP15]], [[ADD]]
-// CHECK1-NEXT: br i1 [[CMP13]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK1-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK1-NEXT: store i32 [[TMP19]], ptr [[N_CASTED]], align 4
-// CHECK1-NEXT: [[TMP20:%.*]] = load i64, ptr [[N_CASTED]], align 8
-// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK1-NEXT: [[TMP22:%.*]] = inttoptr i64 [[TMP17]] to ptr
-// CHECK1-NEXT: store ptr [[TMP22]], ptr [[TMP21]], align 8
-// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK1-NEXT: [[TMP24:%.*]] = inttoptr i64 [[TMP18]] to ptr
-// CHECK1-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 8
-// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK1-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP20]] to ptr
-// CHECK1-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8
-// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP27]], align 8
-// CHECK1-NEXT: [[TMP28:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[TMP28]], align 4
-// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP29]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l52_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP30:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
-// CHECK1-NEXT: [[TMP31:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
-// CHECK1-NEXT: [[ADD14:%.*]] = add nsw i64 [[TMP30]], [[TMP31]]
-// CHECK1-NEXT: store i64 [[ADD14]], ptr [[DOTOMP_IV]], align 8
-// CHECK1-NEXT: [[TMP32:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK1-NEXT: [[TMP33:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
-// CHECK1-NEXT: [[ADD15:%.*]] = add nsw i64 [[TMP32]], [[TMP33]]
-// CHECK1-NEXT: store i64 [[ADD15]], ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK1-NEXT: [[TMP34:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: [[TMP35:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
-// CHECK1-NEXT: [[ADD16:%.*]] = add nsw i64 [[TMP34]], [[TMP35]]
-// CHECK1-NEXT: store i64 [[ADD16]], ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: [[TMP36:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: [[TMP37:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK1-NEXT: [[CMP17:%.*]] = icmp sgt i64 [[TMP36]], [[TMP37]]
-// CHECK1-NEXT: br i1 [[CMP17]], label [[COND_TRUE18:%.*]], label [[COND_FALSE19:%.*]]
-// CHECK1: cond.true18:
-// CHECK1-NEXT: [[TMP38:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK1-NEXT: br label [[COND_END20:%.*]]
-// CHECK1: cond.false19:
-// CHECK1-NEXT: [[TMP39:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: br label [[COND_END20]]
-// CHECK1: cond.end20:
-// CHECK1-NEXT: [[COND21:%.*]] = phi i64 [ [[TMP38]], [[COND_TRUE18]] ], [ [[TMP39]], [[COND_FALSE19]] ]
-// CHECK1-NEXT: store i64 [[COND21]], ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: [[TMP40:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK1-NEXT: store i64 [[TMP40]], ptr [[DOTOMP_IV]], align 8
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP10]], ptr [[N_CASTED]], align 4
+// CHECK1-NEXT: [[TMP11:%.*]] = load i64, ptr [[N_CASTED]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK1-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP8]] to ptr
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK1-NEXT: [[TMP15:%.*]] = inttoptr i64 [[TMP9]] to ptr
+// CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK1-NEXT: [[TMP17:%.*]] = inttoptr i64 [[TMP11]] to ptr
+// CHECK1-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP18]], align 8
+// CHECK1-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4
+// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP20]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l52_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
-// CHECK1-NEXT: [[TMP41:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP42:%.*]] = load i32, ptr [[TMP41]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP42]])
// CHECK1-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1: omp.precond.end:
// CHECK1-NEXT: ret void
@@ -1296,7 +1076,7 @@ int bar(int n){
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
-// CHECK1-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB3]], i32 [[TMP11]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
+// CHECK1-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB3]], i32 [[TMP11]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK1-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK1-NEXT: store i64 [[TMP12]], ptr [[DOTOMP_IV]], align 8
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -1443,92 +1223,32 @@ int bar(int n){
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP6]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = zext i32 [[TMP5]] to i64
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP7]], [[TMP8]]
-// CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK1: cond.true:
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: br label [[COND_END:%.*]]
-// CHECK1: cond.false:
-// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END]]
-// CHECK1: cond.end:
-// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP9]], [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
-// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], 1
-// CHECK1-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP12]], [[ADD]]
-// CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP15:%.*]] = zext i32 [[TMP14]] to i64
-// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
-// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK1-NEXT: store i32 [[TMP18]], ptr [[N_CASTED]], align 4
-// CHECK1-NEXT: [[TMP19:%.*]] = load i64, ptr [[N_CASTED]], align 8
-// CHECK1-NEXT: [[TMP20:%.*]] = load ptr, ptr [[V_ADDR]], align 8
-// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK1-NEXT: [[TMP22:%.*]] = inttoptr i64 [[TMP15]] to ptr
-// CHECK1-NEXT: store ptr [[TMP22]], ptr [[TMP21]], align 8
-// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK1-NEXT: [[TMP24:%.*]] = inttoptr i64 [[TMP17]] to ptr
-// CHECK1-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 8
-// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK1-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP19]] to ptr
-// CHECK1-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8
-// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP27]], align 8
-// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
-// CHECK1-NEXT: store ptr [[TMP20]], ptr [[TMP28]], align 8
-// CHECK1-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
-// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP30]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l59_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5, i32 0)
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
-// CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
-// CHECK1-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP35]], [[TMP36]]
-// CHECK1-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP37]], [[TMP38]]
-// CHECK1-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]]
-// CHECK1: cond.true10:
-// CHECK1-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: br label [[COND_END12:%.*]]
-// CHECK1: cond.false11:
-// CHECK1-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END12]]
-// CHECK1: cond.end12:
-// CHECK1-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP39]], [[COND_TRUE10]] ], [ [[TMP40]], [[COND_FALSE11]] ]
-// CHECK1-NEXT: store i32 [[COND13]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP41]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP9]], ptr [[N_CASTED]], align 4
+// CHECK1-NEXT: [[TMP10:%.*]] = load i64, ptr [[N_CASTED]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[V_ADDR]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK1-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP6]] to ptr
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK1-NEXT: [[TMP15:%.*]] = inttoptr i64 [[TMP8]] to ptr
+// CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK1-NEXT: [[TMP17:%.*]] = inttoptr i64 [[TMP10]] to ptr
+// CHECK1-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP18]], align 8
+// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
+// CHECK1-NEXT: store ptr [[TMP11]], ptr [[TMP19]], align 8
+// CHECK1-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
+// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP21]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l59_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5, i32 0)
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
-// CHECK1-NEXT: [[TMP42:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP43:%.*]] = load i32, ptr [[TMP42]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP43]])
// CHECK1-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1: omp.precond.end:
// CHECK1-NEXT: ret void
@@ -1587,7 +1307,7 @@ int bar(int n){
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -2017,89 +1737,29 @@ int bar(int n){
// CHECK2-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP6]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK2-NEXT: [[TMP6:%.*]] = zext i32 [[TMP5]] to i64
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP7]], [[TMP8]]
-// CHECK2-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK2: cond.true:
-// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: br label [[COND_END:%.*]]
-// CHECK2: cond.false:
-// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END]]
-// CHECK2: cond.end:
-// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ [[TMP9]], [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
-// CHECK2-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], 1
-// CHECK2-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP12]], [[ADD]]
-// CHECK2-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP15:%.*]] = zext i32 [[TMP14]] to i64
-// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
-// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK2-NEXT: store i32 [[TMP18]], ptr [[N_CASTED]], align 4
-// CHECK2-NEXT: [[TMP19:%.*]] = load i64, ptr [[N_CASTED]], align 8
-// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK2-NEXT: [[TMP21:%.*]] = inttoptr i64 [[TMP15]] to ptr
-// CHECK2-NEXT: store ptr [[TMP21]], ptr [[TMP20]], align 8
-// CHECK2-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK2-NEXT: [[TMP23:%.*]] = inttoptr i64 [[TMP17]] to ptr
-// CHECK2-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 8
-// CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK2-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP19]] to ptr
-// CHECK2-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 8
-// CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 8
-// CHECK2-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
-// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l34_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP29]], [[TMP30]]
-// CHECK2-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
-// CHECK2-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
-// CHECK2-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP35]], [[TMP36]]
-// CHECK2-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]]
-// CHECK2: cond.true10:
-// CHECK2-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: br label [[COND_END12:%.*]]
-// CHECK2: cond.false11:
-// CHECK2-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END12]]
-// CHECK2: cond.end12:
-// CHECK2-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP37]], [[COND_TRUE10]] ], [ [[TMP38]], [[COND_FALSE11]] ]
-// CHECK2-NEXT: store i32 [[COND13]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP39]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK2: omp.inner.for.end:
+// CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK2-NEXT: store i32 [[TMP9]], ptr [[N_CASTED]], align 4
+// CHECK2-NEXT: [[TMP10:%.*]] = load i64, ptr [[N_CASTED]], align 8
+// CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK2-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP6]] to ptr
+// CHECK2-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8
+// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK2-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP8]] to ptr
+// CHECK2-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8
+// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK2-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr
+// CHECK2-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8
+// CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP17]], align 8
+// CHECK2-NEXT: [[TMP18:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK2-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP18]], align 4
+// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP19]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l34_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
-// CHECK2-NEXT: [[TMP40:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP41:%.*]] = load i32, ptr [[TMP40]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP41]])
// CHECK2-NEXT: br label [[OMP_PRECOND_END]]
// CHECK2: omp.precond.end:
// CHECK2-NEXT: ret void
@@ -2156,7 +1816,7 @@ int bar(int n){
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -2246,73 +1906,23 @@ int bar(int n){
// CHECK2-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK2-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
-// CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK2: cond.true:
-// CHECK2-NEXT: br label [[COND_END:%.*]]
-// CHECK2: cond.false:
-// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END]]
-// CHECK2: cond.end:
-// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
-// CHECK2-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP6]], 10
-// CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
-// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
-// CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK2-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP8]] to ptr
-// CHECK2-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8
-// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK2-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP10]] to ptr
-// CHECK2-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8
-// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 8
-// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l39_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3, i32 0)
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK2-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK2-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
-// CHECK2-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP22]], 9
-// CHECK2-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK2: cond.true5:
-// CHECK2-NEXT: br label [[COND_END7:%.*]]
-// CHECK2: cond.false6:
-// CHECK2-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END7]]
-// CHECK2: cond.end7:
-// CHECK2-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP23]], [[COND_FALSE6]] ]
-// CHECK2-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP24]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK2: omp.inner.for.end:
+// CHECK2-NEXT: [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
+// CHECK2-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK2-NEXT: [[TMP6:%.*]] = inttoptr i64 [[TMP2]] to ptr
+// CHECK2-NEXT: store ptr [[TMP6]], ptr [[TMP5]], align 8
+// CHECK2-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK2-NEXT: [[TMP8:%.*]] = inttoptr i64 [[TMP4]] to ptr
+// CHECK2-NEXT: store ptr [[TMP8]], ptr [[TMP7]], align 8
+// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP9]], align 8
+// CHECK2-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
+// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP11]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l39_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3, i32 0)
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
-// CHECK2-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP2]])
// CHECK2-NEXT: ret void
//
//
@@ -2349,7 +1959,7 @@ int bar(int n){
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -2445,79 +2055,29 @@ int bar(int n){
// CHECK2-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK2-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99
-// CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK2: cond.true:
-// CHECK2-NEXT: br label [[COND_END:%.*]]
-// CHECK2: cond.false:
-// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END]]
-// CHECK2: cond.end:
-// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
-// CHECK2-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[CMP2:%.*]] = icmp slt i32 [[TMP6]], 100
-// CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
-// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
-// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[F_ADDR]], align 4
-// CHECK2-NEXT: store i32 [[TMP11]], ptr [[F_CASTED]], align 4
-// CHECK2-NEXT: [[TMP12:%.*]] = load i64, ptr [[F_CASTED]], align 8
-// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK2-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP8]] to ptr
-// CHECK2-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8
-// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK2-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr
-// CHECK2-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8
-// CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP17]], align 8
-// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK2-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP12]] to ptr
-// CHECK2-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8
-// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l44_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
-// CHECK2-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
-// CHECK2-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP24]], [[TMP25]]
-// CHECK2-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP26]], 99
-// CHECK2-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
-// CHECK2: cond.true6:
-// CHECK2-NEXT: br label [[COND_END8:%.*]]
-// CHECK2: cond.false7:
-// CHECK2-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END8]]
-// CHECK2: cond.end8:
-// CHECK2-NEXT: [[COND9:%.*]] = phi i32 [ 99, [[COND_TRUE6]] ], [ [[TMP27]], [[COND_FALSE7]] ]
-// CHECK2-NEXT: store i32 [[COND9]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP28]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK2: omp.inner.for.end:
+// CHECK2-NEXT: [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
+// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[F_ADDR]], align 4
+// CHECK2-NEXT: store i32 [[TMP5]], ptr [[F_CASTED]], align 4
+// CHECK2-NEXT: [[TMP6:%.*]] = load i64, ptr [[F_CASTED]], align 8
+// CHECK2-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK2-NEXT: [[TMP8:%.*]] = inttoptr i64 [[TMP2]] to ptr
+// CHECK2-NEXT: store ptr [[TMP8]], ptr [[TMP7]], align 8
+// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK2-NEXT: [[TMP10:%.*]] = inttoptr i64 [[TMP4]] to ptr
+// CHECK2-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 8
+// CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP11]], align 8
+// CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK2-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP6]] to ptr
+// CHECK2-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
+// CHECK2-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
+// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP15]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l44_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
-// CHECK2-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP2]])
// CHECK2-NEXT: ret void
//
//
@@ -2559,7 +2119,7 @@ int bar(int n){
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -2701,89 +2261,29 @@ int bar(int n){
// CHECK2-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK2-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK2-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4
-// CHECK2-NEXT: [[CMP10:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]]
-// CHECK2-NEXT: br i1 [[CMP10]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK2: cond.true:
-// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4
-// CHECK2-NEXT: br label [[COND_END:%.*]]
-// CHECK2: cond.false:
-// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END]]
-// CHECK2: cond.end:
-// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
-// CHECK2-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4
-// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], 1
-// CHECK2-NEXT: [[CMP11:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
-// CHECK2-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64
-// CHECK2-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64
-// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK2-NEXT: store i32 [[TMP21]], ptr [[N_CASTED]], align 4
-// CHECK2-NEXT: [[TMP22:%.*]] = load i64, ptr [[N_CASTED]], align 8
-// CHECK2-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK2-NEXT: [[TMP24:%.*]] = inttoptr i64 [[TMP18]] to ptr
-// CHECK2-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 8
-// CHECK2-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK2-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP20]] to ptr
-// CHECK2-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8
-// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK2-NEXT: [[TMP28:%.*]] = inttoptr i64 [[TMP22]] to ptr
-// CHECK2-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 8
-// CHECK2-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP29]], align 8
-// CHECK2-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
-// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP31]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l52_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP32]], [[TMP33]]
-// CHECK2-NEXT: store i32 [[ADD12]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP34]], [[TMP35]]
-// CHECK2-NEXT: store i32 [[ADD13]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP36]], [[TMP37]]
-// CHECK2-NEXT: store i32 [[ADD14]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4
-// CHECK2-NEXT: [[CMP15:%.*]] = icmp sgt i32 [[TMP38]], [[TMP39]]
-// CHECK2-NEXT: br i1 [[CMP15]], label [[COND_TRUE16:%.*]], label [[COND_FALSE17:%.*]]
-// CHECK2: cond.true16:
-// CHECK2-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4
-// CHECK2-NEXT: br label [[COND_END18:%.*]]
-// CHECK2: cond.false17:
-// CHECK2-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END18]]
-// CHECK2: cond.end18:
-// CHECK2-NEXT: [[COND19:%.*]] = phi i32 [ [[TMP40]], [[COND_TRUE16]] ], [ [[TMP41]], [[COND_FALSE17]] ]
-// CHECK2-NEXT: store i32 [[COND19]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP42:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP42]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK2: omp.inner.for.end:
+// CHECK2-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
+// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK2-NEXT: store i32 [[TMP12]], ptr [[N_CASTED]], align 4
+// CHECK2-NEXT: [[TMP13:%.*]] = load i64, ptr [[N_CASTED]], align 8
+// CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK2-NEXT: [[TMP15:%.*]] = inttoptr i64 [[TMP9]] to ptr
+// CHECK2-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8
+// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK2-NEXT: [[TMP17:%.*]] = inttoptr i64 [[TMP11]] to ptr
+// CHECK2-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8
+// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK2-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP13]] to ptr
+// CHECK2-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8
+// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP20]], align 8
+// CHECK2-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
+// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP22]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l52_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
-// CHECK2-NEXT: [[TMP43:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP44:%.*]] = load i32, ptr [[TMP43]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP44]])
// CHECK2-NEXT: br label [[OMP_PRECOND_END]]
// CHECK2: omp.precond.end:
// CHECK2-NEXT: ret void
@@ -2855,7 +2355,7 @@ int bar(int n){
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
-// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP11]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP11]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -2998,92 +2498,32 @@ int bar(int n){
// CHECK2-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP6]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK2-NEXT: [[TMP6:%.*]] = zext i32 [[TMP5]] to i64
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP7]], [[TMP8]]
-// CHECK2-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK2: cond.true:
-// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: br label [[COND_END:%.*]]
-// CHECK2: cond.false:
-// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END]]
-// CHECK2: cond.end:
-// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ [[TMP9]], [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
-// CHECK2-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], 1
-// CHECK2-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP12]], [[ADD]]
-// CHECK2-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP15:%.*]] = zext i32 [[TMP14]] to i64
-// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
-// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK2-NEXT: store i32 [[TMP18]], ptr [[N_CASTED]], align 4
-// CHECK2-NEXT: [[TMP19:%.*]] = load i64, ptr [[N_CASTED]], align 8
-// CHECK2-NEXT: [[TMP20:%.*]] = load ptr, ptr [[V_ADDR]], align 8
-// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK2-NEXT: [[TMP22:%.*]] = inttoptr i64 [[TMP15]] to ptr
-// CHECK2-NEXT: store ptr [[TMP22]], ptr [[TMP21]], align 8
-// CHECK2-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK2-NEXT: [[TMP24:%.*]] = inttoptr i64 [[TMP17]] to ptr
-// CHECK2-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 8
-// CHECK2-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK2-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP19]] to ptr
-// CHECK2-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8
-// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP27]], align 8
-// CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
-// CHECK2-NEXT: store ptr [[TMP20]], ptr [[TMP28]], align 8
-// CHECK2-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
-// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP30]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l59_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5, i32 0)
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
-// CHECK2-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
-// CHECK2-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP35]], [[TMP36]]
-// CHECK2-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP37]], [[TMP38]]
-// CHECK2-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]]
-// CHECK2: cond.true10:
-// CHECK2-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: br label [[COND_END12:%.*]]
-// CHECK2: cond.false11:
-// CHECK2-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END12]]
-// CHECK2: cond.end12:
-// CHECK2-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP39]], [[COND_TRUE10]] ], [ [[TMP40]], [[COND_FALSE11]] ]
-// CHECK2-NEXT: store i32 [[COND13]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP41]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK2: omp.inner.for.end:
+// CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK2-NEXT: store i32 [[TMP9]], ptr [[N_CASTED]], align 4
+// CHECK2-NEXT: [[TMP10:%.*]] = load i64, ptr [[N_CASTED]], align 8
+// CHECK2-NEXT: [[TMP11:%.*]] = load ptr, ptr [[V_ADDR]], align 8
+// CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK2-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP6]] to ptr
+// CHECK2-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
+// CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK2-NEXT: [[TMP15:%.*]] = inttoptr i64 [[TMP8]] to ptr
+// CHECK2-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8
+// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK2-NEXT: [[TMP17:%.*]] = inttoptr i64 [[TMP10]] to ptr
+// CHECK2-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8
+// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP18]], align 8
+// CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
+// CHECK2-NEXT: store ptr [[TMP11]], ptr [[TMP19]], align 8
+// CHECK2-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
+// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP21]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l59_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5, i32 0)
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
-// CHECK2-NEXT: [[TMP42:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP43:%.*]] = load i32, ptr [[TMP42]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP43]])
// CHECK2-NEXT: br label [[OMP_PRECOND_END]]
// CHECK2: omp.precond.end:
// CHECK2-NEXT: ret void
@@ -3142,7 +2582,7 @@ int bar(int n){
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -3565,87 +3005,27 @@ int bar(int n){
// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
-// CHECK3-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP6]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP7]], [[TMP8]]
-// CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK3: cond.true:
-// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: br label [[COND_END:%.*]]
-// CHECK3: cond.false:
-// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: br label [[COND_END]]
-// CHECK3: cond.end:
-// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP9]], [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
-// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK3: omp.inner.for.cond:
-// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], 1
-// CHECK3-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP12]], [[ADD]]
-// CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK3: omp.inner.for.body:
-// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK3-NEXT: store i32 [[TMP16]], ptr [[N_CASTED]], align 4
-// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[N_CASTED]], align 4
-// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK3-NEXT: [[TMP19:%.*]] = inttoptr i32 [[TMP14]] to ptr
-// CHECK3-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 4
-// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK3-NEXT: [[TMP21:%.*]] = inttoptr i32 [[TMP15]] to ptr
-// CHECK3-NEXT: store ptr [[TMP21]], ptr [[TMP20]], align 4
-// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK3-NEXT: [[TMP23:%.*]] = inttoptr i32 [[TMP17]] to ptr
-// CHECK3-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 4
-// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
-// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP24]], align 4
-// CHECK3-NEXT: [[TMP25:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
-// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP26]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l34_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4, i32 0)
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK3: omp.inner.for.inc:
-// CHECK3-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP27]], [[TMP28]]
-// CHECK3-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP29]], [[TMP30]]
-// CHECK3-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
-// CHECK3-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP33]], [[TMP34]]
-// CHECK3-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]]
-// CHECK3: cond.true10:
-// CHECK3-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: br label [[COND_END12:%.*]]
-// CHECK3: cond.false11:
-// CHECK3-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: br label [[COND_END12]]
-// CHECK3: cond.end12:
-// CHECK3-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP35]], [[COND_TRUE10]] ], [ [[TMP36]], [[COND_FALSE11]] ]
-// CHECK3-NEXT: store i32 [[COND13]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: store i32 [[TMP37]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP7]], ptr [[N_CASTED]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[N_CASTED]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP10:%.*]] = inttoptr i32 [[TMP5]] to ptr
+// CHECK3-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP6]] to ptr
+// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP14:%.*]] = inttoptr i32 [[TMP8]] to ptr
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 4
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4
+// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP17]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l34_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4, i32 0)
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
-// CHECK3-NEXT: [[TMP38:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP39:%.*]] = load i32, ptr [[TMP38]], align 4
-// CHECK3-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP39]])
// CHECK3-NEXT: br label [[OMP_PRECOND_END]]
// CHECK3: omp.precond.end:
// CHECK3-NEXT: ret void
@@ -3700,7 +3080,7 @@ int bar(int n){
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -3788,71 +3168,21 @@ int bar(int n){
// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
-// CHECK3-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
-// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK3: cond.true:
-// CHECK3-NEXT: br label [[COND_END:%.*]]
-// CHECK3: cond.false:
-// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: br label [[COND_END]]
-// CHECK3: cond.end:
-// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
-// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK3: omp.inner.for.cond:
-// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP6]], 10
-// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK3: omp.inner.for.body:
-// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK3-NEXT: [[TMP10:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK3-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 4
-// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK3-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP8]] to ptr
-// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4
-// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP13]], align 4
-// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l39_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3, i32 0)
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK3: omp.inner.for.inc:
-// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK3-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
-// CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK3: cond.true5:
-// CHECK3-NEXT: br label [[COND_END7:%.*]]
-// CHECK3: cond.false6:
-// CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: br label [[COND_END7]]
-// CHECK3: cond.end7:
-// CHECK3-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK3-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = inttoptr i32 [[TMP1]] to ptr
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP3]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP6:%.*]] = inttoptr i32 [[TMP2]] to ptr
+// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
+// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP9]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l39_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3, i32 0)
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
-// CHECK3-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP2]])
// CHECK3-NEXT: ret void
//
//
@@ -3887,7 +3217,7 @@ int bar(int n){
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -3981,77 +3311,27 @@ int bar(int n){
// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
-// CHECK3-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99
-// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK3: cond.true:
-// CHECK3-NEXT: br label [[COND_END:%.*]]
-// CHECK3: cond.false:
-// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: br label [[COND_END]]
-// CHECK3: cond.end:
-// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
-// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK3: omp.inner.for.cond:
-// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[CMP2:%.*]] = icmp slt i32 [[TMP6]], 100
-// CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK3: omp.inner.for.body:
-// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[F_ADDR]], align 4
-// CHECK3-NEXT: store i32 [[TMP9]], ptr [[F_CASTED]], align 4
-// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[F_CASTED]], align 4
-// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK3-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4
-// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK3-NEXT: [[TMP14:%.*]] = inttoptr i32 [[TMP8]] to ptr
-// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 4
-// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 4
-// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
-// CHECK3-NEXT: [[TMP17:%.*]] = inttoptr i32 [[TMP10]] to ptr
-// CHECK3-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 4
-// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l44_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4, i32 0)
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK3: omp.inner.for.inc:
-// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
-// CHECK3-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
-// CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP24]], 99
-// CHECK3-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
-// CHECK3: cond.true6:
-// CHECK3-NEXT: br label [[COND_END8:%.*]]
-// CHECK3: cond.false7:
-// CHECK3-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: br label [[COND_END8]]
-// CHECK3: cond.end8:
-// CHECK3-NEXT: [[COND9:%.*]] = phi i32 [ 99, [[COND_TRUE6]] ], [ [[TMP25]], [[COND_FALSE7]] ]
-// CHECK3-NEXT: store i32 [[COND9]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: store i32 [[TMP26]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[F_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP3]], ptr [[F_CASTED]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[F_CASTED]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = inttoptr i32 [[TMP1]] to ptr
+// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP8:%.*]] = inttoptr i32 [[TMP2]] to ptr
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
+// CHECK3-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP4]] to ptr
+// CHECK3-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
+// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP13]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l44_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4, i32 0)
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
-// CHECK3-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP2]])
// CHECK3-NEXT: ret void
//
//
@@ -4091,7 +3371,7 @@ int bar(int n){
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -4233,89 +3513,29 @@ int bar(int n){
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK3-NEXT: [[CONV11:%.*]] = zext i32 [[NVPTX_NUM_THREADS]] to i64
-// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
-// CHECK3-NEXT: call void @__kmpc_distribute_static_init_8(ptr @[[GLOB2]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 [[CONV11]])
+// CHECK3-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
+// CHECK3-NEXT: [[TMP9:%.*]] = trunc i64 [[TMP8]] to i32
// CHECK3-NEXT: [[TMP10:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK3-NEXT: [[CMP12:%.*]] = icmp sgt i64 [[TMP10]], [[TMP11]]
-// CHECK3-NEXT: br i1 [[CMP12]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK3: cond.true:
-// CHECK3-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK3-NEXT: br label [[COND_END:%.*]]
-// CHECK3: cond.false:
-// CHECK3-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: br label [[COND_END]]
-// CHECK3: cond.end:
-// CHECK3-NEXT: [[COND:%.*]] = phi i64 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
-// CHECK3-NEXT: store i64 [[COND]], ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK3-NEXT: store i64 [[TMP14]], ptr [[DOTOMP_IV]], align 8
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK3: omp.inner.for.cond:
-// CHECK3-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
-// CHECK3-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK3-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP16]], 1
-// CHECK3-NEXT: [[CMP13:%.*]] = icmp slt i64 [[TMP15]], [[ADD]]
-// CHECK3-NEXT: br i1 [[CMP13]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK3: omp.inner.for.body:
-// CHECK3-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK3-NEXT: [[TMP18:%.*]] = trunc i64 [[TMP17]] to i32
-// CHECK3-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: [[TMP20:%.*]] = trunc i64 [[TMP19]] to i32
-// CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK3-NEXT: store i32 [[TMP21]], ptr [[N_CASTED]], align 4
-// CHECK3-NEXT: [[TMP22:%.*]] = load i32, ptr [[N_CASTED]], align 4
-// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK3-NEXT: [[TMP24:%.*]] = inttoptr i32 [[TMP18]] to ptr
-// CHECK3-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 4
-// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK3-NEXT: [[TMP26:%.*]] = inttoptr i32 [[TMP20]] to ptr
-// CHECK3-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 4
-// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK3-NEXT: [[TMP28:%.*]] = inttoptr i32 [[TMP22]] to ptr
-// CHECK3-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 4
-// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
-// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP29]], align 4
-// CHECK3-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
-// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP31]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l52_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4, i32 0)
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK3: omp.inner.for.inc:
-// CHECK3-NEXT: [[TMP32:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
-// CHECK3-NEXT: [[TMP33:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
-// CHECK3-NEXT: [[ADD14:%.*]] = add nsw i64 [[TMP32]], [[TMP33]]
-// CHECK3-NEXT: store i64 [[ADD14]], ptr [[DOTOMP_IV]], align 8
-// CHECK3-NEXT: [[TMP34:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK3-NEXT: [[TMP35:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
-// CHECK3-NEXT: [[ADD15:%.*]] = add nsw i64 [[TMP34]], [[TMP35]]
-// CHECK3-NEXT: store i64 [[ADD15]], ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK3-NEXT: [[TMP36:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: [[TMP37:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
-// CHECK3-NEXT: [[ADD16:%.*]] = add nsw i64 [[TMP36]], [[TMP37]]
-// CHECK3-NEXT: store i64 [[ADD16]], ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: [[TMP38:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: [[TMP39:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK3-NEXT: [[CMP17:%.*]] = icmp sgt i64 [[TMP38]], [[TMP39]]
-// CHECK3-NEXT: br i1 [[CMP17]], label [[COND_TRUE18:%.*]], label [[COND_FALSE19:%.*]]
-// CHECK3: cond.true18:
-// CHECK3-NEXT: [[TMP40:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK3-NEXT: br label [[COND_END20:%.*]]
-// CHECK3: cond.false19:
-// CHECK3-NEXT: [[TMP41:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: br label [[COND_END20]]
-// CHECK3: cond.end20:
-// CHECK3-NEXT: [[COND21:%.*]] = phi i64 [ [[TMP40]], [[COND_TRUE18]] ], [ [[TMP41]], [[COND_FALSE19]] ]
-// CHECK3-NEXT: store i64 [[COND21]], ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: [[TMP42:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK3-NEXT: store i64 [[TMP42]], ptr [[DOTOMP_IV]], align 8
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: [[TMP11:%.*]] = trunc i64 [[TMP10]] to i32
+// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP12]], ptr [[N_CASTED]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[N_CASTED]], align 4
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP15:%.*]] = inttoptr i32 [[TMP9]] to ptr
+// CHECK3-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP17:%.*]] = inttoptr i32 [[TMP11]] to ptr
+// CHECK3-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP19:%.*]] = inttoptr i32 [[TMP13]] to ptr
+// CHECK3-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 4
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP20]], align 4
+// CHECK3-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
+// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP22]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l52_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4, i32 0)
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
-// CHECK3-NEXT: [[TMP43:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP44:%.*]] = load i32, ptr [[TMP43]], align 4
-// CHECK3-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP44]])
// CHECK3-NEXT: br label [[OMP_PRECOND_END]]
// CHECK3: omp.precond.end:
// CHECK3-NEXT: ret void
@@ -4389,7 +3609,7 @@ int bar(int n){
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
-// CHECK3-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB3]], i32 [[TMP11]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
+// CHECK3-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB3]], i32 [[TMP11]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK3-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK3-NEXT: store i64 [[TMP12]], ptr [[DOTOMP_IV]], align 8
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -4535,90 +3755,30 @@ int bar(int n){
// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
-// CHECK3-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP6]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP7]], [[TMP8]]
-// CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK3: cond.true:
-// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: br label [[COND_END:%.*]]
-// CHECK3: cond.false:
-// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: br label [[COND_END]]
-// CHECK3: cond.end:
-// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP9]], [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
-// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK3: omp.inner.for.cond:
-// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], 1
-// CHECK3-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP12]], [[ADD]]
-// CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK3: omp.inner.for.body:
-// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK3-NEXT: store i32 [[TMP16]], ptr [[N_CASTED]], align 4
-// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[N_CASTED]], align 4
-// CHECK3-NEXT: [[TMP18:%.*]] = load ptr, ptr [[V_ADDR]], align 4
-// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK3-NEXT: [[TMP20:%.*]] = inttoptr i32 [[TMP14]] to ptr
-// CHECK3-NEXT: store ptr [[TMP20]], ptr [[TMP19]], align 4
-// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK3-NEXT: [[TMP22:%.*]] = inttoptr i32 [[TMP15]] to ptr
-// CHECK3-NEXT: store ptr [[TMP22]], ptr [[TMP21]], align 4
-// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK3-NEXT: [[TMP24:%.*]] = inttoptr i32 [[TMP17]] to ptr
-// CHECK3-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 4
-// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
-// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP25]], align 4
-// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 4
-// CHECK3-NEXT: store ptr [[TMP18]], ptr [[TMP26]], align 4
-// CHECK3-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
-// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l59_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 5, i32 0)
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK3: omp.inner.for.inc:
-// CHECK3-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP29]], [[TMP30]]
-// CHECK3-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
-// CHECK3-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
-// CHECK3-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP35]], [[TMP36]]
-// CHECK3-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]]
-// CHECK3: cond.true10:
-// CHECK3-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: br label [[COND_END12:%.*]]
-// CHECK3: cond.false11:
-// CHECK3-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: br label [[COND_END12]]
-// CHECK3: cond.end12:
-// CHECK3-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP37]], [[COND_TRUE10]] ], [ [[TMP38]], [[COND_FALSE11]] ]
-// CHECK3-NEXT: store i32 [[COND13]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: store i32 [[TMP39]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP7]], ptr [[N_CASTED]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[N_CASTED]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[V_ADDR]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP5]] to ptr
+// CHECK3-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP13:%.*]] = inttoptr i32 [[TMP6]] to ptr
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP15:%.*]] = inttoptr i32 [[TMP8]] to ptr
+// CHECK3-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP18]], align 4
+// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP19]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l59_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 5, i32 0)
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
-// CHECK3-NEXT: [[TMP40:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP41:%.*]] = load i32, ptr [[TMP40]], align 4
-// CHECK3-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP41]])
// CHECK3-NEXT: br label [[OMP_PRECOND_END]]
// CHECK3: omp.precond.end:
// CHECK3-NEXT: ret void
@@ -4675,7 +3835,7 @@ int bar(int n){
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
diff --git a/clang/test/OpenMP/nvptx_target_teams_distribute_parallel_for_simd_codegen.cpp b/clang/test/OpenMP/nvptx_target_teams_distribute_parallel_for_simd_codegen.cpp
index f2cc95451e2f7..227b111956fef 100644
--- a/clang/test/OpenMP/nvptx_target_teams_distribute_parallel_for_simd_codegen.cpp
+++ b/clang/test/OpenMP/nvptx_target_teams_distribute_parallel_for_simd_codegen.cpp
@@ -469,99 +469,39 @@ int bar(int n){
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP6]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = zext i32 [[TMP5]] to i64
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP7]], [[TMP8]]
-// CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK1: cond.true:
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: br label [[COND_END:%.*]]
-// CHECK1: cond.false:
-// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END]]
-// CHECK1: cond.end:
-// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP9]], [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
-// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17:![0-9]+]]
-// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], 1
-// CHECK1-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP12]], [[ADD]]
-// CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP15:%.*]] = zext i32 [[TMP14]] to i64
-// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
-// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[N_ADDR]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: store i32 [[TMP18]], ptr [[N_CASTED]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP19:%.*]] = load i64, ptr [[N_CASTED]], align 8, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK1-NEXT: [[TMP21:%.*]] = inttoptr i64 [[TMP15]] to ptr
-// CHECK1-NEXT: store ptr [[TMP21]], ptr [[TMP20]], align 8, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK1-NEXT: [[TMP23:%.*]] = inttoptr i64 [[TMP17]] to ptr
-// CHECK1-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 8, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK1-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP19]] to ptr
-// CHECK1-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 8, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 8, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l32_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0), !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP29]], [[TMP30]]
-// CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
-// CHECK1-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
-// CHECK1-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP35]], [[TMP36]]
-// CHECK1-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]]
-// CHECK1: cond.true10:
-// CHECK1-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: br label [[COND_END12:%.*]]
-// CHECK1: cond.false11:
-// CHECK1-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: br label [[COND_END12]]
-// CHECK1: cond.end12:
-// CHECK1-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP37]], [[COND_TRUE10]] ], [ [[TMP38]], [[COND_FALSE11]] ]
-// CHECK1-NEXT: store i32 [[COND13]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: store i32 [[TMP39]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]]
-// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP9]], ptr [[N_CASTED]], align 4
+// CHECK1-NEXT: [[TMP10:%.*]] = load i64, ptr [[N_CASTED]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK1-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP6]] to ptr
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK1-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP8]] to ptr
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK1-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr
+// CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP17]], align 8
+// CHECK1-NEXT: [[TMP18:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP18]], align 4
+// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP19]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l32_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
-// CHECK1-NEXT: [[TMP40:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP41:%.*]] = load i32, ptr [[TMP40]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP41]])
-// CHECK1-NEXT: [[TMP42:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK1-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
-// CHECK1-NEXT: br i1 [[TMP43]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP21:%.*]] = icmp ne i32 [[TMP20]], 0
+// CHECK1-NEXT: br i1 [[TMP21]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK1: .omp.final.then:
-// CHECK1-NEXT: [[TMP44:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
-// CHECK1-NEXT: [[SUB14:%.*]] = sub nsw i32 [[TMP44]], 0
-// CHECK1-NEXT: [[DIV15:%.*]] = sdiv i32 [[SUB14]], 1
-// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV15]], 1
-// CHECK1-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK1-NEXT: store i32 [[ADD16]], ptr [[I3]], align 4
+// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP22]], 0
+// CHECK1-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
+// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV5]], 1
+// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[I3]], align 4
// CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK1: .omp.final.done:
// CHECK1-NEXT: br label [[OMP_PRECOND_END]]
@@ -620,38 +560,38 @@ int bar(int n){
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP20:![0-9]+]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17:![0-9]+]]
// CHECK1-NEXT: [[CONV5:%.*]] = sext i32 [[TMP10]] to i64
-// CHECK1-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP20]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP17]]
// CHECK1-NEXT: [[CMP6:%.*]] = icmp ule i64 [[CONV5]], [[TMP11]]
// CHECK1-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP20]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP12]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK1-NEXT: store i32 [[ADD]], ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP20]]
-// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP20]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP17]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [1000 x i16], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
-// CHECK1-NEXT: [[TMP14:%.*]] = load i16, ptr [[ARRAYIDX]], align 2, !llvm.access.group [[ACC_GRP20]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load i16, ptr [[ARRAYIDX]], align 2, !llvm.access.group [[ACC_GRP17]]
// CHECK1-NEXT: [[CONV7:%.*]] = sext i16 [[TMP14]] to i32
// CHECK1-NEXT: [[ADD8:%.*]] = add nsw i32 [[CONV7]], 1
// CHECK1-NEXT: [[CONV9:%.*]] = trunc i32 [[ADD8]] to i16
-// CHECK1-NEXT: store i16 [[CONV9]], ptr [[ARRAYIDX]], align 2, !llvm.access.group [[ACC_GRP20]]
+// CHECK1-NEXT: store i16 [[CONV9]], ptr [[ARRAYIDX]], align 2, !llvm.access.group [[ACC_GRP17]]
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP20]]
-// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP20]]
+// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
+// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK1-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP15]], [[TMP16]]
-// CHECK1-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP20]]
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP21:![0-9]+]]
+// CHECK1-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]]
// CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
@@ -722,76 +662,26 @@ int bar(int n){
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
-// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK1: cond.true:
-// CHECK1-NEXT: br label [[COND_END:%.*]]
-// CHECK1: cond.false:
-// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END]]
-// CHECK1: cond.end:
-// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
-// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23:![0-9]+]]
-// CHECK1-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP6]], 10
-// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
-// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK1-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP8]] to ptr
-// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK1-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP10]] to ptr
-// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 8, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l37_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3, i32 0), !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK1-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
-// CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP22]], 9
-// CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK1: cond.true5:
-// CHECK1-NEXT: br label [[COND_END7:%.*]]
-// CHECK1: cond.false6:
-// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: br label [[COND_END7]]
-// CHECK1: cond.end7:
-// CHECK1-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP23]], [[COND_FALSE6]] ]
-// CHECK1-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: store i32 [[TMP24]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP24:![0-9]+]]
-// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK1-NEXT: [[TMP6:%.*]] = inttoptr i64 [[TMP2]] to ptr
+// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP5]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK1-NEXT: [[TMP8:%.*]] = inttoptr i64 [[TMP4]] to ptr
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
+// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP11]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l37_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3, i32 0)
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
-// CHECK1-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP2]])
-// CHECK1-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK1-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
-// CHECK1-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
+// CHECK1-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK1: .omp.final.then:
// CHECK1-NEXT: store i32 10, ptr [[I]], align 4
// CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
@@ -832,36 +722,36 @@ int bar(int n){
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP26:![0-9]+]]
+// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP20:![0-9]+]]
// CHECK1-NEXT: [[CONV2:%.*]] = sext i32 [[TMP6]] to i64
-// CHECK1-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP26]]
+// CHECK1-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP20]]
// CHECK1-NEXT: [[CMP:%.*]] = icmp ule i64 [[CONV2]], [[TMP7]]
// CHECK1-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP26]]
+// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP20]]
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP26]]
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP26]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP20]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP20]]
// CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP9]] to i64
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
-// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP26]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP20]]
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK1-NEXT: store i32 [[ADD3]], ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP26]]
+// CHECK1-NEXT: store i32 [[ADD3]], ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP20]]
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP26]]
-// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP26]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP20]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP20]]
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
-// CHECK1-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP26]]
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP27:![0-9]+]]
+// CHECK1-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP20]]
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP21:![0-9]+]]
// CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
@@ -935,82 +825,32 @@ int bar(int n){
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99
-// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK1: cond.true:
-// CHECK1-NEXT: br label [[COND_END:%.*]]
-// CHECK1: cond.false:
-// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END]]
-// CHECK1: cond.end:
-// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
-// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP29:![0-9]+]]
-// CHECK1-NEXT: [[CMP2:%.*]] = icmp slt i32 [[TMP6]], 100
-// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
-// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[F_ADDR]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: store i32 [[TMP11]], ptr [[F_CASTED]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[TMP12:%.*]] = load i64, ptr [[F_CASTED]], align 8, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK1-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP8]] to ptr
-// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK1-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr
-// CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP17]], align 8, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK1-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP12]] to ptr
-// CHECK1-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l42_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0), !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
-// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
-// CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP24]], [[TMP25]]
-// CHECK1-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP26]], 99
-// CHECK1-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
-// CHECK1: cond.true6:
-// CHECK1-NEXT: br label [[COND_END8:%.*]]
-// CHECK1: cond.false7:
-// CHECK1-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: br label [[COND_END8]]
-// CHECK1: cond.end8:
-// CHECK1-NEXT: [[COND9:%.*]] = phi i32 [ 99, [[COND_TRUE6]] ], [ [[TMP27]], [[COND_FALSE7]] ]
-// CHECK1-NEXT: store i32 [[COND9]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: store i32 [[TMP28]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP30:![0-9]+]]
-// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[F_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP5]], ptr [[F_CASTED]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[F_CASTED]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK1-NEXT: [[TMP8:%.*]] = inttoptr i64 [[TMP2]] to ptr
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK1-NEXT: [[TMP10:%.*]] = inttoptr i64 [[TMP4]] to ptr
+// CHECK1-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK1-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP6]] to ptr
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
+// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP15]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l42_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
-// CHECK1-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP2]])
-// CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK1-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
-// CHECK1-NEXT: br i1 [[TMP30]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
+// CHECK1-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK1: .omp.final.then:
// CHECK1-NEXT: store i32 10, ptr [[I]], align 4
// CHECK1-NEXT: store i32 10, ptr [[J]], align 4
@@ -1057,54 +897,54 @@ int bar(int n){
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP32:![0-9]+]]
+// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23:![0-9]+]]
// CHECK1-NEXT: [[CONV3:%.*]] = sext i32 [[TMP6]] to i64
-// CHECK1-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP32]]
+// CHECK1-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP23]]
// CHECK1-NEXT: [[CMP:%.*]] = icmp ule i64 [[CONV3]], [[TMP7]]
// CHECK1-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 10
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP32]]
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP32]]
-// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP23]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK1-NEXT: [[DIV4:%.*]] = sdiv i32 [[TMP10]], 10
// CHECK1-NEXT: [[MUL5:%.*]] = mul nsw i32 [[DIV4]], 10
// CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL5]]
// CHECK1-NEXT: [[MUL6:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 0, [[MUL6]]
-// CHECK1-NEXT: store i32 [[ADD7]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP32]]
-// CHECK1-NEXT: store i32 10, ptr [[K]], align 4, !llvm.access.group [[ACC_GRP32]]
-// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP32]]
-// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP32]]
-// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[F_ADDR]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK1-NEXT: store i32 [[ADD7]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP23]]
+// CHECK1-NEXT: store i32 10, ptr [[K]], align 4, !llvm.access.group [[ACC_GRP23]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP23]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP23]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[F_ADDR]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK1-NEXT: [[MUL8:%.*]] = mul nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP11]], [[MUL8]]
-// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[K]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[K]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK1-NEXT: [[ADD10:%.*]] = add nsw i32 [[ADD9]], [[TMP14]]
-// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP15]] to i64
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
-// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK1-NEXT: [[IDXPROM11:%.*]] = sext i32 [[TMP16]] to i64
// CHECK1-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM11]]
-// CHECK1-NEXT: store i32 [[ADD10]], ptr [[ARRAYIDX12]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK1-NEXT: store i32 [[ADD10]], ptr [[ARRAYIDX12]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP32]]
-// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23]]
+// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK1-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP17]], [[TMP18]]
-// CHECK1-NEXT: store i32 [[ADD13]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP32]]
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP33:![0-9]+]]
+// CHECK1-NEXT: store i32 [[ADD13]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23]]
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP24:![0-9]+]]
// CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
@@ -1523,97 +1363,37 @@ int bar(int n){
// CHECK2-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP6]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP7]], [[TMP8]]
-// CHECK2-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK2: cond.true:
-// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: br label [[COND_END:%.*]]
-// CHECK2: cond.false:
-// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END]]
-// CHECK2: cond.end:
-// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ [[TMP9]], [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
-// CHECK2-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17:![0-9]+]]
-// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], 1
-// CHECK2-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP12]], [[ADD]]
-// CHECK2-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[N_ADDR]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: store i32 [[TMP16]], ptr [[N_CASTED]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[N_CASTED]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK2-NEXT: [[TMP19:%.*]] = inttoptr i32 [[TMP14]] to ptr
-// CHECK2-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK2-NEXT: [[TMP21:%.*]] = inttoptr i32 [[TMP15]] to ptr
-// CHECK2-NEXT: store ptr [[TMP21]], ptr [[TMP20]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK2-NEXT: [[TMP23:%.*]] = inttoptr i32 [[TMP17]] to ptr
-// CHECK2-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
-// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP24]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP25:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP26]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l32_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4, i32 0), !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP27]], [[TMP28]]
-// CHECK2-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP29]], [[TMP30]]
-// CHECK2-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
-// CHECK2-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP33]], [[TMP34]]
-// CHECK2-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]]
-// CHECK2: cond.true10:
-// CHECK2-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: br label [[COND_END12:%.*]]
-// CHECK2: cond.false11:
-// CHECK2-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: br label [[COND_END12]]
-// CHECK2: cond.end12:
-// CHECK2-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP35]], [[COND_TRUE10]] ], [ [[TMP36]], [[COND_FALSE11]] ]
-// CHECK2-NEXT: store i32 [[COND13]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: store i32 [[TMP37]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]]
-// CHECK2: omp.inner.for.end:
+// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK2-NEXT: store i32 [[TMP7]], ptr [[N_CASTED]], align 4
+// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[N_CASTED]], align 4
+// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK2-NEXT: [[TMP10:%.*]] = inttoptr i32 [[TMP5]] to ptr
+// CHECK2-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 4
+// CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK2-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP6]] to ptr
+// CHECK2-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4
+// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK2-NEXT: [[TMP14:%.*]] = inttoptr i32 [[TMP8]] to ptr
+// CHECK2-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 4
+// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
+// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 4
+// CHECK2-NEXT: [[TMP16:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4
+// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP17]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l32_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4, i32 0)
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
-// CHECK2-NEXT: [[TMP38:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK2-NEXT: [[TMP39:%.*]] = load i32, ptr [[TMP38]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP39]])
-// CHECK2-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK2-NEXT: [[TMP41:%.*]] = icmp ne i32 [[TMP40]], 0
-// CHECK2-NEXT: br i1 [[TMP41]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK2-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK2-NEXT: br i1 [[TMP19]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK2: .omp.final.then:
-// CHECK2-NEXT: [[TMP42:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
-// CHECK2-NEXT: [[SUB14:%.*]] = sub nsw i32 [[TMP42]], 0
-// CHECK2-NEXT: [[DIV15:%.*]] = sdiv i32 [[SUB14]], 1
-// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV15]], 1
-// CHECK2-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK2-NEXT: store i32 [[ADD16]], ptr [[I3]], align 4
+// CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK2-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP20]], 0
+// CHECK2-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
+// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV5]], 1
+// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK2-NEXT: store i32 [[ADD]], ptr [[I3]], align 4
// CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK2: .omp.final.done:
// CHECK2-NEXT: br label [[OMP_PRECOND_END]]
@@ -1670,36 +1450,36 @@ int bar(int n){
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP20:![0-9]+]]
-// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP20]]
+// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17:![0-9]+]]
+// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK2-NEXT: [[CMP4:%.*]] = icmp ule i32 [[TMP10]], [[TMP11]]
// CHECK2-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP20]]
+// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP12]], 1
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK2-NEXT: store i32 [[ADD]], ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP20]]
-// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP20]]
+// CHECK2-NEXT: store i32 [[ADD]], ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP17]]
+// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [1000 x i16], ptr [[TMP0]], i32 0, i32 [[TMP13]]
-// CHECK2-NEXT: [[TMP14:%.*]] = load i16, ptr [[ARRAYIDX]], align 2, !llvm.access.group [[ACC_GRP20]]
+// CHECK2-NEXT: [[TMP14:%.*]] = load i16, ptr [[ARRAYIDX]], align 2, !llvm.access.group [[ACC_GRP17]]
// CHECK2-NEXT: [[CONV:%.*]] = sext i16 [[TMP14]] to i32
// CHECK2-NEXT: [[ADD5:%.*]] = add nsw i32 [[CONV]], 1
// CHECK2-NEXT: [[CONV6:%.*]] = trunc i32 [[ADD5]] to i16
-// CHECK2-NEXT: store i16 [[CONV6]], ptr [[ARRAYIDX]], align 2, !llvm.access.group [[ACC_GRP20]]
+// CHECK2-NEXT: store i16 [[CONV6]], ptr [[ARRAYIDX]], align 2, !llvm.access.group [[ACC_GRP17]]
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue:
// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP20]]
-// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP20]]
+// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
+// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK2-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP15]], [[TMP16]]
-// CHECK2-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP20]]
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP21:![0-9]+]]
+// CHECK2-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
+// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]]
// CHECK2: omp.inner.for.end:
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
@@ -1770,74 +1550,24 @@ int bar(int n){
// CHECK2-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
-// CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK2: cond.true:
-// CHECK2-NEXT: br label [[COND_END:%.*]]
-// CHECK2: cond.false:
-// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END]]
-// CHECK2: cond.end:
-// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
-// CHECK2-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23:![0-9]+]]
-// CHECK2-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP6]], 10
-// CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK2-NEXT: [[TMP10:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK2-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK2-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP8]] to ptr
-// CHECK2-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP13]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l37_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3, i32 0), !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK2-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK2-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK2-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
-// CHECK2-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK2: cond.true5:
-// CHECK2-NEXT: br label [[COND_END7:%.*]]
-// CHECK2: cond.false6:
-// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: br label [[COND_END7]]
-// CHECK2: cond.end7:
-// CHECK2-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK2-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23]]
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP24:![0-9]+]]
-// CHECK2: omp.inner.for.end:
+// CHECK2-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK2-NEXT: [[TMP4:%.*]] = inttoptr i32 [[TMP1]] to ptr
+// CHECK2-NEXT: store ptr [[TMP4]], ptr [[TMP3]], align 4
+// CHECK2-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK2-NEXT: [[TMP6:%.*]] = inttoptr i32 [[TMP2]] to ptr
+// CHECK2-NEXT: store ptr [[TMP6]], ptr [[TMP5]], align 4
+// CHECK2-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP7]], align 4
+// CHECK2-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
+// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP9]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l37_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3, i32 0)
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
-// CHECK2-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP2]])
-// CHECK2-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK2-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
-// CHECK2-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK2-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP10]], 0
+// CHECK2-NEXT: br i1 [[TMP11]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK2: .omp.final.then:
// CHECK2-NEXT: store i32 10, ptr [[I]], align 4
// CHECK2-NEXT: br label [[DOTOMP_FINAL_DONE]]
@@ -1876,34 +1606,34 @@ int bar(int n){
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP26:![0-9]+]]
-// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP26]]
+// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP20:![0-9]+]]
+// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP20]]
// CHECK2-NEXT: [[CMP:%.*]] = icmp ule i32 [[TMP6]], [[TMP7]]
// CHECK2-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP26]]
+// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP20]]
// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK2-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP26]]
-// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP26]]
+// CHECK2-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP20]]
+// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP20]]
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP9]]
-// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP26]]
+// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP20]]
// CHECK2-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK2-NEXT: store i32 [[ADD1]], ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP26]]
+// CHECK2-NEXT: store i32 [[ADD1]], ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP20]]
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue:
// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP26]]
-// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP26]]
+// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP20]]
+// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP20]]
// CHECK2-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
-// CHECK2-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP26]]
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP27:![0-9]+]]
+// CHECK2-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP20]]
+// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP21:![0-9]+]]
// CHECK2: omp.inner.for.end:
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
@@ -1977,80 +1707,30 @@ int bar(int n){
// CHECK2-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99
-// CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK2: cond.true:
-// CHECK2-NEXT: br label [[COND_END:%.*]]
-// CHECK2: cond.false:
-// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END]]
-// CHECK2: cond.end:
-// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
-// CHECK2-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP29:![0-9]+]]
-// CHECK2-NEXT: [[CMP2:%.*]] = icmp slt i32 [[TMP6]], 100
-// CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[F_ADDR]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: store i32 [[TMP9]], ptr [[F_CASTED]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[F_CASTED]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK2-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK2-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK2-NEXT: [[TMP14:%.*]] = inttoptr i32 [[TMP8]] to ptr
-// CHECK2-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
-// CHECK2-NEXT: [[TMP17:%.*]] = inttoptr i32 [[TMP10]] to ptr
-// CHECK2-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l42_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4, i32 0), !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK2-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
-// CHECK2-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
-// CHECK2-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP24]], 99
-// CHECK2-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
-// CHECK2: cond.true6:
-// CHECK2-NEXT: br label [[COND_END8:%.*]]
-// CHECK2: cond.false7:
-// CHECK2-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: br label [[COND_END8]]
-// CHECK2: cond.end8:
-// CHECK2-NEXT: [[COND9:%.*]] = phi i32 [ 99, [[COND_TRUE6]] ], [ [[TMP25]], [[COND_FALSE7]] ]
-// CHECK2-NEXT: store i32 [[COND9]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: store i32 [[TMP26]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP29]]
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP30:![0-9]+]]
-// CHECK2: omp.inner.for.end:
+// CHECK2-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[F_ADDR]], align 4
+// CHECK2-NEXT: store i32 [[TMP3]], ptr [[F_CASTED]], align 4
+// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[F_CASTED]], align 4
+// CHECK2-NEXT: [[TMP5:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK2-NEXT: [[TMP6:%.*]] = inttoptr i32 [[TMP1]] to ptr
+// CHECK2-NEXT: store ptr [[TMP6]], ptr [[TMP5]], align 4
+// CHECK2-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK2-NEXT: [[TMP8:%.*]] = inttoptr i32 [[TMP2]] to ptr
+// CHECK2-NEXT: store ptr [[TMP8]], ptr [[TMP7]], align 4
+// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP9]], align 4
+// CHECK2-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
+// CHECK2-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP4]] to ptr
+// CHECK2-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4
+// CHECK2-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
+// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP13]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l42_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4, i32 0)
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
-// CHECK2-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP2]])
-// CHECK2-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK2-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
-// CHECK2-NEXT: br i1 [[TMP28]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK2-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
+// CHECK2-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK2: .omp.final.then:
// CHECK2-NEXT: store i32 10, ptr [[I]], align 4
// CHECK2-NEXT: store i32 10, ptr [[J]], align 4
@@ -2095,51 +1775,51 @@ int bar(int n){
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP32:![0-9]+]]
-// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23:![0-9]+]]
+// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK2-NEXT: [[CMP:%.*]] = icmp ule i32 [[TMP6]], [[TMP7]]
// CHECK2-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK2-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 10
// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK2-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP32]]
-// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP32]]
-// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK2-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP23]]
+// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23]]
+// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK2-NEXT: [[DIV2:%.*]] = sdiv i32 [[TMP10]], 10
// CHECK2-NEXT: [[MUL3:%.*]] = mul nsw i32 [[DIV2]], 10
// CHECK2-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL3]]
// CHECK2-NEXT: [[MUL4:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK2-NEXT: [[ADD5:%.*]] = add nsw i32 0, [[MUL4]]
-// CHECK2-NEXT: store i32 [[ADD5]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP32]]
-// CHECK2-NEXT: store i32 10, ptr [[K]], align 4, !llvm.access.group [[ACC_GRP32]]
-// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP32]]
-// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP32]]
-// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[F_ADDR]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK2-NEXT: store i32 [[ADD5]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP23]]
+// CHECK2-NEXT: store i32 10, ptr [[K]], align 4, !llvm.access.group [[ACC_GRP23]]
+// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP23]]
+// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP23]]
+// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[F_ADDR]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK2-NEXT: [[MUL6:%.*]] = mul nsw i32 [[TMP12]], [[TMP13]]
// CHECK2-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP11]], [[MUL6]]
-// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[K]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[K]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK2-NEXT: [[ADD8:%.*]] = add nsw i32 [[ADD7]], [[TMP14]]
-// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[TMP0]], i32 0, i32 [[TMP15]]
-// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK2-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP16]]
-// CHECK2-NEXT: store i32 [[ADD8]], ptr [[ARRAYIDX9]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK2-NEXT: store i32 [[ADD8]], ptr [[ARRAYIDX9]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue:
// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP32]]
-// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP32]]
+// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23]]
+// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP23]]
// CHECK2-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP17]], [[TMP18]]
-// CHECK2-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP32]]
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP33:![0-9]+]]
+// CHECK2-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP23]]
+// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP24:![0-9]+]]
// CHECK2: omp.inner.for.end:
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
diff --git a/clang/test/OpenMP/nvptx_target_teams_generic_loop_codegen.cpp b/clang/test/OpenMP/nvptx_target_teams_generic_loop_codegen.cpp
index a58f4e5f0997b..9ff6282942161 100644
--- a/clang/test/OpenMP/nvptx_target_teams_generic_loop_codegen.cpp
+++ b/clang/test/OpenMP/nvptx_target_teams_generic_loop_codegen.cpp
@@ -137,89 +137,29 @@ int bar(int n){
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP6]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = zext i32 [[TMP5]] to i64
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP7]], [[TMP8]]
-// CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK1: cond.true:
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: br label [[COND_END:%.*]]
-// CHECK1: cond.false:
-// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END]]
-// CHECK1: cond.end:
-// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP9]], [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
-// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], 1
-// CHECK1-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP12]], [[ADD]]
-// CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP15:%.*]] = zext i32 [[TMP14]] to i64
-// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
-// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK1-NEXT: store i32 [[TMP18]], ptr [[N_CASTED]], align 4
-// CHECK1-NEXT: [[TMP19:%.*]] = load i64, ptr [[N_CASTED]], align 8
-// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK1-NEXT: [[TMP21:%.*]] = inttoptr i64 [[TMP15]] to ptr
-// CHECK1-NEXT: store ptr [[TMP21]], ptr [[TMP20]], align 8
-// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK1-NEXT: [[TMP23:%.*]] = inttoptr i64 [[TMP17]] to ptr
-// CHECK1-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 8
-// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK1-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP19]] to ptr
-// CHECK1-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 8
-// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 8
-// CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
-// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l28_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP29]], [[TMP30]]
-// CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
-// CHECK1-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
-// CHECK1-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP35]], [[TMP36]]
-// CHECK1-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]]
-// CHECK1: cond.true10:
-// CHECK1-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: br label [[COND_END12:%.*]]
-// CHECK1: cond.false11:
-// CHECK1-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END12]]
-// CHECK1: cond.end12:
-// CHECK1-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP37]], [[COND_TRUE10]] ], [ [[TMP38]], [[COND_FALSE11]] ]
-// CHECK1-NEXT: store i32 [[COND13]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP39]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP9]], ptr [[N_CASTED]], align 4
+// CHECK1-NEXT: [[TMP10:%.*]] = load i64, ptr [[N_CASTED]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK1-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP6]] to ptr
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK1-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP8]] to ptr
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK1-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr
+// CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP17]], align 8
+// CHECK1-NEXT: [[TMP18:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP18]], align 4
+// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP19]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l28_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
-// CHECK1-NEXT: [[TMP40:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP41:%.*]] = load i32, ptr [[TMP40]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP41]])
// CHECK1-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1: omp.precond.end:
// CHECK1-NEXT: ret void
@@ -276,7 +216,7 @@ int bar(int n){
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP8]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -313,7 +253,7 @@ int bar(int n){
// CHECK1: omp.loop.exit:
// CHECK1-NEXT: [[TMP17:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP17]], align 4
-// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP18]])
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP18]])
// CHECK1-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1: omp.precond.end:
// CHECK1-NEXT: ret void
@@ -376,73 +316,23 @@ int bar(int n){
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
-// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK1: cond.true:
-// CHECK1-NEXT: br label [[COND_END:%.*]]
-// CHECK1: cond.false:
-// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END]]
-// CHECK1: cond.end:
-// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
-// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP6]], 10
-// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
-// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK1-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP8]] to ptr
-// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8
-// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK1-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP10]] to ptr
-// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8
-// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 8
-// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l33_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3, i32 0)
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK1-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
-// CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP22]], 9
-// CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK1: cond.true5:
-// CHECK1-NEXT: br label [[COND_END7:%.*]]
-// CHECK1: cond.false6:
-// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END7]]
-// CHECK1: cond.end7:
-// CHECK1-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP23]], [[COND_FALSE6]] ]
-// CHECK1-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP24]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK1-NEXT: [[TMP6:%.*]] = inttoptr i64 [[TMP2]] to ptr
+// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP5]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK1-NEXT: [[TMP8:%.*]] = inttoptr i64 [[TMP4]] to ptr
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
+// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP11]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l33_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3, i32 0)
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
-// CHECK1-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP2]])
// CHECK1-NEXT: ret void
//
//
@@ -479,7 +369,7 @@ int bar(int n){
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -512,7 +402,7 @@ int bar(int n){
// CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
-// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP4]])
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK1-NEXT: ret void
//
//
@@ -575,79 +465,29 @@ int bar(int n){
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99
-// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK1: cond.true:
-// CHECK1-NEXT: br label [[COND_END:%.*]]
-// CHECK1: cond.false:
-// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END]]
-// CHECK1: cond.end:
-// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
-// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[CMP2:%.*]] = icmp slt i32 [[TMP6]], 100
-// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
-// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[F_ADDR]], align 4
-// CHECK1-NEXT: store i32 [[TMP11]], ptr [[F_CASTED]], align 4
-// CHECK1-NEXT: [[TMP12:%.*]] = load i64, ptr [[F_CASTED]], align 8
-// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK1-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP8]] to ptr
-// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8
-// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK1-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr
-// CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8
-// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP17]], align 8
-// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK1-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP12]] to ptr
-// CHECK1-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8
-// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l38_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
-// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
-// CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP24]], [[TMP25]]
-// CHECK1-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP26]], 99
-// CHECK1-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
-// CHECK1: cond.true6:
-// CHECK1-NEXT: br label [[COND_END8:%.*]]
-// CHECK1: cond.false7:
-// CHECK1-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END8]]
-// CHECK1: cond.end8:
-// CHECK1-NEXT: [[COND9:%.*]] = phi i32 [ 99, [[COND_TRUE6]] ], [ [[TMP27]], [[COND_FALSE7]] ]
-// CHECK1-NEXT: store i32 [[COND9]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP28]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[F_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP5]], ptr [[F_CASTED]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[F_CASTED]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK1-NEXT: [[TMP8:%.*]] = inttoptr i64 [[TMP2]] to ptr
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK1-NEXT: [[TMP10:%.*]] = inttoptr i64 [[TMP4]] to ptr
+// CHECK1-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK1-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP6]] to ptr
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
+// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP15]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l38_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
-// CHECK1-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP2]])
// CHECK1-NEXT: ret void
//
//
@@ -689,7 +529,7 @@ int bar(int n){
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -740,7 +580,7 @@ int bar(int n){
// CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
-// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP4]])
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK1-NEXT: ret void
//
//
@@ -834,87 +674,27 @@ int bar(int n){
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK1-NEXT: [[CONV11:%.*]] = zext i32 [[NVPTX_NUM_THREADS]] to i64
-// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_init_8(ptr @[[GLOB2]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 [[CONV11]])
-// CHECK1-NEXT: [[TMP10:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK1-NEXT: [[CMP12:%.*]] = icmp sgt i64 [[TMP10]], [[TMP11]]
-// CHECK1-NEXT: br i1 [[CMP12]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK1: cond.true:
-// CHECK1-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK1-NEXT: br label [[COND_END:%.*]]
-// CHECK1: cond.false:
-// CHECK1-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: br label [[COND_END]]
-// CHECK1: cond.end:
-// CHECK1-NEXT: [[COND:%.*]] = phi i64 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
-// CHECK1-NEXT: store i64 [[COND]], ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK1-NEXT: store i64 [[TMP14]], ptr [[DOTOMP_IV]], align 8
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
-// CHECK1-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK1-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP16]], 1
-// CHECK1-NEXT: [[CMP13:%.*]] = icmp slt i64 [[TMP15]], [[ADD]]
-// CHECK1-NEXT: br i1 [[CMP13]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK1-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK1-NEXT: store i32 [[TMP19]], ptr [[N_CASTED]], align 4
-// CHECK1-NEXT: [[TMP20:%.*]] = load i64, ptr [[N_CASTED]], align 8
-// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK1-NEXT: [[TMP22:%.*]] = inttoptr i64 [[TMP17]] to ptr
-// CHECK1-NEXT: store ptr [[TMP22]], ptr [[TMP21]], align 8
-// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK1-NEXT: [[TMP24:%.*]] = inttoptr i64 [[TMP18]] to ptr
-// CHECK1-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 8
-// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK1-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP20]] to ptr
-// CHECK1-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8
-// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP27]], align 8
-// CHECK1-NEXT: [[TMP28:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[TMP28]], align 4
-// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP29]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l46_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP30:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
-// CHECK1-NEXT: [[TMP31:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
-// CHECK1-NEXT: [[ADD14:%.*]] = add nsw i64 [[TMP30]], [[TMP31]]
-// CHECK1-NEXT: store i64 [[ADD14]], ptr [[DOTOMP_IV]], align 8
-// CHECK1-NEXT: [[TMP32:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK1-NEXT: [[TMP33:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
-// CHECK1-NEXT: [[ADD15:%.*]] = add nsw i64 [[TMP32]], [[TMP33]]
-// CHECK1-NEXT: store i64 [[ADD15]], ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK1-NEXT: [[TMP34:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: [[TMP35:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
-// CHECK1-NEXT: [[ADD16:%.*]] = add nsw i64 [[TMP34]], [[TMP35]]
-// CHECK1-NEXT: store i64 [[ADD16]], ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: [[TMP36:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: [[TMP37:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK1-NEXT: [[CMP17:%.*]] = icmp sgt i64 [[TMP36]], [[TMP37]]
-// CHECK1-NEXT: br i1 [[CMP17]], label [[COND_TRUE18:%.*]], label [[COND_FALSE19:%.*]]
-// CHECK1: cond.true18:
-// CHECK1-NEXT: [[TMP38:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK1-NEXT: br label [[COND_END20:%.*]]
-// CHECK1: cond.false19:
-// CHECK1-NEXT: [[TMP39:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: br label [[COND_END20]]
-// CHECK1: cond.end20:
-// CHECK1-NEXT: [[COND21:%.*]] = phi i64 [ [[TMP38]], [[COND_TRUE18]] ], [ [[TMP39]], [[COND_FALSE19]] ]
-// CHECK1-NEXT: store i64 [[COND21]], ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK1-NEXT: [[TMP40:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK1-NEXT: store i64 [[TMP40]], ptr [[DOTOMP_IV]], align 8
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP10]], ptr [[N_CASTED]], align 4
+// CHECK1-NEXT: [[TMP11:%.*]] = load i64, ptr [[N_CASTED]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK1-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP8]] to ptr
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK1-NEXT: [[TMP15:%.*]] = inttoptr i64 [[TMP9]] to ptr
+// CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK1-NEXT: [[TMP17:%.*]] = inttoptr i64 [[TMP11]] to ptr
+// CHECK1-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP18]], align 8
+// CHECK1-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4
+// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP20]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l46_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
-// CHECK1-NEXT: [[TMP41:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP42:%.*]] = load i32, ptr [[TMP41]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP42]])
// CHECK1-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1: omp.precond.end:
// CHECK1-NEXT: ret void
@@ -986,7 +766,7 @@ int bar(int n){
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
-// CHECK1-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB3]], i32 [[TMP11]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
+// CHECK1-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB2]], i32 [[TMP11]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK1-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK1-NEXT: store i64 [[TMP12]], ptr [[DOTOMP_IV]], align 8
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -1050,7 +830,7 @@ int bar(int n){
// CHECK1: omp.loop.exit:
// CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
-// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP28]])
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP28]])
// CHECK1-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1: omp.precond.end:
// CHECK1-NEXT: ret void
@@ -1133,92 +913,32 @@ int bar(int n){
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP6]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = zext i32 [[TMP5]] to i64
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP7]], [[TMP8]]
-// CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK1: cond.true:
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: br label [[COND_END:%.*]]
-// CHECK1: cond.false:
-// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END]]
-// CHECK1: cond.end:
-// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP9]], [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
-// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], 1
-// CHECK1-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP12]], [[ADD]]
-// CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP15:%.*]] = zext i32 [[TMP14]] to i64
-// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
-// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK1-NEXT: store i32 [[TMP18]], ptr [[N_CASTED]], align 4
-// CHECK1-NEXT: [[TMP19:%.*]] = load i64, ptr [[N_CASTED]], align 8
-// CHECK1-NEXT: [[TMP20:%.*]] = load ptr, ptr [[V_ADDR]], align 8
-// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK1-NEXT: [[TMP22:%.*]] = inttoptr i64 [[TMP15]] to ptr
-// CHECK1-NEXT: store ptr [[TMP22]], ptr [[TMP21]], align 8
-// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK1-NEXT: [[TMP24:%.*]] = inttoptr i64 [[TMP17]] to ptr
-// CHECK1-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 8
-// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK1-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP19]] to ptr
-// CHECK1-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8
-// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP27]], align 8
-// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
-// CHECK1-NEXT: store ptr [[TMP20]], ptr [[TMP28]], align 8
-// CHECK1-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
-// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP30]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l53_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5, i32 0)
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
-// CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
-// CHECK1-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK1-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP35]], [[TMP36]]
-// CHECK1-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP37]], [[TMP38]]
-// CHECK1-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]]
-// CHECK1: cond.true10:
-// CHECK1-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: br label [[COND_END12:%.*]]
-// CHECK1: cond.false11:
-// CHECK1-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END12]]
-// CHECK1: cond.end12:
-// CHECK1-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP39]], [[COND_TRUE10]] ], [ [[TMP40]], [[COND_FALSE11]] ]
-// CHECK1-NEXT: store i32 [[COND13]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP41]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP9]], ptr [[N_CASTED]], align 4
+// CHECK1-NEXT: [[TMP10:%.*]] = load i64, ptr [[N_CASTED]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[V_ADDR]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK1-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP6]] to ptr
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK1-NEXT: [[TMP15:%.*]] = inttoptr i64 [[TMP8]] to ptr
+// CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK1-NEXT: [[TMP17:%.*]] = inttoptr i64 [[TMP10]] to ptr
+// CHECK1-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP18]], align 8
+// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
+// CHECK1-NEXT: store ptr [[TMP11]], ptr [[TMP19]], align 8
+// CHECK1-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
+// CHECK1-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP21]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l53_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5, i32 0)
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
-// CHECK1-NEXT: [[TMP42:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP43:%.*]] = load i32, ptr [[TMP42]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP43]])
// CHECK1-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1: omp.precond.end:
// CHECK1-NEXT: ret void
@@ -1277,7 +997,7 @@ int bar(int n){
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP8]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -1315,7 +1035,7 @@ int bar(int n){
// CHECK1: omp.loop.exit:
// CHECK1-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4
-// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP20]])
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP20]])
// CHECK1-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1: omp.precond.end:
// CHECK1-NEXT: ret void
@@ -1393,89 +1113,29 @@ int bar(int n){
// CHECK2-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP6]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK2-NEXT: [[TMP6:%.*]] = zext i32 [[TMP5]] to i64
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP7]], [[TMP8]]
-// CHECK2-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK2: cond.true:
-// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: br label [[COND_END:%.*]]
-// CHECK2: cond.false:
-// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END]]
-// CHECK2: cond.end:
-// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ [[TMP9]], [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
-// CHECK2-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], 1
-// CHECK2-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP12]], [[ADD]]
-// CHECK2-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP15:%.*]] = zext i32 [[TMP14]] to i64
-// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
-// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK2-NEXT: store i32 [[TMP18]], ptr [[N_CASTED]], align 4
-// CHECK2-NEXT: [[TMP19:%.*]] = load i64, ptr [[N_CASTED]], align 8
-// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK2-NEXT: [[TMP21:%.*]] = inttoptr i64 [[TMP15]] to ptr
-// CHECK2-NEXT: store ptr [[TMP21]], ptr [[TMP20]], align 8
-// CHECK2-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK2-NEXT: [[TMP23:%.*]] = inttoptr i64 [[TMP17]] to ptr
-// CHECK2-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 8
-// CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK2-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP19]] to ptr
-// CHECK2-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 8
-// CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 8
-// CHECK2-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
-// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l28_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP29]], [[TMP30]]
-// CHECK2-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
-// CHECK2-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
-// CHECK2-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP35]], [[TMP36]]
-// CHECK2-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]]
-// CHECK2: cond.true10:
-// CHECK2-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: br label [[COND_END12:%.*]]
-// CHECK2: cond.false11:
-// CHECK2-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END12]]
-// CHECK2: cond.end12:
-// CHECK2-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP37]], [[COND_TRUE10]] ], [ [[TMP38]], [[COND_FALSE11]] ]
-// CHECK2-NEXT: store i32 [[COND13]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP39]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK2: omp.inner.for.end:
+// CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK2-NEXT: store i32 [[TMP9]], ptr [[N_CASTED]], align 4
+// CHECK2-NEXT: [[TMP10:%.*]] = load i64, ptr [[N_CASTED]], align 8
+// CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK2-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP6]] to ptr
+// CHECK2-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8
+// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK2-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP8]] to ptr
+// CHECK2-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8
+// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK2-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr
+// CHECK2-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8
+// CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP17]], align 8
+// CHECK2-NEXT: [[TMP18:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK2-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP18]], align 4
+// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP19]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l28_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
-// CHECK2-NEXT: [[TMP40:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP41:%.*]] = load i32, ptr [[TMP40]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP41]])
// CHECK2-NEXT: br label [[OMP_PRECOND_END]]
// CHECK2: omp.precond.end:
// CHECK2-NEXT: ret void
@@ -1532,7 +1192,7 @@ int bar(int n){
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP8]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -1569,7 +1229,7 @@ int bar(int n){
// CHECK2: omp.loop.exit:
// CHECK2-NEXT: [[TMP17:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP17]], align 4
-// CHECK2-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP18]])
+// CHECK2-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP18]])
// CHECK2-NEXT: br label [[OMP_PRECOND_END]]
// CHECK2: omp.precond.end:
// CHECK2-NEXT: ret void
@@ -1632,73 +1292,23 @@ int bar(int n){
// CHECK2-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK2-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
-// CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK2: cond.true:
-// CHECK2-NEXT: br label [[COND_END:%.*]]
-// CHECK2: cond.false:
-// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END]]
-// CHECK2: cond.end:
-// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
-// CHECK2-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP6]], 10
-// CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
-// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
-// CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK2-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP8]] to ptr
-// CHECK2-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8
-// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK2-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP10]] to ptr
-// CHECK2-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8
-// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 8
-// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l33_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3, i32 0)
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK2-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK2-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
-// CHECK2-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP22]], 9
-// CHECK2-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK2: cond.true5:
-// CHECK2-NEXT: br label [[COND_END7:%.*]]
-// CHECK2: cond.false6:
-// CHECK2-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END7]]
-// CHECK2: cond.end7:
-// CHECK2-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP23]], [[COND_FALSE6]] ]
-// CHECK2-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP24]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK2: omp.inner.for.end:
+// CHECK2-NEXT: [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
+// CHECK2-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK2-NEXT: [[TMP6:%.*]] = inttoptr i64 [[TMP2]] to ptr
+// CHECK2-NEXT: store ptr [[TMP6]], ptr [[TMP5]], align 8
+// CHECK2-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK2-NEXT: [[TMP8:%.*]] = inttoptr i64 [[TMP4]] to ptr
+// CHECK2-NEXT: store ptr [[TMP8]], ptr [[TMP7]], align 8
+// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP9]], align 8
+// CHECK2-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
+// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP11]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l33_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3, i32 0)
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
-// CHECK2-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP2]])
// CHECK2-NEXT: ret void
//
//
@@ -1735,7 +1345,7 @@ int bar(int n){
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -1768,7 +1378,7 @@ int bar(int n){
// CHECK2: omp.inner.for.end:
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
-// CHECK2-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP4]])
+// CHECK2-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK2-NEXT: ret void
//
//
@@ -1831,79 +1441,29 @@ int bar(int n){
// CHECK2-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK2-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99
-// CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK2: cond.true:
-// CHECK2-NEXT: br label [[COND_END:%.*]]
-// CHECK2: cond.false:
-// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END]]
-// CHECK2: cond.end:
-// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
-// CHECK2-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[CMP2:%.*]] = icmp slt i32 [[TMP6]], 100
-// CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
-// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
-// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[F_ADDR]], align 4
-// CHECK2-NEXT: store i32 [[TMP11]], ptr [[F_CASTED]], align 4
-// CHECK2-NEXT: [[TMP12:%.*]] = load i64, ptr [[F_CASTED]], align 8
-// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK2-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP8]] to ptr
-// CHECK2-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8
-// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK2-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr
-// CHECK2-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8
-// CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP17]], align 8
-// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK2-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP12]] to ptr
-// CHECK2-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8
-// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l38_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
-// CHECK2-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
-// CHECK2-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP24]], [[TMP25]]
-// CHECK2-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP26]], 99
-// CHECK2-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
-// CHECK2: cond.true6:
-// CHECK2-NEXT: br label [[COND_END8:%.*]]
-// CHECK2: cond.false7:
-// CHECK2-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END8]]
-// CHECK2: cond.end8:
-// CHECK2-NEXT: [[COND9:%.*]] = phi i32 [ 99, [[COND_TRUE6]] ], [ [[TMP27]], [[COND_FALSE7]] ]
-// CHECK2-NEXT: store i32 [[COND9]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP28]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK2: omp.inner.for.end:
+// CHECK2-NEXT: [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
+// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[F_ADDR]], align 4
+// CHECK2-NEXT: store i32 [[TMP5]], ptr [[F_CASTED]], align 4
+// CHECK2-NEXT: [[TMP6:%.*]] = load i64, ptr [[F_CASTED]], align 8
+// CHECK2-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK2-NEXT: [[TMP8:%.*]] = inttoptr i64 [[TMP2]] to ptr
+// CHECK2-NEXT: store ptr [[TMP8]], ptr [[TMP7]], align 8
+// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK2-NEXT: [[TMP10:%.*]] = inttoptr i64 [[TMP4]] to ptr
+// CHECK2-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 8
+// CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP11]], align 8
+// CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK2-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP6]] to ptr
+// CHECK2-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
+// CHECK2-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
+// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP15]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l38_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
-// CHECK2-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP2]])
// CHECK2-NEXT: ret void
//
//
@@ -1945,7 +1505,7 @@ int bar(int n){
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -1996,7 +1556,7 @@ int bar(int n){
// CHECK2: omp.inner.for.end:
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
-// CHECK2-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP4]])
+// CHECK2-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK2-NEXT: ret void
//
//
@@ -2087,89 +1647,29 @@ int bar(int n){
// CHECK2-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK2-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK2-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4
-// CHECK2-NEXT: [[CMP10:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]]
-// CHECK2-NEXT: br i1 [[CMP10]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK2: cond.true:
-// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4
-// CHECK2-NEXT: br label [[COND_END:%.*]]
-// CHECK2: cond.false:
-// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END]]
-// CHECK2: cond.end:
-// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
-// CHECK2-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4
-// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], 1
-// CHECK2-NEXT: [[CMP11:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
-// CHECK2-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64
-// CHECK2-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64
-// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK2-NEXT: store i32 [[TMP21]], ptr [[N_CASTED]], align 4
-// CHECK2-NEXT: [[TMP22:%.*]] = load i64, ptr [[N_CASTED]], align 8
-// CHECK2-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK2-NEXT: [[TMP24:%.*]] = inttoptr i64 [[TMP18]] to ptr
-// CHECK2-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 8
-// CHECK2-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK2-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP20]] to ptr
-// CHECK2-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8
-// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK2-NEXT: [[TMP28:%.*]] = inttoptr i64 [[TMP22]] to ptr
-// CHECK2-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 8
-// CHECK2-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP29]], align 8
-// CHECK2-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
-// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP31]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l46_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP32]], [[TMP33]]
-// CHECK2-NEXT: store i32 [[ADD12]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP34]], [[TMP35]]
-// CHECK2-NEXT: store i32 [[ADD13]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP36]], [[TMP37]]
-// CHECK2-NEXT: store i32 [[ADD14]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4
-// CHECK2-NEXT: [[CMP15:%.*]] = icmp sgt i32 [[TMP38]], [[TMP39]]
-// CHECK2-NEXT: br i1 [[CMP15]], label [[COND_TRUE16:%.*]], label [[COND_FALSE17:%.*]]
-// CHECK2: cond.true16:
-// CHECK2-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4
-// CHECK2-NEXT: br label [[COND_END18:%.*]]
-// CHECK2: cond.false17:
-// CHECK2-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END18]]
-// CHECK2: cond.end18:
-// CHECK2-NEXT: [[COND19:%.*]] = phi i32 [ [[TMP40]], [[COND_TRUE16]] ], [ [[TMP41]], [[COND_FALSE17]] ]
-// CHECK2-NEXT: store i32 [[COND19]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP42:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP42]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK2: omp.inner.for.end:
+// CHECK2-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
+// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK2-NEXT: store i32 [[TMP12]], ptr [[N_CASTED]], align 4
+// CHECK2-NEXT: [[TMP13:%.*]] = load i64, ptr [[N_CASTED]], align 8
+// CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK2-NEXT: [[TMP15:%.*]] = inttoptr i64 [[TMP9]] to ptr
+// CHECK2-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8
+// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK2-NEXT: [[TMP17:%.*]] = inttoptr i64 [[TMP11]] to ptr
+// CHECK2-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8
+// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK2-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP13]] to ptr
+// CHECK2-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8
+// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP20]], align 8
+// CHECK2-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
+// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP22]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l46_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4, i32 0)
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
-// CHECK2-NEXT: [[TMP43:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP44:%.*]] = load i32, ptr [[TMP43]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP44]])
// CHECK2-NEXT: br label [[OMP_PRECOND_END]]
// CHECK2: omp.precond.end:
// CHECK2-NEXT: ret void
@@ -2241,7 +1741,7 @@ int bar(int n){
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
-// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP11]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP11]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -2301,7 +1801,7 @@ int bar(int n){
// CHECK2: omp.loop.exit:
// CHECK2-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
-// CHECK2-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP28]])
+// CHECK2-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP28]])
// CHECK2-NEXT: br label [[OMP_PRECOND_END]]
// CHECK2: omp.precond.end:
// CHECK2-NEXT: ret void
@@ -2384,92 +1884,32 @@ int bar(int n){
// CHECK2-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP6]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK2-NEXT: [[TMP6:%.*]] = zext i32 [[TMP5]] to i64
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP7]], [[TMP8]]
-// CHECK2-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK2: cond.true:
-// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: br label [[COND_END:%.*]]
-// CHECK2: cond.false:
-// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END]]
-// CHECK2: cond.end:
-// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ [[TMP9]], [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
-// CHECK2-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], 1
-// CHECK2-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP12]], [[ADD]]
-// CHECK2-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP15:%.*]] = zext i32 [[TMP14]] to i64
-// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
-// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK2-NEXT: store i32 [[TMP18]], ptr [[N_CASTED]], align 4
-// CHECK2-NEXT: [[TMP19:%.*]] = load i64, ptr [[N_CASTED]], align 8
-// CHECK2-NEXT: [[TMP20:%.*]] = load ptr, ptr [[V_ADDR]], align 8
-// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
-// CHECK2-NEXT: [[TMP22:%.*]] = inttoptr i64 [[TMP15]] to ptr
-// CHECK2-NEXT: store ptr [[TMP22]], ptr [[TMP21]], align 8
-// CHECK2-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
-// CHECK2-NEXT: [[TMP24:%.*]] = inttoptr i64 [[TMP17]] to ptr
-// CHECK2-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 8
-// CHECK2-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
-// CHECK2-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP19]] to ptr
-// CHECK2-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8
-// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
-// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP27]], align 8
-// CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
-// CHECK2-NEXT: store ptr [[TMP20]], ptr [[TMP28]], align 8
-// CHECK2-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
-// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP30]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l53_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5, i32 0)
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
-// CHECK2-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
-// CHECK2-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK2-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP35]], [[TMP36]]
-// CHECK2-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP37]], [[TMP38]]
-// CHECK2-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]]
-// CHECK2: cond.true10:
-// CHECK2-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK2-NEXT: br label [[COND_END12:%.*]]
-// CHECK2: cond.false11:
-// CHECK2-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END12]]
-// CHECK2: cond.end12:
-// CHECK2-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP39]], [[COND_TRUE10]] ], [ [[TMP40]], [[COND_FALSE11]] ]
-// CHECK2-NEXT: store i32 [[COND13]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP41]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK2: omp.inner.for.end:
+// CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK2-NEXT: store i32 [[TMP9]], ptr [[N_CASTED]], align 4
+// CHECK2-NEXT: [[TMP10:%.*]] = load i64, ptr [[N_CASTED]], align 8
+// CHECK2-NEXT: [[TMP11:%.*]] = load ptr, ptr [[V_ADDR]], align 8
+// CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
+// CHECK2-NEXT: [[TMP13:%.*]] = inttoptr i64 [[TMP6]] to ptr
+// CHECK2-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
+// CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
+// CHECK2-NEXT: [[TMP15:%.*]] = inttoptr i64 [[TMP8]] to ptr
+// CHECK2-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8
+// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
+// CHECK2-NEXT: [[TMP17:%.*]] = inttoptr i64 [[TMP10]] to ptr
+// CHECK2-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8
+// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
+// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP18]], align 8
+// CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
+// CHECK2-NEXT: store ptr [[TMP11]], ptr [[TMP19]], align 8
+// CHECK2-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
+// CHECK2-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP21]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l53_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5, i32 0)
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
-// CHECK2-NEXT: [[TMP42:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK2-NEXT: [[TMP43:%.*]] = load i32, ptr [[TMP42]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP43]])
// CHECK2-NEXT: br label [[OMP_PRECOND_END]]
// CHECK2: omp.precond.end:
// CHECK2-NEXT: ret void
@@ -2528,7 +1968,7 @@ int bar(int n){
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK2-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP8]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -2566,7 +2006,7 @@ int bar(int n){
// CHECK2: omp.loop.exit:
// CHECK2-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4
-// CHECK2-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP20]])
+// CHECK2-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP20]])
// CHECK2-NEXT: br label [[OMP_PRECOND_END]]
// CHECK2: omp.precond.end:
// CHECK2-NEXT: ret void
@@ -2644,87 +2084,27 @@ int bar(int n){
// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
-// CHECK3-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP6]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP7]], [[TMP8]]
-// CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK3: cond.true:
-// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: br label [[COND_END:%.*]]
-// CHECK3: cond.false:
-// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: br label [[COND_END]]
-// CHECK3: cond.end:
-// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP9]], [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
-// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK3: omp.inner.for.cond:
-// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], 1
-// CHECK3-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP12]], [[ADD]]
-// CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK3: omp.inner.for.body:
-// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK3-NEXT: store i32 [[TMP16]], ptr [[N_CASTED]], align 4
-// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[N_CASTED]], align 4
-// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK3-NEXT: [[TMP19:%.*]] = inttoptr i32 [[TMP14]] to ptr
-// CHECK3-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 4
-// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK3-NEXT: [[TMP21:%.*]] = inttoptr i32 [[TMP15]] to ptr
-// CHECK3-NEXT: store ptr [[TMP21]], ptr [[TMP20]], align 4
-// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK3-NEXT: [[TMP23:%.*]] = inttoptr i32 [[TMP17]] to ptr
-// CHECK3-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 4
-// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
-// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP24]], align 4
-// CHECK3-NEXT: [[TMP25:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
-// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP26]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l28_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4, i32 0)
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK3: omp.inner.for.inc:
-// CHECK3-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP27]], [[TMP28]]
-// CHECK3-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP29]], [[TMP30]]
-// CHECK3-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
-// CHECK3-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP33]], [[TMP34]]
-// CHECK3-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]]
-// CHECK3: cond.true10:
-// CHECK3-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: br label [[COND_END12:%.*]]
-// CHECK3: cond.false11:
-// CHECK3-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: br label [[COND_END12]]
-// CHECK3: cond.end12:
-// CHECK3-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP35]], [[COND_TRUE10]] ], [ [[TMP36]], [[COND_FALSE11]] ]
-// CHECK3-NEXT: store i32 [[COND13]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: store i32 [[TMP37]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP7]], ptr [[N_CASTED]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[N_CASTED]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP10:%.*]] = inttoptr i32 [[TMP5]] to ptr
+// CHECK3-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP6]] to ptr
+// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP14:%.*]] = inttoptr i32 [[TMP8]] to ptr
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 4
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4
+// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP17]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l28_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4, i32 0)
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
-// CHECK3-NEXT: [[TMP38:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP39:%.*]] = load i32, ptr [[TMP38]], align 4
-// CHECK3-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP39]])
// CHECK3-NEXT: br label [[OMP_PRECOND_END]]
// CHECK3: omp.precond.end:
// CHECK3-NEXT: ret void
@@ -2779,7 +2159,7 @@ int bar(int n){
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP8]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -2814,7 +2194,7 @@ int bar(int n){
// CHECK3: omp.loop.exit:
// CHECK3-NEXT: [[TMP17:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP17]], align 4
-// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP18]])
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP18]])
// CHECK3-NEXT: br label [[OMP_PRECOND_END]]
// CHECK3: omp.precond.end:
// CHECK3-NEXT: ret void
@@ -2877,71 +2257,21 @@ int bar(int n){
// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
-// CHECK3-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
-// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK3: cond.true:
-// CHECK3-NEXT: br label [[COND_END:%.*]]
-// CHECK3: cond.false:
-// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: br label [[COND_END]]
-// CHECK3: cond.end:
-// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
-// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK3: omp.inner.for.cond:
-// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP6]], 10
-// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK3: omp.inner.for.body:
-// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK3-NEXT: [[TMP10:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK3-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 4
-// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK3-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP8]] to ptr
-// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4
-// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP13]], align 4
-// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l33_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3, i32 0)
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK3: omp.inner.for.inc:
-// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
-// CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK3-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
-// CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
-// CHECK3: cond.true5:
-// CHECK3-NEXT: br label [[COND_END7:%.*]]
-// CHECK3: cond.false6:
-// CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: br label [[COND_END7]]
-// CHECK3: cond.end7:
-// CHECK3-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
-// CHECK3-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = inttoptr i32 [[TMP1]] to ptr
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP3]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP6:%.*]] = inttoptr i32 [[TMP2]] to ptr
+// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
+// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP9]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l33_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3, i32 0)
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
-// CHECK3-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP2]])
// CHECK3-NEXT: ret void
//
//
@@ -2976,7 +2306,7 @@ int bar(int n){
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -3007,7 +2337,7 @@ int bar(int n){
// CHECK3: omp.inner.for.end:
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
-// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP4]])
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK3-NEXT: ret void
//
//
@@ -3070,77 +2400,27 @@ int bar(int n){
// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
-// CHECK3-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99
-// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK3: cond.true:
-// CHECK3-NEXT: br label [[COND_END:%.*]]
-// CHECK3: cond.false:
-// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: br label [[COND_END]]
-// CHECK3: cond.end:
-// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
-// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK3: omp.inner.for.cond:
-// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[CMP2:%.*]] = icmp slt i32 [[TMP6]], 100
-// CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK3: omp.inner.for.body:
-// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[F_ADDR]], align 4
-// CHECK3-NEXT: store i32 [[TMP9]], ptr [[F_CASTED]], align 4
-// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[F_CASTED]], align 4
-// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK3-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP7]] to ptr
-// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4
-// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK3-NEXT: [[TMP14:%.*]] = inttoptr i32 [[TMP8]] to ptr
-// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 4
-// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 4
-// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
-// CHECK3-NEXT: [[TMP17:%.*]] = inttoptr i32 [[TMP10]] to ptr
-// CHECK3-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 4
-// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l38_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4, i32 0)
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK3: omp.inner.for.inc:
-// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
-// CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
-// CHECK3-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
-// CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP24]], 99
-// CHECK3-NEXT: br i1 [[CMP5]], label [[COND_TRUE6:%.*]], label [[COND_FALSE7:%.*]]
-// CHECK3: cond.true6:
-// CHECK3-NEXT: br label [[COND_END8:%.*]]
-// CHECK3: cond.false7:
-// CHECK3-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: br label [[COND_END8]]
-// CHECK3: cond.end8:
-// CHECK3-NEXT: [[COND9:%.*]] = phi i32 [ 99, [[COND_TRUE6]] ], [ [[TMP25]], [[COND_FALSE7]] ]
-// CHECK3-NEXT: store i32 [[COND9]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: store i32 [[TMP26]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[F_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP3]], ptr [[F_CASTED]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[F_CASTED]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = inttoptr i32 [[TMP1]] to ptr
+// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP8:%.*]] = inttoptr i32 [[TMP2]] to ptr
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
+// CHECK3-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP4]] to ptr
+// CHECK3-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
+// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP13]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l38_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4, i32 0)
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
-// CHECK3-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP2]])
// CHECK3-NEXT: ret void
//
//
@@ -3180,7 +2460,7 @@ int bar(int n){
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
-// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP4]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -3228,7 +2508,7 @@ int bar(int n){
// CHECK3: omp.inner.for.end:
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
-// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP4]])
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK3-NEXT: ret void
//
//
@@ -3322,89 +2602,29 @@ int bar(int n){
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK3-NEXT: [[CONV11:%.*]] = zext i32 [[NVPTX_NUM_THREADS]] to i64
-// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
-// CHECK3-NEXT: call void @__kmpc_distribute_static_init_8(ptr @[[GLOB2]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 [[CONV11]])
+// CHECK3-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
+// CHECK3-NEXT: [[TMP9:%.*]] = trunc i64 [[TMP8]] to i32
// CHECK3-NEXT: [[TMP10:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK3-NEXT: [[CMP12:%.*]] = icmp sgt i64 [[TMP10]], [[TMP11]]
-// CHECK3-NEXT: br i1 [[CMP12]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK3: cond.true:
-// CHECK3-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK3-NEXT: br label [[COND_END:%.*]]
-// CHECK3: cond.false:
-// CHECK3-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: br label [[COND_END]]
-// CHECK3: cond.end:
-// CHECK3-NEXT: [[COND:%.*]] = phi i64 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
-// CHECK3-NEXT: store i64 [[COND]], ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK3-NEXT: store i64 [[TMP14]], ptr [[DOTOMP_IV]], align 8
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK3: omp.inner.for.cond:
-// CHECK3-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
-// CHECK3-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK3-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP16]], 1
-// CHECK3-NEXT: [[CMP13:%.*]] = icmp slt i64 [[TMP15]], [[ADD]]
-// CHECK3-NEXT: br i1 [[CMP13]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK3: omp.inner.for.body:
-// CHECK3-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK3-NEXT: [[TMP18:%.*]] = trunc i64 [[TMP17]] to i32
-// CHECK3-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: [[TMP20:%.*]] = trunc i64 [[TMP19]] to i32
-// CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK3-NEXT: store i32 [[TMP21]], ptr [[N_CASTED]], align 4
-// CHECK3-NEXT: [[TMP22:%.*]] = load i32, ptr [[N_CASTED]], align 4
-// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK3-NEXT: [[TMP24:%.*]] = inttoptr i32 [[TMP18]] to ptr
-// CHECK3-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 4
-// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK3-NEXT: [[TMP26:%.*]] = inttoptr i32 [[TMP20]] to ptr
-// CHECK3-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 4
-// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK3-NEXT: [[TMP28:%.*]] = inttoptr i32 [[TMP22]] to ptr
-// CHECK3-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 4
-// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
-// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP29]], align 4
-// CHECK3-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
-// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP31]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l46_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4, i32 0)
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK3: omp.inner.for.inc:
-// CHECK3-NEXT: [[TMP32:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
-// CHECK3-NEXT: [[TMP33:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
-// CHECK3-NEXT: [[ADD14:%.*]] = add nsw i64 [[TMP32]], [[TMP33]]
-// CHECK3-NEXT: store i64 [[ADD14]], ptr [[DOTOMP_IV]], align 8
-// CHECK3-NEXT: [[TMP34:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK3-NEXT: [[TMP35:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
-// CHECK3-NEXT: [[ADD15:%.*]] = add nsw i64 [[TMP34]], [[TMP35]]
-// CHECK3-NEXT: store i64 [[ADD15]], ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK3-NEXT: [[TMP36:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: [[TMP37:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
-// CHECK3-NEXT: [[ADD16:%.*]] = add nsw i64 [[TMP36]], [[TMP37]]
-// CHECK3-NEXT: store i64 [[ADD16]], ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: [[TMP38:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: [[TMP39:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK3-NEXT: [[CMP17:%.*]] = icmp sgt i64 [[TMP38]], [[TMP39]]
-// CHECK3-NEXT: br i1 [[CMP17]], label [[COND_TRUE18:%.*]], label [[COND_FALSE19:%.*]]
-// CHECK3: cond.true18:
-// CHECK3-NEXT: [[TMP40:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
-// CHECK3-NEXT: br label [[COND_END20:%.*]]
-// CHECK3: cond.false19:
-// CHECK3-NEXT: [[TMP41:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: br label [[COND_END20]]
-// CHECK3: cond.end20:
-// CHECK3-NEXT: [[COND21:%.*]] = phi i64 [ [[TMP40]], [[COND_TRUE18]] ], [ [[TMP41]], [[COND_FALSE19]] ]
-// CHECK3-NEXT: store i64 [[COND21]], ptr [[DOTOMP_COMB_UB]], align 8
-// CHECK3-NEXT: [[TMP42:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
-// CHECK3-NEXT: store i64 [[TMP42]], ptr [[DOTOMP_IV]], align 8
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: [[TMP11:%.*]] = trunc i64 [[TMP10]] to i32
+// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP12]], ptr [[N_CASTED]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[N_CASTED]], align 4
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP15:%.*]] = inttoptr i32 [[TMP9]] to ptr
+// CHECK3-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP17:%.*]] = inttoptr i32 [[TMP11]] to ptr
+// CHECK3-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP19:%.*]] = inttoptr i32 [[TMP13]] to ptr
+// CHECK3-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 4
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP20]], align 4
+// CHECK3-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
+// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP22]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l46_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4, i32 0)
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
-// CHECK3-NEXT: [[TMP43:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP44:%.*]] = load i32, ptr [[TMP43]], align 4
-// CHECK3-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP44]])
// CHECK3-NEXT: br label [[OMP_PRECOND_END]]
// CHECK3: omp.precond.end:
// CHECK3-NEXT: ret void
@@ -3478,7 +2698,7 @@ int bar(int n){
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
-// CHECK3-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB3]], i32 [[TMP11]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
+// CHECK3-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB2]], i32 [[TMP11]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK3-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK3-NEXT: store i64 [[TMP12]], ptr [[DOTOMP_IV]], align 8
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -3541,7 +2761,7 @@ int bar(int n){
// CHECK3: omp.loop.exit:
// CHECK3-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
-// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP28]])
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP28]])
// CHECK3-NEXT: br label [[OMP_PRECOND_END]]
// CHECK3: omp.precond.end:
// CHECK3-NEXT: ret void
@@ -3624,90 +2844,30 @@ int bar(int n){
// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
-// CHECK3-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP6]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
-// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP7]], [[TMP8]]
-// CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK3: cond.true:
-// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: br label [[COND_END:%.*]]
-// CHECK3: cond.false:
-// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: br label [[COND_END]]
-// CHECK3: cond.end:
-// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP9]], [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
-// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK3: omp.inner.for.cond:
-// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], 1
-// CHECK3-NEXT: [[CMP5:%.*]] = icmp slt i32 [[TMP12]], [[ADD]]
-// CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK3: omp.inner.for.body:
-// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[N_ADDR]], align 4
-// CHECK3-NEXT: store i32 [[TMP16]], ptr [[N_CASTED]], align 4
-// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[N_CASTED]], align 4
-// CHECK3-NEXT: [[TMP18:%.*]] = load ptr, ptr [[V_ADDR]], align 4
-// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
-// CHECK3-NEXT: [[TMP20:%.*]] = inttoptr i32 [[TMP14]] to ptr
-// CHECK3-NEXT: store ptr [[TMP20]], ptr [[TMP19]], align 4
-// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
-// CHECK3-NEXT: [[TMP22:%.*]] = inttoptr i32 [[TMP15]] to ptr
-// CHECK3-NEXT: store ptr [[TMP22]], ptr [[TMP21]], align 4
-// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
-// CHECK3-NEXT: [[TMP24:%.*]] = inttoptr i32 [[TMP17]] to ptr
-// CHECK3-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 4
-// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
-// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP25]], align 4
-// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 4
-// CHECK3-NEXT: store ptr [[TMP18]], ptr [[TMP26]], align 4
-// CHECK3-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
-// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l53_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 5, i32 0)
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK3: omp.inner.for.inc:
-// CHECK3-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP29]], [[TMP30]]
-// CHECK3-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
-// CHECK3-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
-// CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
-// CHECK3-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: [[CMP9:%.*]] = icmp sgt i32 [[TMP35]], [[TMP36]]
-// CHECK3-NEXT: br i1 [[CMP9]], label [[COND_TRUE10:%.*]], label [[COND_FALSE11:%.*]]
-// CHECK3: cond.true10:
-// CHECK3-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: br label [[COND_END12:%.*]]
-// CHECK3: cond.false11:
-// CHECK3-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: br label [[COND_END12]]
-// CHECK3: cond.end12:
-// CHECK3-NEXT: [[COND13:%.*]] = phi i32 [ [[TMP37]], [[COND_TRUE10]] ], [ [[TMP38]], [[COND_FALSE11]] ]
-// CHECK3-NEXT: store i32 [[COND13]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK3-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK3-NEXT: store i32 [[TMP39]], ptr [[DOTOMP_IV]], align 4
-// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[N_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP7]], ptr [[N_CASTED]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[N_CASTED]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[V_ADDR]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP11:%.*]] = inttoptr i32 [[TMP5]] to ptr
+// CHECK3-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP13:%.*]] = inttoptr i32 [[TMP6]] to ptr
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP15:%.*]] = inttoptr i32 [[TMP8]] to ptr
+// CHECK3-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP18]], align 4
+// CHECK3-NEXT: call void @__kmpc_parallel_60(ptr @[[GLOB1]], i32 [[TMP19]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l53_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 5, i32 0)
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
-// CHECK3-NEXT: [[TMP40:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK3-NEXT: [[TMP41:%.*]] = load i32, ptr [[TMP40]], align 4
-// CHECK3-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP41]])
// CHECK3-NEXT: br label [[OMP_PRECOND_END]]
// CHECK3: omp.precond.end:
// CHECK3-NEXT: ret void
@@ -3764,7 +2924,7 @@ int bar(int n){
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
-// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP8]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP8]], i32 93, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -3799,7 +2959,7 @@ int bar(int n){
// CHECK3: omp.loop.exit:
// CHECK3-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4
-// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP20]])
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP20]])
// CHECK3-NEXT: br label [[OMP_PRECOND_END]]
// CHECK3: omp.precond.end:
// CHECK3-NEXT: ret void
diff --git a/clang/test/OpenMP/nvptx_target_teams_generic_loop_generic_mode_codegen.cpp b/clang/test/OpenMP/nvptx_target_teams_generic_loop_generic_mode_codegen.cpp
index 33bbeb230e2b2..b70ad88fc2f36 100644
--- a/clang/test/OpenMP/nvptx_target_teams_generic_loop_generic_mode_codegen.cpp
+++ b/clang/test/OpenMP/nvptx_target_teams_generic_loop_generic_mode_codegen.cpp
@@ -60,7 +60,7 @@ int main(int argc, char **argv) {
// CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
-// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l24_omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]], i64 [[TMP6]]) #[[ATTR2:[0-9]+]]
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l24_omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]], i64 [[TMP6]]) #[[ATTR3:[0-9]+]]
// CHECK1-NEXT: call void @__kmpc_target_deinit()
// CHECK1-NEXT: ret void
// CHECK1: worker.exit:
@@ -108,55 +108,20 @@ int main(int argc, char **argv) {
// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP6]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
-// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
-// CHECK1-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP7]], [[TMP8]]
-// CHECK1-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK1: cond.true:
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
-// CHECK1-NEXT: br label [[COND_END:%.*]]
-// CHECK1: cond.false:
-// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: br label [[COND_END]]
-// CHECK1: cond.end:
-// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP9]], [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
-// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK1-NEXT: store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK1: omp.inner.for.cond:
-// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK1-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
-// CHECK1-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK1: omp.inner.for.body:
-// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP14]], 1
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP5]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[I4]], align 4
// CHECK1-NEXT: [[CALL:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[I4]]) #[[ATTR4:[0-9]+]]
-// CHECK1-NEXT: [[CALL7:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[TMP0]]) #[[ATTR4]]
-// CHECK1-NEXT: [[ADD8:%.*]] = add nsw i32 [[CALL]], [[CALL7]]
-// CHECK1-NEXT: [[CALL9:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[ARGC_ADDR]]) #[[ATTR4]]
-// CHECK1-NEXT: [[ADD10:%.*]] = add nsw i32 [[ADD8]], [[CALL9]]
-// CHECK1-NEXT: store i32 [[ADD10]], ptr [[TMP0]], align 4
+// CHECK1-NEXT: [[CALL5:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[TMP0]]) #[[ATTR4]]
+// CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[CALL]], [[CALL5]]
+// CHECK1-NEXT: [[CALL7:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[ARGC_ADDR]]) #[[ATTR4]]
+// CHECK1-NEXT: [[ADD8:%.*]] = add nsw i32 [[ADD6]], [[CALL7]]
+// CHECK1-NEXT: store i32 [[ADD8]], ptr [[TMP0]], align 4
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK1: omp.inner.for.inc:
-// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP15]], 1
-// CHECK1-NEXT: store i32 [[ADD11]], ptr [[DOTOMP_IV]], align 4
-// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
-// CHECK1-NEXT: [[TMP16:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
-// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4
-// CHECK1-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP17]])
// CHECK1-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1: omp.precond.end:
// CHECK1-NEXT: ret void
@@ -193,7 +158,7 @@ int main(int argc, char **argv) {
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
-// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l24_omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP4]], ptr [[TMP0]], i32 [[TMP6]]) #[[ATTR2:[0-9]+]]
+// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l24_omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP4]], ptr [[TMP0]], i32 [[TMP6]]) #[[ATTR3:[0-9]+]]
// CHECK2-NEXT: call void @__kmpc_target_deinit()
// CHECK2-NEXT: ret void
// CHECK2: worker.exit:
@@ -241,55 +206,20 @@ int main(int argc, char **argv) {
// CHECK2-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP6]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
-// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
-// CHECK2-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP7]], [[TMP8]]
-// CHECK2-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// CHECK2: cond.true:
-// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
-// CHECK2-NEXT: br label [[COND_END:%.*]]
-// CHECK2: cond.false:
-// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: br label [[COND_END]]
-// CHECK2: cond.end:
-// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ [[TMP9]], [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
-// CHECK2-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
-// CHECK2-NEXT: store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// CHECK2: omp.inner.for.cond:
-// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
-// CHECK2-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
-// CHECK2-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// CHECK2: omp.inner.for.body:
-// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP14]], 1
+// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP5]], 1
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT: store i32 [[ADD]], ptr [[I4]], align 4
// CHECK2-NEXT: [[CALL:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[I4]]) #[[ATTR4:[0-9]+]]
-// CHECK2-NEXT: [[CALL7:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[TMP0]]) #[[ATTR4]]
-// CHECK2-NEXT: [[ADD8:%.*]] = add nsw i32 [[CALL]], [[CALL7]]
-// CHECK2-NEXT: [[CALL9:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[ARGC_ADDR]]) #[[ATTR4]]
-// CHECK2-NEXT: [[ADD10:%.*]] = add nsw i32 [[ADD8]], [[CALL9]]
-// CHECK2-NEXT: store i32 [[ADD10]], ptr [[TMP0]], align 4
+// CHECK2-NEXT: [[CALL5:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[TMP0]]) #[[ATTR4]]
+// CHECK2-NEXT: [[ADD6:%.*]] = add nsw i32 [[CALL]], [[CALL5]]
+// CHECK2-NEXT: [[CALL7:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[ARGC_ADDR]]) #[[ATTR4]]
+// CHECK2-NEXT: [[ADD8:%.*]] = add nsw i32 [[ADD6]], [[CALL7]]
+// CHECK2-NEXT: store i32 [[ADD8]], ptr [[TMP0]], align 4
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue:
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// CHECK2: omp.inner.for.inc:
-// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP15]], 1
-// CHECK2-NEXT: store i32 [[ADD11]], ptr [[DOTOMP_IV]], align 4
-// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
-// CHECK2: omp.inner.for.end:
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
-// CHECK2-NEXT: [[TMP16:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
-// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4
-// CHECK2-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP17]])
// CHECK2-NEXT: br label [[OMP_PRECOND_END]]
// CHECK2: omp.precond.end:
// CHECK2-NEXT: ret void
diff --git a/clang/test/OpenMP/target_teams_generic_loop_codegen_as_distribute.cpp b/clang/test/OpenMP/target_teams_generic_loop_codegen_as_distribute.cpp
index 3686ca7ea08e0..d1ddecdbc1f63 100644
--- a/clang/test/OpenMP/target_teams_generic_loop_codegen_as_distribute.cpp
+++ b/clang/test/OpenMP/target_teams_generic_loop_codegen_as_distribute.cpp
@@ -73,7 +73,7 @@ int main()
// IR-GPU-NEXT: store i32 0, ptr [[DOTZERO_ADDR_ASCAST]], align 4
// IR-GPU-NEXT: store i32 [[TMP5]], ptr addrspace(5) [[DOTTHREADID_TEMP_]], align 4
// IR-GPU-NEXT: [[TMP8:%.*]] = addrspacecast ptr addrspace(5) [[DOTTHREADID_TEMP_]] to ptr
-// IR-GPU-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l27_omp_outlined(ptr [[TMP8]], ptr [[DOTZERO_ADDR_ASCAST]], i64 [[TMP7]], i64 [[TMP0]], ptr [[TMP1]], i64 [[TMP2]], ptr [[TMP3]]) #[[ATTR2:[0-9]+]]
+// IR-GPU-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l27_omp_outlined(ptr [[TMP8]], ptr [[DOTZERO_ADDR_ASCAST]], i64 [[TMP7]], i64 [[TMP0]], ptr [[TMP1]], i64 [[TMP2]], ptr [[TMP3]]) #[[ATTR3:[0-9]+]]
// IR-GPU-NEXT: call void @__kmpc_target_deinit()
// IR-GPU-NEXT: ret void
// IR-GPU: worker.exit:
@@ -147,77 +147,42 @@ int main()
// IR-GPU-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
// IR-GPU-NEXT: store i32 1, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
// IR-GPU-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
-// IR-GPU-NEXT: call void @__kmpc_distribute_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB2:[0-9]+]] to ptr), i32 [[TMP9]], i32 92, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_COMB_LB_ASCAST]], ptr [[DOTOMP_COMB_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
-// IR-GPU-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3_ASCAST]], align 4
-// IR-GPU-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]]
-// IR-GPU-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// IR-GPU: cond.true:
-// IR-GPU-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3_ASCAST]], align 4
-// IR-GPU-NEXT: br label [[COND_END:%.*]]
-// IR-GPU: cond.false:
-// IR-GPU-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: br label [[COND_END]]
-// IR-GPU: cond.end:
-// IR-GPU-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
-// IR-GPU-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// IR-GPU-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV_ASCAST]], align 4
-// IR-GPU-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// IR-GPU: omp.inner.for.cond:
-// IR-GPU-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]]
-// IR-GPU-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// IR-GPU: omp.inner.for.body:
-// IR-GPU-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4
-// IR-GPU-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1
+// IR-GPU-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4
+// IR-GPU-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// IR-GPU-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// IR-GPU-NEXT: store i32 [[ADD]], ptr [[I5_ASCAST]], align 4
// IR-GPU-NEXT: store i32 0, ptr [[J_ASCAST]], align 4
// IR-GPU-NEXT: br label [[FOR_COND:%.*]]
// IR-GPU: for.cond:
-// IR-GPU-NEXT: [[TMP18:%.*]] = load i32, ptr [[J_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP19:%.*]] = load i32, ptr [[N_ADDR_ASCAST]], align 4
-// IR-GPU-NEXT: [[CMP8:%.*]] = icmp slt i32 [[TMP18]], [[TMP19]]
-// IR-GPU-NEXT: br i1 [[CMP8]], label [[FOR_BODY:%.*]], label [[FOR_END:%.*]]
+// IR-GPU-NEXT: [[TMP9:%.*]] = load i32, ptr [[J_ASCAST]], align 4
+// IR-GPU-NEXT: [[TMP10:%.*]] = load i32, ptr [[N_ADDR_ASCAST]], align 4
+// IR-GPU-NEXT: [[CMP6:%.*]] = icmp slt i32 [[TMP9]], [[TMP10]]
+// IR-GPU-NEXT: br i1 [[CMP6]], label [[FOR_BODY:%.*]], label [[FOR_END:%.*]]
// IR-GPU: for.body:
-// IR-GPU-NEXT: [[TMP20:%.*]] = load i32, ptr [[I5_ASCAST]], align 4
-// IR-GPU-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP20]] to i64
+// IR-GPU-NEXT: [[TMP11:%.*]] = load i32, ptr [[I5_ASCAST]], align 4
+// IR-GPU-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
// IR-GPU-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP3]], i64 [[IDXPROM]]
-// IR-GPU-NEXT: [[TMP21:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-// IR-GPU-NEXT: [[TMP22:%.*]] = load i32, ptr [[N_ADDR_ASCAST]], align 4
-// IR-GPU-NEXT: [[MUL9:%.*]] = mul nsw i32 [[TMP21]], [[TMP22]]
-// IR-GPU-NEXT: [[TMP23:%.*]] = load i32, ptr [[J_ASCAST]], align 4
-// IR-GPU-NEXT: [[CALL:%.*]] = call noundef i32 @_Z3fooi(i32 noundef [[TMP23]]) #[[ATTR4:[0-9]+]]
-// IR-GPU-NEXT: [[ADD10:%.*]] = add nsw i32 [[MUL9]], [[CALL]]
-// IR-GPU-NEXT: [[TMP24:%.*]] = load i32, ptr [[I5_ASCAST]], align 4
-// IR-GPU-NEXT: [[IDXPROM11:%.*]] = sext i32 [[TMP24]] to i64
-// IR-GPU-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 [[IDXPROM11]]
-// IR-GPU-NEXT: store i32 [[ADD10]], ptr [[ARRAYIDX12]], align 4
+// IR-GPU-NEXT: [[TMP12:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
+// IR-GPU-NEXT: [[TMP13:%.*]] = load i32, ptr [[N_ADDR_ASCAST]], align 4
+// IR-GPU-NEXT: [[MUL7:%.*]] = mul nsw i32 [[TMP12]], [[TMP13]]
+// IR-GPU-NEXT: [[TMP14:%.*]] = load i32, ptr [[J_ASCAST]], align 4
+// IR-GPU-NEXT: [[CALL:%.*]] = call noundef i32 @_Z3fooi(i32 noundef [[TMP14]]) #[[ATTR4:[0-9]+]]
+// IR-GPU-NEXT: [[ADD8:%.*]] = add nsw i32 [[MUL7]], [[CALL]]
+// IR-GPU-NEXT: [[TMP15:%.*]] = load i32, ptr [[I5_ASCAST]], align 4
+// IR-GPU-NEXT: [[IDXPROM9:%.*]] = sext i32 [[TMP15]] to i64
+// IR-GPU-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 [[IDXPROM9]]
+// IR-GPU-NEXT: store i32 [[ADD8]], ptr [[ARRAYIDX10]], align 4
// IR-GPU-NEXT: br label [[FOR_INC:%.*]]
// IR-GPU: for.inc:
-// IR-GPU-NEXT: [[TMP25:%.*]] = load i32, ptr [[J_ASCAST]], align 4
-// IR-GPU-NEXT: [[INC:%.*]] = add nsw i32 [[TMP25]], 1
+// IR-GPU-NEXT: [[TMP16:%.*]] = load i32, ptr [[J_ASCAST]], align 4
+// IR-GPU-NEXT: [[INC:%.*]] = add nsw i32 [[TMP16]], 1
// IR-GPU-NEXT: store i32 [[INC]], ptr [[J_ASCAST]], align 4
// IR-GPU-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// IR-GPU: for.end:
// IR-GPU-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// IR-GPU: omp.body.continue:
-// IR-GPU-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// IR-GPU: omp.inner.for.inc:
-// IR-GPU-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4
-// IR-GPU-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP26]], 1
-// IR-GPU-NEXT: store i32 [[ADD13]], ptr [[DOTOMP_IV_ASCAST]], align 4
-// IR-GPU-NEXT: br label [[OMP_INNER_FOR_COND]]
-// IR-GPU: omp.inner.for.end:
// IR-GPU-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// IR-GPU: omp.loop.exit:
-// IR-GPU-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
-// IR-GPU-NEXT: call void @__kmpc_distribute_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP28]])
// IR-GPU-NEXT: br label [[OMP_PRECOND_END]]
// IR-GPU: omp.precond.end:
// IR-GPU-NEXT: ret void
diff --git a/clang/test/OpenMP/target_teams_generic_loop_codegen_as_parallel_for.cpp b/clang/test/OpenMP/target_teams_generic_loop_codegen_as_parallel_for.cpp
index 11ae386739b40..2110e67c162e4 100644
--- a/clang/test/OpenMP/target_teams_generic_loop_codegen_as_parallel_for.cpp
+++ b/clang/test/OpenMP/target_teams_generic_loop_codegen_as_parallel_for.cpp
@@ -187,97 +187,37 @@ int main()
// IR-GPU-NEXT: store i32 1, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
// IR-GPU-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
// IR-GPU-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// IR-GPU-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
-// IR-GPU-NEXT: call void @__kmpc_distribute_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB2:[0-9]+]] to ptr), i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_COMB_LB_ASCAST]], ptr [[DOTOMP_COMB_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// IR-GPU-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
+// IR-GPU-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// IR-GPU-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3_ASCAST]], align 4
-// IR-GPU-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]]
-// IR-GPU-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// IR-GPU: cond.true:
-// IR-GPU-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3_ASCAST]], align 4
-// IR-GPU-NEXT: br label [[COND_END:%.*]]
-// IR-GPU: cond.false:
-// IR-GPU-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: br label [[COND_END]]
-// IR-GPU: cond.end:
-// IR-GPU-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
-// IR-GPU-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// IR-GPU-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV_ASCAST]], align 4
-// IR-GPU-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// IR-GPU: omp.inner.for.cond:
-// IR-GPU-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3_ASCAST]], align 4
-// IR-GPU-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], 1
-// IR-GPU-NEXT: [[CMP7:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
-// IR-GPU-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// IR-GPU: omp.inner.for.body:
-// IR-GPU-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64
-// IR-GPU-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64
-// IR-GPU-NEXT: [[TMP21:%.*]] = load i32, ptr [[N_ADDR_ASCAST]], align 4
-// IR-GPU-NEXT: store i32 [[TMP21]], ptr addrspace(5) [[N_CASTED]], align 4
-// IR-GPU-NEXT: [[TMP22:%.*]] = load i64, ptr addrspace(5) [[N_CASTED]], align 8
-// IR-GPU-NEXT: [[TMP23:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 0
-// IR-GPU-NEXT: [[TMP24:%.*]] = inttoptr i64 [[TMP18]] to ptr
+// IR-GPU-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
+// IR-GPU-NEXT: [[TMP12:%.*]] = load i32, ptr [[N_ADDR_ASCAST]], align 4
+// IR-GPU-NEXT: store i32 [[TMP12]], ptr addrspace(5) [[N_CASTED]], align 4
+// IR-GPU-NEXT: [[TMP13:%.*]] = load i64, ptr addrspace(5) [[N_CASTED]], align 8
+// IR-GPU-NEXT: [[TMP14:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 0
+// IR-GPU-NEXT: [[TMP15:%.*]] = inttoptr i64 [[TMP9]] to ptr
+// IR-GPU-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8
+// IR-GPU-NEXT: [[TMP16:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 1
+// IR-GPU-NEXT: [[TMP17:%.*]] = inttoptr i64 [[TMP11]] to ptr
+// IR-GPU-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8
+// IR-GPU-NEXT: [[TMP18:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 2
+// IR-GPU-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP13]] to ptr
+// IR-GPU-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8
+// IR-GPU-NEXT: [[TMP20:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 3
+// IR-GPU-NEXT: [[TMP21:%.*]] = inttoptr i64 [[TMP0]] to ptr
+// IR-GPU-NEXT: store ptr [[TMP21]], ptr [[TMP20]], align 8
+// IR-GPU-NEXT: [[TMP22:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 4
+// IR-GPU-NEXT: store ptr [[TMP1]], ptr [[TMP22]], align 8
+// IR-GPU-NEXT: [[TMP23:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 5
+// IR-GPU-NEXT: [[TMP24:%.*]] = inttoptr i64 [[TMP2]] to ptr
// IR-GPU-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 8
-// IR-GPU-NEXT: [[TMP25:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 1
-// IR-GPU-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP20]] to ptr
-// IR-GPU-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8
-// IR-GPU-NEXT: [[TMP27:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 2
-// IR-GPU-NEXT: [[TMP28:%.*]] = inttoptr i64 [[TMP22]] to ptr
-// IR-GPU-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 8
-// IR-GPU-NEXT: [[TMP29:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 3
-// IR-GPU-NEXT: [[TMP30:%.*]] = inttoptr i64 [[TMP0]] to ptr
-// IR-GPU-NEXT: store ptr [[TMP30]], ptr [[TMP29]], align 8
-// IR-GPU-NEXT: [[TMP31:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 4
-// IR-GPU-NEXT: store ptr [[TMP1]], ptr [[TMP31]], align 8
-// IR-GPU-NEXT: [[TMP32:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 5
-// IR-GPU-NEXT: [[TMP33:%.*]] = inttoptr i64 [[TMP2]] to ptr
-// IR-GPU-NEXT: store ptr [[TMP33]], ptr [[TMP32]], align 8
-// IR-GPU-NEXT: [[TMP34:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 6
-// IR-GPU-NEXT: store ptr [[TMP3]], ptr [[TMP34]], align 8
-// IR-GPU-NEXT: [[TMP35:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
-// IR-GPU-NEXT: call void @__kmpc_parallel_60(ptr addrspacecast (ptr addrspace(1) @[[GLOB1]] to ptr), i32 [[TMP36]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l41_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 7, i32 0)
-// IR-GPU-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// IR-GPU: omp.inner.for.inc:
-// IR-GPU-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
-// IR-GPU-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP37]], [[TMP38]]
-// IR-GPU-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
-// IR-GPU-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP39]], [[TMP40]]
-// IR-GPU-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP42:%.*]] = load i32, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
-// IR-GPU-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP41]], [[TMP42]]
-// IR-GPU-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP43:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP44:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3_ASCAST]], align 4
-// IR-GPU-NEXT: [[CMP11:%.*]] = icmp sgt i32 [[TMP43]], [[TMP44]]
-// IR-GPU-NEXT: br i1 [[CMP11]], label [[COND_TRUE12:%.*]], label [[COND_FALSE13:%.*]]
-// IR-GPU: cond.true12:
-// IR-GPU-NEXT: [[TMP45:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3_ASCAST]], align 4
-// IR-GPU-NEXT: br label [[COND_END14:%.*]]
-// IR-GPU: cond.false13:
-// IR-GPU-NEXT: [[TMP46:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: br label [[COND_END14]]
-// IR-GPU: cond.end14:
-// IR-GPU-NEXT: [[COND15:%.*]] = phi i32 [ [[TMP45]], [[COND_TRUE12]] ], [ [[TMP46]], [[COND_FALSE13]] ]
-// IR-GPU-NEXT: store i32 [[COND15]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP47:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// IR-GPU-NEXT: store i32 [[TMP47]], ptr [[DOTOMP_IV_ASCAST]], align 4
-// IR-GPU-NEXT: br label [[OMP_INNER_FOR_COND]]
-// IR-GPU: omp.inner.for.end:
+// IR-GPU-NEXT: [[TMP25:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 6
+// IR-GPU-NEXT: store ptr [[TMP3]], ptr [[TMP25]], align 8
+// IR-GPU-NEXT: [[TMP26:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
+// IR-GPU-NEXT: [[TMP27:%.*]] = load i32, ptr [[TMP26]], align 4
+// IR-GPU-NEXT: call void @__kmpc_parallel_60(ptr addrspacecast (ptr addrspace(1) @[[GLOB1]] to ptr), i32 [[TMP27]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l41_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 7, i32 0)
// IR-GPU-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// IR-GPU: omp.loop.exit:
-// IR-GPU-NEXT: [[TMP48:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP49:%.*]] = load i32, ptr [[TMP48]], align 4
-// IR-GPU-NEXT: call void @__kmpc_distribute_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP49]])
// IR-GPU-NEXT: br label [[OMP_PRECOND_END]]
// IR-GPU: omp.precond.end:
// IR-GPU-NEXT: ret void
@@ -362,7 +302,7 @@ int main()
// IR-GPU-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
// IR-GPU-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// IR-GPU-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
-// IR-GPU-NEXT: call void @__kmpc_for_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB3:[0-9]+]] to ptr), i32 [[TMP11]], i32 33, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
+// IR-GPU-NEXT: call void @__kmpc_for_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB2:[0-9]+]] to ptr), i32 [[TMP11]], i32 93, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
// IR-GPU-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_LB_ASCAST]], align 4
// IR-GPU-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV_ASCAST]], align 4
// IR-GPU-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -399,7 +339,7 @@ int main()
// IR-GPU: omp.loop.exit:
// IR-GPU-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// IR-GPU-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
-// IR-GPU-NEXT: call void @__kmpc_for_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB3]] to ptr), i32 [[TMP22]])
+// IR-GPU-NEXT: call void @__kmpc_for_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP22]])
// IR-GPU-NEXT: br label [[OMP_PRECOND_END]]
// IR-GPU: omp.precond.end:
// IR-GPU-NEXT: ret void
@@ -543,95 +483,35 @@ int main()
// IR-GPU-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
// IR-GPU-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// IR-GPU-NEXT: [[CONV13:%.*]] = zext i32 [[NVPTX_NUM_THREADS]] to i64
-// IR-GPU-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
-// IR-GPU-NEXT: call void @__kmpc_distribute_static_init_8(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP12]], i32 91, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_COMB_LB_ASCAST]], ptr [[DOTOMP_COMB_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i64 1, i64 [[CONV13]])
-// IR-GPU-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5_ASCAST]], align 8
-// IR-GPU-NEXT: [[CMP14:%.*]] = icmp sgt i64 [[TMP13]], [[TMP14]]
-// IR-GPU-NEXT: br i1 [[CMP14]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// IR-GPU: cond.true:
-// IR-GPU-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5_ASCAST]], align 8
-// IR-GPU-NEXT: br label [[COND_END:%.*]]
-// IR-GPU: cond.false:
-// IR-GPU-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NEXT: br label [[COND_END]]
-// IR-GPU: cond.end:
-// IR-GPU-NEXT: [[COND:%.*]] = phi i64 [ [[TMP15]], [[COND_TRUE]] ], [ [[TMP16]], [[COND_FALSE]] ]
-// IR-GPU-NEXT: store i64 [[COND]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_COMB_LB_ASCAST]], align 8
-// IR-GPU-NEXT: store i64 [[TMP17]], ptr [[DOTOMP_IV_ASCAST]], align 8
-// IR-GPU-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// IR-GPU: omp.inner.for.cond:
-// IR-GPU-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_IV_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5_ASCAST]], align 8
-// IR-GPU-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP19]], 1
-// IR-GPU-NEXT: [[CMP15:%.*]] = icmp slt i64 [[TMP18]], [[ADD]]
-// IR-GPU-NEXT: br i1 [[CMP15]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// IR-GPU: omp.inner.for.body:
-// IR-GPU-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTOMP_COMB_LB_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP22:%.*]] = load i32, ptr [[N_ADDR_ASCAST]], align 4
-// IR-GPU-NEXT: store i32 [[TMP22]], ptr addrspace(5) [[N_CASTED]], align 4
-// IR-GPU-NEXT: [[TMP23:%.*]] = load i64, ptr addrspace(5) [[N_CASTED]], align 8
-// IR-GPU-NEXT: [[TMP24:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 0
-// IR-GPU-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP20]] to ptr
+// IR-GPU-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTOMP_COMB_LB_ASCAST]], align 8
+// IR-GPU-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
+// IR-GPU-NEXT: [[TMP13:%.*]] = load i32, ptr [[N_ADDR_ASCAST]], align 4
+// IR-GPU-NEXT: store i32 [[TMP13]], ptr addrspace(5) [[N_CASTED]], align 4
+// IR-GPU-NEXT: [[TMP14:%.*]] = load i64, ptr addrspace(5) [[N_CASTED]], align 8
+// IR-GPU-NEXT: [[TMP15:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 0
+// IR-GPU-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP11]] to ptr
+// IR-GPU-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8
+// IR-GPU-NEXT: [[TMP17:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 1
+// IR-GPU-NEXT: [[TMP18:%.*]] = inttoptr i64 [[TMP12]] to ptr
+// IR-GPU-NEXT: store ptr [[TMP18]], ptr [[TMP17]], align 8
+// IR-GPU-NEXT: [[TMP19:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 2
+// IR-GPU-NEXT: [[TMP20:%.*]] = inttoptr i64 [[TMP14]] to ptr
+// IR-GPU-NEXT: store ptr [[TMP20]], ptr [[TMP19]], align 8
+// IR-GPU-NEXT: [[TMP21:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 3
+// IR-GPU-NEXT: [[TMP22:%.*]] = inttoptr i64 [[TMP0]] to ptr
+// IR-GPU-NEXT: store ptr [[TMP22]], ptr [[TMP21]], align 8
+// IR-GPU-NEXT: [[TMP23:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 4
+// IR-GPU-NEXT: store ptr [[TMP1]], ptr [[TMP23]], align 8
+// IR-GPU-NEXT: [[TMP24:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 5
+// IR-GPU-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP2]] to ptr
// IR-GPU-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 8
-// IR-GPU-NEXT: [[TMP26:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 1
-// IR-GPU-NEXT: [[TMP27:%.*]] = inttoptr i64 [[TMP21]] to ptr
-// IR-GPU-NEXT: store ptr [[TMP27]], ptr [[TMP26]], align 8
-// IR-GPU-NEXT: [[TMP28:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 2
-// IR-GPU-NEXT: [[TMP29:%.*]] = inttoptr i64 [[TMP23]] to ptr
-// IR-GPU-NEXT: store ptr [[TMP29]], ptr [[TMP28]], align 8
-// IR-GPU-NEXT: [[TMP30:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 3
-// IR-GPU-NEXT: [[TMP31:%.*]] = inttoptr i64 [[TMP0]] to ptr
-// IR-GPU-NEXT: store ptr [[TMP31]], ptr [[TMP30]], align 8
-// IR-GPU-NEXT: [[TMP32:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 4
-// IR-GPU-NEXT: store ptr [[TMP1]], ptr [[TMP32]], align 8
-// IR-GPU-NEXT: [[TMP33:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 5
-// IR-GPU-NEXT: [[TMP34:%.*]] = inttoptr i64 [[TMP2]] to ptr
-// IR-GPU-NEXT: store ptr [[TMP34]], ptr [[TMP33]], align 8
-// IR-GPU-NEXT: [[TMP35:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 6
-// IR-GPU-NEXT: store ptr [[TMP3]], ptr [[TMP35]], align 8
-// IR-GPU-NEXT: [[TMP36:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP37:%.*]] = load i32, ptr [[TMP36]], align 4
-// IR-GPU-NEXT: call void @__kmpc_parallel_60(ptr addrspacecast (ptr addrspace(1) @[[GLOB1]] to ptr), i32 [[TMP37]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l46_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 7, i32 0)
-// IR-GPU-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// IR-GPU: omp.inner.for.inc:
-// IR-GPU-NEXT: [[TMP38:%.*]] = load i64, ptr [[DOTOMP_IV_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP39:%.*]] = load i64, ptr [[DOTOMP_STRIDE_ASCAST]], align 8
-// IR-GPU-NEXT: [[ADD16:%.*]] = add nsw i64 [[TMP38]], [[TMP39]]
-// IR-GPU-NEXT: store i64 [[ADD16]], ptr [[DOTOMP_IV_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP40:%.*]] = load i64, ptr [[DOTOMP_COMB_LB_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP41:%.*]] = load i64, ptr [[DOTOMP_STRIDE_ASCAST]], align 8
-// IR-GPU-NEXT: [[ADD17:%.*]] = add nsw i64 [[TMP40]], [[TMP41]]
-// IR-GPU-NEXT: store i64 [[ADD17]], ptr [[DOTOMP_COMB_LB_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP42:%.*]] = load i64, ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP43:%.*]] = load i64, ptr [[DOTOMP_STRIDE_ASCAST]], align 8
-// IR-GPU-NEXT: [[ADD18:%.*]] = add nsw i64 [[TMP42]], [[TMP43]]
-// IR-GPU-NEXT: store i64 [[ADD18]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP44:%.*]] = load i64, ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP45:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5_ASCAST]], align 8
-// IR-GPU-NEXT: [[CMP19:%.*]] = icmp sgt i64 [[TMP44]], [[TMP45]]
-// IR-GPU-NEXT: br i1 [[CMP19]], label [[COND_TRUE20:%.*]], label [[COND_FALSE21:%.*]]
-// IR-GPU: cond.true20:
-// IR-GPU-NEXT: [[TMP46:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5_ASCAST]], align 8
-// IR-GPU-NEXT: br label [[COND_END22:%.*]]
-// IR-GPU: cond.false21:
-// IR-GPU-NEXT: [[TMP47:%.*]] = load i64, ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NEXT: br label [[COND_END22]]
-// IR-GPU: cond.end22:
-// IR-GPU-NEXT: [[COND23:%.*]] = phi i64 [ [[TMP46]], [[COND_TRUE20]] ], [ [[TMP47]], [[COND_FALSE21]] ]
-// IR-GPU-NEXT: store i64 [[COND23]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP48:%.*]] = load i64, ptr [[DOTOMP_COMB_LB_ASCAST]], align 8
-// IR-GPU-NEXT: store i64 [[TMP48]], ptr [[DOTOMP_IV_ASCAST]], align 8
-// IR-GPU-NEXT: br label [[OMP_INNER_FOR_COND]]
-// IR-GPU: omp.inner.for.end:
+// IR-GPU-NEXT: [[TMP26:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 6
+// IR-GPU-NEXT: store ptr [[TMP3]], ptr [[TMP26]], align 8
+// IR-GPU-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
+// IR-GPU-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
+// IR-GPU-NEXT: call void @__kmpc_parallel_60(ptr addrspacecast (ptr addrspace(1) @[[GLOB1]] to ptr), i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l46_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 7, i32 0)
// IR-GPU-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// IR-GPU: omp.loop.exit:
-// IR-GPU-NEXT: [[TMP49:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP50:%.*]] = load i32, ptr [[TMP49]], align 4
-// IR-GPU-NEXT: call void @__kmpc_distribute_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP50]])
// IR-GPU-NEXT: br label [[OMP_PRECOND_END]]
// IR-GPU: omp.precond.end:
// IR-GPU-NEXT: ret void
@@ -735,7 +615,7 @@ int main()
// IR-GPU-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
// IR-GPU-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// IR-GPU-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP13]], align 4
-// IR-GPU-NEXT: call void @__kmpc_for_static_init_8(ptr addrspacecast (ptr addrspace(1) @[[GLOB3]] to ptr), i32 [[TMP14]], i32 33, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i64 1, i64 1)
+// IR-GPU-NEXT: call void @__kmpc_for_static_init_8(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP14]], i32 93, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i64 1, i64 1)
// IR-GPU-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_LB_ASCAST]], align 8
// IR-GPU-NEXT: store i64 [[TMP15]], ptr [[DOTOMP_IV_ASCAST]], align 8
// IR-GPU-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -801,7 +681,7 @@ int main()
// IR-GPU: omp.loop.exit:
// IR-GPU-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// IR-GPU-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
-// IR-GPU-NEXT: call void @__kmpc_for_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB3]] to ptr), i32 [[TMP32]])
+// IR-GPU-NEXT: call void @__kmpc_for_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP32]])
// IR-GPU-NEXT: br label [[OMP_PRECOND_END]]
// IR-GPU: omp.precond.end:
// IR-GPU-NEXT: ret void
@@ -934,103 +814,43 @@ int main()
// IR-GPU-NEXT: store i32 1, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
// IR-GPU-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
// IR-GPU-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
-// IR-GPU-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
-// IR-GPU-NEXT: call void @__kmpc_distribute_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_COMB_LB_ASCAST]], ptr [[DOTOMP_COMB_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 [[NVPTX_NUM_THREADS]])
+// IR-GPU-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
+// IR-GPU-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// IR-GPU-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3_ASCAST]], align 4
-// IR-GPU-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]]
-// IR-GPU-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// IR-GPU: cond.true:
-// IR-GPU-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3_ASCAST]], align 4
-// IR-GPU-NEXT: br label [[COND_END:%.*]]
-// IR-GPU: cond.false:
-// IR-GPU-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: br label [[COND_END]]
-// IR-GPU: cond.end:
-// IR-GPU-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
-// IR-GPU-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// IR-GPU-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV_ASCAST]], align 4
-// IR-GPU-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// IR-GPU: omp.inner.for.cond:
-// IR-GPU-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3_ASCAST]], align 4
-// IR-GPU-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], 1
-// IR-GPU-NEXT: [[CMP7:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
-// IR-GPU-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// IR-GPU: omp.inner.for.body:
-// IR-GPU-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64
-// IR-GPU-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64
-// IR-GPU-NEXT: [[TMP21:%.*]] = load i32, ptr [[N_ADDR_ASCAST]], align 4
-// IR-GPU-NEXT: store i32 [[TMP21]], ptr addrspace(5) [[N_CASTED]], align 4
-// IR-GPU-NEXT: [[TMP22:%.*]] = load i64, ptr addrspace(5) [[N_CASTED]], align 8
-// IR-GPU-NEXT: [[TMP23:%.*]] = load i32, ptr [[NT_ADDR_ASCAST]], align 4
-// IR-GPU-NEXT: store i32 [[TMP23]], ptr addrspace(5) [[NT_CASTED]], align 4
-// IR-GPU-NEXT: [[TMP24:%.*]] = load i64, ptr addrspace(5) [[NT_CASTED]], align 8
-// IR-GPU-NEXT: [[TMP25:%.*]] = getelementptr inbounds [8 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 0
-// IR-GPU-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP18]] to ptr
-// IR-GPU-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8
-// IR-GPU-NEXT: [[TMP27:%.*]] = getelementptr inbounds [8 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 1
-// IR-GPU-NEXT: [[TMP28:%.*]] = inttoptr i64 [[TMP20]] to ptr
+// IR-GPU-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
+// IR-GPU-NEXT: [[TMP12:%.*]] = load i32, ptr [[N_ADDR_ASCAST]], align 4
+// IR-GPU-NEXT: store i32 [[TMP12]], ptr addrspace(5) [[N_CASTED]], align 4
+// IR-GPU-NEXT: [[TMP13:%.*]] = load i64, ptr addrspace(5) [[N_CASTED]], align 8
+// IR-GPU-NEXT: [[TMP14:%.*]] = load i32, ptr [[NT_ADDR_ASCAST]], align 4
+// IR-GPU-NEXT: store i32 [[TMP14]], ptr addrspace(5) [[NT_CASTED]], align 4
+// IR-GPU-NEXT: [[TMP15:%.*]] = load i64, ptr addrspace(5) [[NT_CASTED]], align 8
+// IR-GPU-NEXT: [[TMP16:%.*]] = getelementptr inbounds [8 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 0
+// IR-GPU-NEXT: [[TMP17:%.*]] = inttoptr i64 [[TMP9]] to ptr
+// IR-GPU-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8
+// IR-GPU-NEXT: [[TMP18:%.*]] = getelementptr inbounds [8 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 1
+// IR-GPU-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP11]] to ptr
+// IR-GPU-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8
+// IR-GPU-NEXT: [[TMP20:%.*]] = getelementptr inbounds [8 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 2
+// IR-GPU-NEXT: [[TMP21:%.*]] = inttoptr i64 [[TMP13]] to ptr
+// IR-GPU-NEXT: store ptr [[TMP21]], ptr [[TMP20]], align 8
+// IR-GPU-NEXT: [[TMP22:%.*]] = getelementptr inbounds [8 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 3
+// IR-GPU-NEXT: [[TMP23:%.*]] = inttoptr i64 [[TMP15]] to ptr
+// IR-GPU-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 8
+// IR-GPU-NEXT: [[TMP24:%.*]] = getelementptr inbounds [8 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 4
+// IR-GPU-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP0]] to ptr
+// IR-GPU-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 8
+// IR-GPU-NEXT: [[TMP26:%.*]] = getelementptr inbounds [8 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 5
+// IR-GPU-NEXT: store ptr [[TMP1]], ptr [[TMP26]], align 8
+// IR-GPU-NEXT: [[TMP27:%.*]] = getelementptr inbounds [8 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 6
+// IR-GPU-NEXT: [[TMP28:%.*]] = inttoptr i64 [[TMP2]] to ptr
// IR-GPU-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 8
-// IR-GPU-NEXT: [[TMP29:%.*]] = getelementptr inbounds [8 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 2
-// IR-GPU-NEXT: [[TMP30:%.*]] = inttoptr i64 [[TMP22]] to ptr
-// IR-GPU-NEXT: store ptr [[TMP30]], ptr [[TMP29]], align 8
-// IR-GPU-NEXT: [[TMP31:%.*]] = getelementptr inbounds [8 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 3
-// IR-GPU-NEXT: [[TMP32:%.*]] = inttoptr i64 [[TMP24]] to ptr
-// IR-GPU-NEXT: store ptr [[TMP32]], ptr [[TMP31]], align 8
-// IR-GPU-NEXT: [[TMP33:%.*]] = getelementptr inbounds [8 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 4
-// IR-GPU-NEXT: [[TMP34:%.*]] = inttoptr i64 [[TMP0]] to ptr
-// IR-GPU-NEXT: store ptr [[TMP34]], ptr [[TMP33]], align 8
-// IR-GPU-NEXT: [[TMP35:%.*]] = getelementptr inbounds [8 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 5
-// IR-GPU-NEXT: store ptr [[TMP1]], ptr [[TMP35]], align 8
-// IR-GPU-NEXT: [[TMP36:%.*]] = getelementptr inbounds [8 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 6
-// IR-GPU-NEXT: [[TMP37:%.*]] = inttoptr i64 [[TMP2]] to ptr
-// IR-GPU-NEXT: store ptr [[TMP37]], ptr [[TMP36]], align 8
-// IR-GPU-NEXT: [[TMP38:%.*]] = getelementptr inbounds [8 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 7
-// IR-GPU-NEXT: store ptr [[TMP3]], ptr [[TMP38]], align 8
-// IR-GPU-NEXT: [[TMP39:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP40:%.*]] = load i32, ptr [[TMP39]], align 4
-// IR-GPU-NEXT: call void @__kmpc_parallel_60(ptr addrspacecast (ptr addrspace(1) @[[GLOB1]] to ptr), i32 [[TMP40]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l55_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 8, i32 0)
-// IR-GPU-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// IR-GPU: omp.inner.for.inc:
-// IR-GPU-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP42:%.*]] = load i32, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
-// IR-GPU-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP41]], [[TMP42]]
-// IR-GPU-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP43:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP44:%.*]] = load i32, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
-// IR-GPU-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP43]], [[TMP44]]
-// IR-GPU-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP45:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP46:%.*]] = load i32, ptr [[DOTOMP_STRIDE_ASCAST]], align 4
-// IR-GPU-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP45]], [[TMP46]]
-// IR-GPU-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP47:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP48:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3_ASCAST]], align 4
-// IR-GPU-NEXT: [[CMP11:%.*]] = icmp sgt i32 [[TMP47]], [[TMP48]]
-// IR-GPU-NEXT: br i1 [[CMP11]], label [[COND_TRUE12:%.*]], label [[COND_FALSE13:%.*]]
-// IR-GPU: cond.true12:
-// IR-GPU-NEXT: [[TMP49:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3_ASCAST]], align 4
-// IR-GPU-NEXT: br label [[COND_END14:%.*]]
-// IR-GPU: cond.false13:
-// IR-GPU-NEXT: [[TMP50:%.*]] = load i32, ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: br label [[COND_END14]]
-// IR-GPU: cond.end14:
-// IR-GPU-NEXT: [[COND15:%.*]] = phi i32 [ [[TMP49]], [[COND_TRUE12]] ], [ [[TMP50]], [[COND_FALSE13]] ]
-// IR-GPU-NEXT: store i32 [[COND15]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 4
-// IR-GPU-NEXT: [[TMP51:%.*]] = load i32, ptr [[DOTOMP_COMB_LB_ASCAST]], align 4
-// IR-GPU-NEXT: store i32 [[TMP51]], ptr [[DOTOMP_IV_ASCAST]], align 4
-// IR-GPU-NEXT: br label [[OMP_INNER_FOR_COND]]
-// IR-GPU: omp.inner.for.end:
+// IR-GPU-NEXT: [[TMP29:%.*]] = getelementptr inbounds [8 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 7
+// IR-GPU-NEXT: store ptr [[TMP3]], ptr [[TMP29]], align 8
+// IR-GPU-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
+// IR-GPU-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
+// IR-GPU-NEXT: call void @__kmpc_parallel_60(ptr addrspacecast (ptr addrspace(1) @[[GLOB1]] to ptr), i32 [[TMP31]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l55_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 8, i32 0)
// IR-GPU-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// IR-GPU: omp.loop.exit:
-// IR-GPU-NEXT: [[TMP52:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// IR-GPU-NEXT: [[TMP53:%.*]] = load i32, ptr [[TMP52]], align 4
-// IR-GPU-NEXT: call void @__kmpc_distribute_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP53]])
// IR-GPU-NEXT: br label [[OMP_PRECOND_END]]
// IR-GPU: omp.precond.end:
// IR-GPU-NEXT: ret void
@@ -1120,7 +940,7 @@ int main()
// IR-GPU-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
// IR-GPU-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// IR-GPU-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
-// IR-GPU-NEXT: call void @__kmpc_for_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB3]] to ptr), i32 [[TMP11]], i32 33, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
+// IR-GPU-NEXT: call void @__kmpc_for_static_init_4(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP11]], i32 93, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i32 1, i32 1)
// IR-GPU-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_LB_ASCAST]], align 4
// IR-GPU-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV_ASCAST]], align 4
// IR-GPU-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -1184,7 +1004,7 @@ int main()
// IR-GPU: omp.loop.exit:
// IR-GPU-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// IR-GPU-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
-// IR-GPU-NEXT: call void @__kmpc_for_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB3]] to ptr), i32 [[TMP28]])
+// IR-GPU-NEXT: call void @__kmpc_for_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP28]])
// IR-GPU-NEXT: br label [[OMP_PRECOND_END]]
// IR-GPU: omp.precond.end:
// IR-GPU-NEXT: ret void
@@ -3074,95 +2894,35 @@ int main()
// IR-GPU-NESTED-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
// IR-GPU-NESTED-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// IR-GPU-NESTED-NEXT: [[CONV13:%.*]] = zext i32 [[NVPTX_NUM_THREADS]] to i64
-// IR-GPU-NESTED-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
-// IR-GPU-NESTED-NEXT: call void @__kmpc_distribute_static_init_8(ptr addrspacecast (ptr addrspace(1) @[[GLOB2:[0-9]+]] to ptr), i32 [[TMP12]], i32 91, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_COMB_LB_ASCAST]], ptr [[DOTOMP_COMB_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i64 1, i64 [[CONV13]])
-// IR-GPU-NESTED-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[CMP14:%.*]] = icmp sgt i64 [[TMP13]], [[TMP14]]
-// IR-GPU-NESTED-NEXT: br i1 [[CMP14]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
-// IR-GPU-NESTED: cond.true:
-// IR-GPU-NESTED-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: br label [[COND_END:%.*]]
-// IR-GPU-NESTED: cond.false:
-// IR-GPU-NESTED-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: br label [[COND_END]]
-// IR-GPU-NESTED: cond.end:
-// IR-GPU-NESTED-NEXT: [[COND:%.*]] = phi i64 [ [[TMP15]], [[COND_TRUE]] ], [ [[TMP16]], [[COND_FALSE]] ]
-// IR-GPU-NESTED-NEXT: store i64 [[COND]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_COMB_LB_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: store i64 [[TMP17]], ptr [[DOTOMP_IV_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
-// IR-GPU-NESTED: omp.inner.for.cond:
-// IR-GPU-NESTED-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_IV_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP19]], 1
-// IR-GPU-NESTED-NEXT: [[CMP15:%.*]] = icmp slt i64 [[TMP18]], [[ADD]]
-// IR-GPU-NESTED-NEXT: br i1 [[CMP15]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
-// IR-GPU-NESTED: omp.inner.for.body:
-// IR-GPU-NESTED-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTOMP_COMB_LB_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP22:%.*]] = load i32, ptr [[N_ADDR_ASCAST]], align 4
-// IR-GPU-NESTED-NEXT: store i32 [[TMP22]], ptr addrspace(5) [[N_CASTED]], align 4
-// IR-GPU-NESTED-NEXT: [[TMP23:%.*]] = load i64, ptr addrspace(5) [[N_CASTED]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP24:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 0
-// IR-GPU-NESTED-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP20]] to ptr
+// IR-GPU-NESTED-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTOMP_COMB_LB_ASCAST]], align 8
+// IR-GPU-NESTED-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
+// IR-GPU-NESTED-NEXT: [[TMP13:%.*]] = load i32, ptr [[N_ADDR_ASCAST]], align 4
+// IR-GPU-NESTED-NEXT: store i32 [[TMP13]], ptr addrspace(5) [[N_CASTED]], align 4
+// IR-GPU-NESTED-NEXT: [[TMP14:%.*]] = load i64, ptr addrspace(5) [[N_CASTED]], align 8
+// IR-GPU-NESTED-NEXT: [[TMP15:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 0
+// IR-GPU-NESTED-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP11]] to ptr
+// IR-GPU-NESTED-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8
+// IR-GPU-NESTED-NEXT: [[TMP17:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 1
+// IR-GPU-NESTED-NEXT: [[TMP18:%.*]] = inttoptr i64 [[TMP12]] to ptr
+// IR-GPU-NESTED-NEXT: store ptr [[TMP18]], ptr [[TMP17]], align 8
+// IR-GPU-NESTED-NEXT: [[TMP19:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 2
+// IR-GPU-NESTED-NEXT: [[TMP20:%.*]] = inttoptr i64 [[TMP14]] to ptr
+// IR-GPU-NESTED-NEXT: store ptr [[TMP20]], ptr [[TMP19]], align 8
+// IR-GPU-NESTED-NEXT: [[TMP21:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 3
+// IR-GPU-NESTED-NEXT: [[TMP22:%.*]] = inttoptr i64 [[TMP0]] to ptr
+// IR-GPU-NESTED-NEXT: store ptr [[TMP22]], ptr [[TMP21]], align 8
+// IR-GPU-NESTED-NEXT: [[TMP23:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 4
+// IR-GPU-NESTED-NEXT: store ptr [[TMP1]], ptr [[TMP23]], align 8
+// IR-GPU-NESTED-NEXT: [[TMP24:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 5
+// IR-GPU-NESTED-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP2]] to ptr
// IR-GPU-NESTED-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP26:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 1
-// IR-GPU-NESTED-NEXT: [[TMP27:%.*]] = inttoptr i64 [[TMP21]] to ptr
-// IR-GPU-NESTED-NEXT: store ptr [[TMP27]], ptr [[TMP26]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP28:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 2
-// IR-GPU-NESTED-NEXT: [[TMP29:%.*]] = inttoptr i64 [[TMP23]] to ptr
-// IR-GPU-NESTED-NEXT: store ptr [[TMP29]], ptr [[TMP28]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP30:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 3
-// IR-GPU-NESTED-NEXT: [[TMP31:%.*]] = inttoptr i64 [[TMP0]] to ptr
-// IR-GPU-NESTED-NEXT: store ptr [[TMP31]], ptr [[TMP30]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP32:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 4
-// IR-GPU-NESTED-NEXT: store ptr [[TMP1]], ptr [[TMP32]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP33:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 5
-// IR-GPU-NESTED-NEXT: [[TMP34:%.*]] = inttoptr i64 [[TMP2]] to ptr
-// IR-GPU-NESTED-NEXT: store ptr [[TMP34]], ptr [[TMP33]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP35:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 6
-// IR-GPU-NESTED-NEXT: store ptr [[TMP3]], ptr [[TMP35]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP36:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP37:%.*]] = load i32, ptr [[TMP36]], align 4
-// IR-GPU-NESTED-NEXT: call void @__kmpc_parallel_60(ptr addrspacecast (ptr addrspace(1) @[[GLOB1]] to ptr), i32 [[TMP37]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l64_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 7, i32 0)
-// IR-GPU-NESTED-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
-// IR-GPU-NESTED: omp.inner.for.inc:
-// IR-GPU-NESTED-NEXT: [[TMP38:%.*]] = load i64, ptr [[DOTOMP_IV_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP39:%.*]] = load i64, ptr [[DOTOMP_STRIDE_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[ADD16:%.*]] = add nsw i64 [[TMP38]], [[TMP39]]
-// IR-GPU-NESTED-NEXT: store i64 [[ADD16]], ptr [[DOTOMP_IV_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP40:%.*]] = load i64, ptr [[DOTOMP_COMB_LB_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP41:%.*]] = load i64, ptr [[DOTOMP_STRIDE_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[ADD17:%.*]] = add nsw i64 [[TMP40]], [[TMP41]]
-// IR-GPU-NESTED-NEXT: store i64 [[ADD17]], ptr [[DOTOMP_COMB_LB_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP42:%.*]] = load i64, ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP43:%.*]] = load i64, ptr [[DOTOMP_STRIDE_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[ADD18:%.*]] = add nsw i64 [[TMP42]], [[TMP43]]
-// IR-GPU-NESTED-NEXT: store i64 [[ADD18]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP44:%.*]] = load i64, ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP45:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[CMP19:%.*]] = icmp sgt i64 [[TMP44]], [[TMP45]]
-// IR-GPU-NESTED-NEXT: br i1 [[CMP19]], label [[COND_TRUE20:%.*]], label [[COND_FALSE21:%.*]]
-// IR-GPU-NESTED: cond.true20:
-// IR-GPU-NESTED-NEXT: [[TMP46:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: br label [[COND_END22:%.*]]
-// IR-GPU-NESTED: cond.false21:
-// IR-GPU-NESTED-NEXT: [[TMP47:%.*]] = load i64, ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: br label [[COND_END22]]
-// IR-GPU-NESTED: cond.end22:
-// IR-GPU-NESTED-NEXT: [[COND23:%.*]] = phi i64 [ [[TMP46]], [[COND_TRUE20]] ], [ [[TMP47]], [[COND_FALSE21]] ]
-// IR-GPU-NESTED-NEXT: store i64 [[COND23]], ptr [[DOTOMP_COMB_UB_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP48:%.*]] = load i64, ptr [[DOTOMP_COMB_LB_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: store i64 [[TMP48]], ptr [[DOTOMP_IV_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: br label [[OMP_INNER_FOR_COND]]
-// IR-GPU-NESTED: omp.inner.for.end:
+// IR-GPU-NESTED-NEXT: [[TMP26:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 0, i64 6
+// IR-GPU-NESTED-NEXT: store ptr [[TMP3]], ptr [[TMP26]], align 8
+// IR-GPU-NESTED-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
+// IR-GPU-NESTED-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
+// IR-GPU-NESTED-NEXT: call void @__kmpc_parallel_60(ptr addrspacecast (ptr addrspace(1) @[[GLOB1]] to ptr), i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l64_omp_outlined_omp_outlined, ptr null, ptr [[CAPTURED_VARS_ADDRS_ASCAST]], i64 7, i32 0)
// IR-GPU-NESTED-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// IR-GPU-NESTED: omp.loop.exit:
-// IR-GPU-NESTED-NEXT: [[TMP49:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
-// IR-GPU-NESTED-NEXT: [[TMP50:%.*]] = load i32, ptr [[TMP49]], align 4
-// IR-GPU-NESTED-NEXT: call void @__kmpc_distribute_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP50]])
// IR-GPU-NESTED-NEXT: br label [[OMP_PRECOND_END]]
// IR-GPU-NESTED: omp.precond.end:
// IR-GPU-NESTED-NEXT: ret void
@@ -3266,7 +3026,7 @@ int main()
// IR-GPU-NESTED-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST_ASCAST]], align 4
// IR-GPU-NESTED-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// IR-GPU-NESTED-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP13]], align 4
-// IR-GPU-NESTED-NEXT: call void @__kmpc_for_static_init_8(ptr addrspacecast (ptr addrspace(1) @[[GLOB3:[0-9]+]] to ptr), i32 [[TMP14]], i32 33, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i64 1, i64 1)
+// IR-GPU-NESTED-NEXT: call void @__kmpc_for_static_init_8(ptr addrspacecast (ptr addrspace(1) @[[GLOB2:[0-9]+]] to ptr), i32 [[TMP14]], i32 93, ptr [[DOTOMP_IS_LAST_ASCAST]], ptr [[DOTOMP_LB_ASCAST]], ptr [[DOTOMP_UB_ASCAST]], ptr [[DOTOMP_STRIDE_ASCAST]], i64 1, i64 1)
// IR-GPU-NESTED-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_LB_ASCAST]], align 8
// IR-GPU-NESTED-NEXT: store i64 [[TMP15]], ptr [[DOTOMP_IV_ASCAST]], align 8
// IR-GPU-NESTED-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
@@ -3333,7 +3093,7 @@ int main()
// IR-GPU-NESTED: omp.loop.exit:
// IR-GPU-NESTED-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR_ASCAST]], align 8
// IR-GPU-NESTED-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
-// IR-GPU-NESTED-NEXT: call void @__kmpc_for_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB3]] to ptr), i32 [[TMP32]])
+// IR-GPU-NESTED-NEXT: call void @__kmpc_for_static_fini(ptr addrspacecast (ptr addrspace(1) @[[GLOB2]] to ptr), i32 [[TMP32]])
// IR-GPU-NESTED-NEXT: br label [[OMP_PRECOND_END]]
// IR-GPU-NESTED: omp.precond.end:
// IR-GPU-NESTED-NEXT: ret void
diff --git a/offload/test/offloading/gpupgo/pgo_atomic_teams.c b/offload/test/offloading/gpupgo/pgo_atomic_teams.c
index b3b72db080392..f9a0514c62886 100644
--- a/offload/test/offloading/gpupgo/pgo_atomic_teams.c
+++ b/offload/test/offloading/gpupgo/pgo_atomic_teams.c
@@ -46,8 +46,10 @@ int main() {
// LLVM-PGO-LABEL: __omp_offloading_{{[_0-9a-zA-Z]*}}_main_{{[_0-9a-zA-Z]*}}_omp_outlined:
// LLVM-PGO: Hash: {{0[xX][0-9a-fA-F]+}}
-// LLVM-PGO: Counters: 4
-// LLVM-PGO: Block counts: [{{.*}}, 0, {{.*}}, 0]
+// On GPU SPMD the combined `target teams distribute parallel for` skips the
+// outer team loop, so the teams-region outlined function has one PGO counter.
+// LLVM-PGO: Counters: 1
+// LLVM-PGO: Block counts: [{{.*}}]
// LLVM-PGO-LABEL: __omp_offloading_{{[_0-9a-zA-Z]*}}_main_{{[_0-9a-zA-Z]*}}_omp_outlined_omp_outlined:
// LLVM-PGO: Hash: {{0[xX][0-9a-fA-F]+}}
More information about the cfe-commits
mailing list