[clang] [llvm] [openmp] [OpenMP][offload] Fix dynamic schedule tracking (PR #97065)
Gheorghe-Teodor Bercea via cfe-commits
cfe-commits at lists.llvm.org
Fri Jun 28 13:18:00 PDT 2024
https://github.com/doru1004 updated https://github.com/llvm/llvm-project/pull/97065
>From 99aa3913c31c96d587fdff3f341fa648e3530d48 Mon Sep 17 00:00:00 2001
From: Doru Bercea <doru.bercea at amd.com>
Date: Thu, 27 Jun 2024 20:17:43 -0400
Subject: [PATCH] Fix dynamic schedule tracking
---
clang/lib/CodeGen/CGOpenMPRuntime.cpp | 14 +
clang/lib/CodeGen/CGOpenMPRuntime.h | 16 +
clang/lib/CodeGen/CGStmtOpenMP.cpp | 5 +
.../distribute_parallel_for_codegen.cpp | 36 +
.../distribute_parallel_for_simd_codegen.cpp | 156 +-
clang/test/OpenMP/nvptx_SPMD_codegen.cpp | 3957 +++++++++--------
clang/test/OpenMP/ordered_codegen.cpp | 98 +-
clang/test/OpenMP/parallel_for_codegen.cpp | 36 +-
.../OpenMP/target_parallel_for_codegen.cpp | 4 +
.../target_parallel_for_simd_codegen.cpp | 4 +
...stribute_parallel_for_schedule_codegen.cpp | 40 +
...ute_parallel_for_simd_schedule_codegen.cpp | 120 +-
...stribute_parallel_for_schedule_codegen.cpp | 40 +
...ute_parallel_for_simd_schedule_codegen.cpp | 120 +-
.../llvm/Frontend/OpenMP/OMPIRBuilder.h | 3 +
.../include/llvm/Frontend/OpenMP/OMPKinds.def | 3 +
llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp | 4 +
offload/DeviceRTL/CMakeLists.txt | 1 +
offload/DeviceRTL/include/Interface.h | 1 +
offload/DeviceRTL/include/Workshare.h | 30 +
offload/DeviceRTL/src/Kernel.cpp | 2 +
offload/DeviceRTL/src/Workshare.cpp | 77 +-
offload/test/offloading/dynamic-schedule.cpp | 57 +
offload/test/offloading/schedule.c | 84 +
openmp/runtime/src/kmp.h | 2 +
openmp/runtime/src/kmp_dispatch.cpp | 5 +
26 files changed, 2798 insertions(+), 2117 deletions(-)
create mode 100644 offload/DeviceRTL/include/Workshare.h
create mode 100644 offload/test/offloading/dynamic-schedule.cpp
create mode 100644 offload/test/offloading/schedule.c
diff --git a/clang/lib/CodeGen/CGOpenMPRuntime.cpp b/clang/lib/CodeGen/CGOpenMPRuntime.cpp
index f6d12d46cfc07..b47b521edd32c 100644
--- a/clang/lib/CodeGen/CGOpenMPRuntime.cpp
+++ b/clang/lib/CodeGen/CGOpenMPRuntime.cpp
@@ -2553,6 +2553,15 @@ void CGOpenMPRuntime::emitForDispatchInit(
Args);
}
+void CGOpenMPRuntime::emitForDispatchDeinit(CodeGenFunction &CGF,
+ SourceLocation Loc) {
+ if (!CGF.HaveInsertPoint())
+ return;
+ // Call __kmpc_dispatch_deinit(ident_t *loc, kmp_int32 tid);
+ llvm::Value *Args[] = {emitUpdateLocation(CGF, Loc), getThreadID(CGF, Loc)};
+ CGF.EmitRuntimeCall(OMPBuilder.createDispatchDeinitFunction(), Args);
+}
+
static void emitForStaticInitCall(
CodeGenFunction &CGF, llvm::Value *UpdateLocation, llvm::Value *ThreadId,
llvm::FunctionCallee ForStaticInitFunction, OpenMPSchedType Schedule,
@@ -11996,6 +12005,11 @@ void CGOpenMPSIMDRuntime::emitForDispatchInit(
llvm_unreachable("Not supported in SIMD-only mode");
}
+void CGOpenMPSIMDRuntime::emitForDispatchDeinit(CodeGenFunction &CGF,
+ SourceLocation Loc) {
+ llvm_unreachable("Not supported in SIMD-only mode");
+}
+
void CGOpenMPSIMDRuntime::emitForStaticInit(
CodeGenFunction &CGF, SourceLocation Loc, OpenMPDirectiveKind DKind,
const OpenMPScheduleTy &ScheduleKind, const StaticRTInput &Values) {
diff --git a/clang/lib/CodeGen/CGOpenMPRuntime.h b/clang/lib/CodeGen/CGOpenMPRuntime.h
index 522ae3d35d22d..f65314d014c08 100644
--- a/clang/lib/CodeGen/CGOpenMPRuntime.h
+++ b/clang/lib/CodeGen/CGOpenMPRuntime.h
@@ -946,6 +946,14 @@ class CGOpenMPRuntime {
unsigned IVSize, bool IVSigned, bool Ordered,
const DispatchRTInput &DispatchValues);
+ /// This is used for non static scheduled types and when the ordered
+ /// clause is present on the loop construct.
+ ///
+ /// \param CGF Reference to current CodeGenFunction.
+ /// \param Loc Clang source location.
+ ///
+ virtual void emitForDispatchDeinit(CodeGenFunction &CGF, SourceLocation Loc);
+
/// Struct with the values to be passed to the static runtime function
struct StaticRTInput {
/// Size of the iteration variable in bits.
@@ -1829,6 +1837,14 @@ class CGOpenMPSIMDRuntime final : public CGOpenMPRuntime {
unsigned IVSize, bool IVSigned, bool Ordered,
const DispatchRTInput &DispatchValues) override;
+ /// This is used for non static scheduled types and when the ordered
+ /// clause is present on the loop construct.
+ ///
+ /// \param CGF Reference to current CodeGenFunction.
+ /// \param Loc Clang source location.
+ ///
+ void emitForDispatchDeinit(CodeGenFunction &CGF, SourceLocation Loc) override;
+
/// Call the appropriate runtime routine to initialize it before start
/// of loop.
///
diff --git a/clang/lib/CodeGen/CGStmtOpenMP.cpp b/clang/lib/CodeGen/CGStmtOpenMP.cpp
index 43cf4dc630cd7..76ff8f5b234da 100644
--- a/clang/lib/CodeGen/CGStmtOpenMP.cpp
+++ b/clang/lib/CodeGen/CGStmtOpenMP.cpp
@@ -2985,12 +2985,14 @@ void CodeGenFunction::EmitOMPForOuterLoop(
// run-sched-var ICV. If the ICV is set to auto, the schedule is
// implementation defined
//
+ // __kmpc_dispatch_init();
// while(__kmpc_dispatch_next(&LB, &UB)) {
// idx = LB;
// while (idx <= UB) { BODY; ++idx;
// __kmpc_dispatch_fini_(4|8)[u](); // For ordered loops only.
// } // inner loop
// }
+ // __kmpc_dispatch_deinit();
//
// OpenMP [2.7.1, Loop Construct, Description, table 2-1]
// When schedule(static, chunk_size) is specified, iterations are divided into
@@ -3044,6 +3046,9 @@ void CodeGenFunction::EmitOMPForOuterLoop(
OuterLoopArgs.DKind = LoopArgs.DKind;
EmitOMPOuterLoop(DynamicOrOrdered, IsMonotonic, S, LoopScope, OuterLoopArgs,
emitOMPLoopBodyWithStopPoint, CodeGenOrdered);
+ if (DynamicOrOrdered) {
+ RT.emitForDispatchDeinit(*this, S.getBeginLoc());
+ }
}
static void emitEmptyOrdered(CodeGenFunction &, SourceLocation Loc,
diff --git a/clang/test/OpenMP/distribute_parallel_for_codegen.cpp b/clang/test/OpenMP/distribute_parallel_for_codegen.cpp
index a8892a06d4b30..109f1690e448b 100644
--- a/clang/test/OpenMP/distribute_parallel_for_codegen.cpp
+++ b/clang/test/OpenMP/distribute_parallel_for_codegen.cpp
@@ -2283,6 +2283,9 @@ int main() {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: [[TMP34:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[TMP34]], align 4
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP35]])
// CHECK1-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1: omp.precond.end:
// CHECK1-NEXT: ret void
@@ -2533,6 +2536,9 @@ int main() {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: [[TMP35:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP36]])
// CHECK1-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1: omp.precond.end:
// CHECK1-NEXT: ret void
@@ -4010,6 +4016,9 @@ int main() {
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
+// CHECK3-NEXT: [[TMP34:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP35:%.*]] = load i32, ptr [[TMP34]], align 4
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP35]])
// CHECK3-NEXT: br label [[OMP_PRECOND_END]]
// CHECK3: omp.precond.end:
// CHECK3-NEXT: ret void
@@ -4253,6 +4262,9 @@ int main() {
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
+// CHECK3-NEXT: [[TMP35:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP36]])
// CHECK3-NEXT: br label [[OMP_PRECOND_END]]
// CHECK3: omp.precond.end:
// CHECK3-NEXT: ret void
@@ -6314,6 +6326,9 @@ int main() {
// CHECK9: omp.dispatch.inc:
// CHECK9-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK9: omp.dispatch.end:
+// CHECK9-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
+// CHECK9-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP31]])
// CHECK9-NEXT: br label [[OMP_PRECOND_END]]
// CHECK9: omp.precond.end:
// CHECK9-NEXT: ret void
@@ -6554,6 +6569,9 @@ int main() {
// CHECK9: omp.dispatch.inc:
// CHECK9-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK9: omp.dispatch.end:
+// CHECK9-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
+// CHECK9-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP32]])
// CHECK9-NEXT: br label [[OMP_PRECOND_END]]
// CHECK9: omp.precond.end:
// CHECK9-NEXT: ret void
@@ -8627,6 +8645,9 @@ int main() {
// CHECK9: omp.dispatch.inc:
// CHECK9-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK9: omp.dispatch.end:
+// CHECK9-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
+// CHECK9-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP31]])
// CHECK9-NEXT: br label [[OMP_PRECOND_END]]
// CHECK9: omp.precond.end:
// CHECK9-NEXT: ret void
@@ -8867,6 +8888,9 @@ int main() {
// CHECK9: omp.dispatch.inc:
// CHECK9-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK9: omp.dispatch.end:
+// CHECK9-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
+// CHECK9-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP32]])
// CHECK9-NEXT: br label [[OMP_PRECOND_END]]
// CHECK9: omp.precond.end:
// CHECK9-NEXT: ret void
@@ -10884,6 +10908,9 @@ int main() {
// CHECK11: omp.dispatch.inc:
// CHECK11-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK11: omp.dispatch.end:
+// CHECK11-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
+// CHECK11-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP31]])
// CHECK11-NEXT: br label [[OMP_PRECOND_END]]
// CHECK11: omp.precond.end:
// CHECK11-NEXT: ret void
@@ -11117,6 +11144,9 @@ int main() {
// CHECK11: omp.dispatch.inc:
// CHECK11-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK11: omp.dispatch.end:
+// CHECK11-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
+// CHECK11-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP32]])
// CHECK11-NEXT: br label [[OMP_PRECOND_END]]
// CHECK11: omp.precond.end:
// CHECK11-NEXT: ret void
@@ -13146,6 +13176,9 @@ int main() {
// CHECK11: omp.dispatch.inc:
// CHECK11-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK11: omp.dispatch.end:
+// CHECK11-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
+// CHECK11-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP31]])
// CHECK11-NEXT: br label [[OMP_PRECOND_END]]
// CHECK11: omp.precond.end:
// CHECK11-NEXT: ret void
@@ -13379,6 +13412,9 @@ int main() {
// CHECK11: omp.dispatch.inc:
// CHECK11-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK11: omp.dispatch.end:
+// CHECK11-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
+// CHECK11-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP32]])
// CHECK11-NEXT: br label [[OMP_PRECOND_END]]
// CHECK11: omp.precond.end:
// CHECK11-NEXT: ret void
diff --git a/clang/test/OpenMP/distribute_parallel_for_simd_codegen.cpp b/clang/test/OpenMP/distribute_parallel_for_simd_codegen.cpp
index a23d538910edd..14dcb17358061 100644
--- a/clang/test/OpenMP/distribute_parallel_for_simd_codegen.cpp
+++ b/clang/test/OpenMP/distribute_parallel_for_simd_codegen.cpp
@@ -2415,12 +2415,15 @@ int main() {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
-// CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK1-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
-// CHECK1-NEXT: br i1 [[TMP35]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK1-NEXT: [[TMP34:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[TMP34]], align 4
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP35]])
+// CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0
+// CHECK1-NEXT: br i1 [[TMP37]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK1: .omp.final.then:
-// CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
-// CHECK1-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP36]], 0
+// CHECK1-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP38]], 0
// CHECK1-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1
// CHECK1-NEXT: [[MUL14:%.*]] = mul nsw i32 [[DIV13]], 1
// CHECK1-NEXT: [[ADD15:%.*]] = add nsw i32 0, [[MUL14]]
@@ -2689,12 +2692,15 @@ int main() {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
-// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK1-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0
-// CHECK1-NEXT: br i1 [[TMP36]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK1-NEXT: [[TMP35:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP36]])
+// CHECK1-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
+// CHECK1-NEXT: br i1 [[TMP38]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK1: .omp.final.then:
-// CHECK1-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK1-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP37]], 0
+// CHECK1-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP39]], 0
// CHECK1-NEXT: [[DIV14:%.*]] = sdiv i32 [[SUB13]], 1
// CHECK1-NEXT: [[MUL15:%.*]] = mul nsw i32 [[DIV14]], 1
// CHECK1-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL15]]
@@ -4310,12 +4316,15 @@ int main() {
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
-// CHECK3-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK3-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
-// CHECK3-NEXT: br i1 [[TMP35]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK3-NEXT: [[TMP34:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP35:%.*]] = load i32, ptr [[TMP34]], align 4
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP35]])
+// CHECK3-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0
+// CHECK3-NEXT: br i1 [[TMP37]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK3: .omp.final.then:
-// CHECK3-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
-// CHECK3-NEXT: [[SUB9:%.*]] = sub nsw i32 [[TMP36]], 0
+// CHECK3-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK3-NEXT: [[SUB9:%.*]] = sub nsw i32 [[TMP38]], 0
// CHECK3-NEXT: [[DIV10:%.*]] = sdiv i32 [[SUB9]], 1
// CHECK3-NEXT: [[MUL11:%.*]] = mul nsw i32 [[DIV10]], 1
// CHECK3-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]]
@@ -4577,12 +4586,15 @@ int main() {
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
-// CHECK3-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK3-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0
-// CHECK3-NEXT: br i1 [[TMP36]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK3-NEXT: [[TMP35:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP36]])
+// CHECK3-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
+// CHECK3-NEXT: br i1 [[TMP38]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK3: .omp.final.then:
-// CHECK3-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK3-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP37]], 0
+// CHECK3-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP39]], 0
// CHECK3-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK3-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1
// CHECK3-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]]
@@ -6836,12 +6848,15 @@ int main() {
// CHECK9: omp.dispatch.inc:
// CHECK9-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK9: omp.dispatch.end:
-// CHECK9-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK9-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0
-// CHECK9-NEXT: br i1 [[TMP31]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK9-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
+// CHECK9-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP31]])
+// CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0
+// CHECK9-NEXT: br i1 [[TMP33]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then:
-// CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
-// CHECK9-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP32]], 0
+// CHECK9-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK9-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP34]], 0
// CHECK9-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1
// CHECK9-NEXT: [[MUL14:%.*]] = mul nsw i32 [[DIV13]], 1
// CHECK9-NEXT: [[ADD15:%.*]] = add nsw i32 0, [[MUL14]]
@@ -7100,12 +7115,15 @@ int main() {
// CHECK9: omp.dispatch.inc:
// CHECK9-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK9: omp.dispatch.end:
-// CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0
-// CHECK9-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK9-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
+// CHECK9-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP32]])
+// CHECK9-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
+// CHECK9-NEXT: br i1 [[TMP34]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then:
-// CHECK9-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP33]], 0
+// CHECK9-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP35]], 0
// CHECK9-NEXT: [[DIV14:%.*]] = sdiv i32 [[SUB13]], 1
// CHECK9-NEXT: [[MUL15:%.*]] = mul nsw i32 [[DIV14]], 1
// CHECK9-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL15]]
@@ -9302,12 +9320,15 @@ int main() {
// CHECK9: omp.dispatch.inc:
// CHECK9-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK9: omp.dispatch.end:
-// CHECK9-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK9-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0
-// CHECK9-NEXT: br i1 [[TMP31]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK9-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
+// CHECK9-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP31]])
+// CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0
+// CHECK9-NEXT: br i1 [[TMP33]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then:
-// CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
-// CHECK9-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP32]], 0
+// CHECK9-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK9-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP34]], 0
// CHECK9-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1
// CHECK9-NEXT: [[MUL14:%.*]] = mul nsw i32 [[DIV13]], 1
// CHECK9-NEXT: [[ADD15:%.*]] = add nsw i32 0, [[MUL14]]
@@ -9566,12 +9587,15 @@ int main() {
// CHECK9: omp.dispatch.inc:
// CHECK9-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK9: omp.dispatch.end:
-// CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0
-// CHECK9-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK9-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
+// CHECK9-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP32]])
+// CHECK9-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
+// CHECK9-NEXT: br i1 [[TMP34]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then:
-// CHECK9-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP33]], 0
+// CHECK9-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP35]], 0
// CHECK9-NEXT: [[DIV14:%.*]] = sdiv i32 [[SUB13]], 1
// CHECK9-NEXT: [[MUL15:%.*]] = mul nsw i32 [[DIV14]], 1
// CHECK9-NEXT: [[ADD16:%.*]] = add nsw i32 0, [[MUL15]]
@@ -11727,12 +11751,15 @@ int main() {
// CHECK11: omp.dispatch.inc:
// CHECK11-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK11: omp.dispatch.end:
-// CHECK11-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK11-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0
-// CHECK11-NEXT: br i1 [[TMP31]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK11-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
+// CHECK11-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP31]])
+// CHECK11-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0
+// CHECK11-NEXT: br i1 [[TMP33]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK11: .omp.final.then:
-// CHECK11-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
-// CHECK11-NEXT: [[SUB9:%.*]] = sub nsw i32 [[TMP32]], 0
+// CHECK11-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK11-NEXT: [[SUB9:%.*]] = sub nsw i32 [[TMP34]], 0
// CHECK11-NEXT: [[DIV10:%.*]] = sdiv i32 [[SUB9]], 1
// CHECK11-NEXT: [[MUL11:%.*]] = mul nsw i32 [[DIV10]], 1
// CHECK11-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]]
@@ -11984,12 +12011,15 @@ int main() {
// CHECK11: omp.dispatch.inc:
// CHECK11-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK11: omp.dispatch.end:
-// CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0
-// CHECK11-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK11-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
+// CHECK11-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP32]])
+// CHECK11-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
+// CHECK11-NEXT: br i1 [[TMP34]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK11: .omp.final.then:
-// CHECK11-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP33]], 0
+// CHECK11-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP35]], 0
// CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK11-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1
// CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]]
@@ -14142,12 +14172,15 @@ int main() {
// CHECK11: omp.dispatch.inc:
// CHECK11-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK11: omp.dispatch.end:
-// CHECK11-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK11-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0
-// CHECK11-NEXT: br i1 [[TMP31]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK11-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
+// CHECK11-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP31]])
+// CHECK11-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0
+// CHECK11-NEXT: br i1 [[TMP33]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK11: .omp.final.then:
-// CHECK11-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
-// CHECK11-NEXT: [[SUB9:%.*]] = sub nsw i32 [[TMP32]], 0
+// CHECK11-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK11-NEXT: [[SUB9:%.*]] = sub nsw i32 [[TMP34]], 0
// CHECK11-NEXT: [[DIV10:%.*]] = sdiv i32 [[SUB9]], 1
// CHECK11-NEXT: [[MUL11:%.*]] = mul nsw i32 [[DIV10]], 1
// CHECK11-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]]
@@ -14399,12 +14432,15 @@ int main() {
// CHECK11: omp.dispatch.inc:
// CHECK11-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK11: omp.dispatch.end:
-// CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0
-// CHECK11-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK11-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
+// CHECK11-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP32]])
+// CHECK11-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
+// CHECK11-NEXT: br i1 [[TMP34]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK11: .omp.final.then:
-// CHECK11-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP33]], 0
+// CHECK11-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP35]], 0
// CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK11-NEXT: [[MUL12:%.*]] = mul nsw i32 [[DIV11]], 1
// CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 0, [[MUL12]]
diff --git a/clang/test/OpenMP/nvptx_SPMD_codegen.cpp b/clang/test/OpenMP/nvptx_SPMD_codegen.cpp
index d47025cd1acaa..54f7694b55da3 100644
--- a/clang/test/OpenMP/nvptx_SPMD_codegen.cpp
+++ b/clang/test/OpenMP/nvptx_SPMD_codegen.cpp
@@ -348,60 +348,60 @@ int a;
// 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_GRP249:![0-9]+]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221:![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_GRP249]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP221]]
// CHECK-64-NEXT: [[TOBOOL2:%.*]] = trunc i8 [[TMP11]] to i1
// CHECK-64-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL2]] to i8
-// CHECK-64-NEXT: store i8 [[FROMBOOL]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1, !llvm.access.group [[ACC_GRP249]]
-// CHECK-64-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group [[ACC_GRP249]]
+// CHECK-64-NEXT: store i8 [[FROMBOOL]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1, !llvm.access.group [[ACC_GRP221]]
+// CHECK-64-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
+// CHECK-64-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
+// CHECK-64-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP249]]
+// CHECK-64-NEXT: store ptr [[TMP18]], ptr [[TMP17]], align 8, !llvm.access.group [[ACC_GRP221]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP221]]
// CHECK-64-NEXT: [[TOBOOL3:%.*]] = trunc i8 [[TMP19]] to i1
// CHECK-64-NEXT: [[TMP20:%.*]] = zext i1 [[TOBOOL3]] to i32
-// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP249]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP249]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP221]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP249]]
-// CHECK-64-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP249]]
-// CHECK-64-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP249]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-64-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-64-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
-// CHECK-64-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP249]]
-// CHECK-64-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP249]]
+// CHECK-64-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-64-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-64-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
-// CHECK-64-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP249]]
+// CHECK-64-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-64-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
+// CHECK-64-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
-// CHECK-64-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP249]]
-// CHECK-64-NEXT: store i32 [[TMP29]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP249]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP250:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND10]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-64-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-64-NEXT: store i32 [[TMP29]], 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: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK-64: omp_if.else:
@@ -460,7 +460,7 @@ int a;
// 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 [[LOOP253:![0-9]+]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND11]], !llvm.loop [[LOOP225:![0-9]+]]
// CHECK-64: omp.inner.for.end28:
// CHECK-64-NEXT: br label [[OMP_IF_END]]
// CHECK-64: omp_if.end:
@@ -518,25 +518,25 @@ int a;
// 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_GRP255:![0-9]+]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227:![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_GRP255]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP227]]
// 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_GRP255]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
// 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_GRP255]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP227]]
// 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_GRP255]]
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP255]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP227]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP255]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP256:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP228:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK-64: omp_if.else:
@@ -565,7 +565,7 @@ 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 [[LOOP258:![0-9]+]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND4]], !llvm.loop [[LOOP230:![0-9]+]]
// CHECK-64: omp.inner.for.end13:
// CHECK-64-NEXT: br label [[OMP_IF_END]]
// CHECK-64: omp_if.end:
@@ -625,25 +625,25 @@ int a;
// 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_GRP259:![0-9]+]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231:![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_GRP259]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP231]]
// 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_GRP259]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231]]
// 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_GRP259]]
+// 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: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259]]
-// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP231]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP260:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], 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_IF_END:%.*]]
// CHECK-64: omp_if.else:
@@ -672,7 +672,7 @@ 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 [[LOOP262:![0-9]+]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND4]], !llvm.loop [[LOOP234:![0-9]+]]
// CHECK-64: omp.inner.for.end13:
// CHECK-64-NEXT: br label [[OMP_IF_END]]
// CHECK-64: omp_if.end:
@@ -750,49 +750,49 @@ 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_GRP263:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP235:![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_GRP263]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP263]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP235]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP263]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP235]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP263]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP263]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP263]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP263]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP263]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP263]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP263]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP263]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP264:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP236:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -853,23 +853,23 @@ 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_GRP266:![0-9]+]]
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP266]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP238:![0-9]+]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP238]]
// 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_GRP266]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP238]]
// 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_GRP266]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP238]]
// 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_GRP266]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP238]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP266]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP267:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP238]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP239:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -943,49 +943,49 @@ 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_GRP269:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP241:![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_GRP269]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP269]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP241]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP269]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP241]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP269]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP269]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP269]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP269]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP269]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP269]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP269]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP269]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP270:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP242:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -1035,25 +1035,25 @@ 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_GRP272:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP244:![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_GRP272]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP244]]
// 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_GRP272]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP244]]
// 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_GRP272]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP244]]
// 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_GRP272]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP272]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP244]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP244]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP272]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP273:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP244]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP245:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -1127,49 +1127,49 @@ 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_GRP275:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP247:![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_GRP275]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP275]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP247]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP275]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP247]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP275]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP275]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP275]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP275]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP275]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP275]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP275]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP275]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP276:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP248:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -1227,28 +1227,29 @@ 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_GRP278:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP278]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP250:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP250]]
// 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_GRP278]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP250]]
// 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_GRP278]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP250]]
// 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_GRP278]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP250]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP278]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP279:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP250]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP251:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK-64-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -1318,49 +1319,49 @@ 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_GRP281:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP253:![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_GRP281]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP281]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP253]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP281]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP253]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP281]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP281]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP281]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP281]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP281]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP281]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP281]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP281]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP282:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP254:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -1418,28 +1419,29 @@ 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_GRP284:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP284]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP256:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP256]]
// 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_GRP284]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP256]]
// 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_GRP284]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP256]]
// 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_GRP284]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP256]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP284]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP285:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP256]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP257:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK-64-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -1509,49 +1511,49 @@ 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_GRP287:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259:![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_GRP287]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP287]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP259]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP287]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP259]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP287]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP287]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP287]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP287]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP287]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP287]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP287]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP287]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP288:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP260:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -1609,28 +1611,29 @@ 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_GRP290:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP290]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP262:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP262]]
// 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_GRP290]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP262]]
// 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_GRP290]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP262]]
// 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_GRP290]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP262]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP290]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP291:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP262]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP263:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK-64-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -1700,49 +1703,49 @@ 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_GRP293:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP265:![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_GRP293]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP293]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP265]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP293]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP265]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP293]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP293]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP293]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP293]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP293]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP293]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP293]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP293]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP294:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP266:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -1800,28 +1803,29 @@ 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_GRP296:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP296]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP268:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP268]]
// 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_GRP296]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP268]]
// 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_GRP296]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP268]]
// 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_GRP296]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP268]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP296]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP297:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP268]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP269:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK-64-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -2541,28 +2545,29 @@ 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_GRP299:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP299]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP271:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP271]]
// 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_GRP299]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP271]]
// 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_GRP299]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP271]]
// 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_GRP299]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP271]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP299]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP300:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP271]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP272:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: ret void
//
//
@@ -2718,28 +2723,29 @@ 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_GRP302:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP302]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP274:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP274]]
// 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_GRP302]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP274]]
// 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_GRP302]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP274]]
// 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_GRP302]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP274]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP302]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP303:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP274]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP275:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: ret void
//
//
@@ -2895,28 +2901,29 @@ 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_GRP305:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP305]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP277:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP277]]
// 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_GRP305]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP277]]
// 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_GRP305]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP277]]
// 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_GRP305]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP277]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP305]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP306:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP277]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP278:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: ret void
//
//
@@ -3072,28 +3079,29 @@ 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_GRP308:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP308]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP280:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP280]]
// 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_GRP308]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP280]]
// 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_GRP308]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP280]]
// 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_GRP308]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP280]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP308]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP309:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP280]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP281:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: ret void
//
//
@@ -3157,49 +3165,49 @@ 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_GRP311:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP283:![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_GRP311]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP311]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP283]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP311]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP283]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP311]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP311]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP311]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP311]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP311]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP311]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP311]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP311]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP312:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP284:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -3249,25 +3257,25 @@ 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_GRP314:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP286:![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_GRP314]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP286]]
// 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_GRP314]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP286]]
// 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_GRP314]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP286]]
// 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_GRP314]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP314]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP286]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP286]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP314]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP315:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP286]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP287:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -3343,49 +3351,49 @@ 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_GRP317:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP289:![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_GRP317]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP317]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP289]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP317]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP289]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP317]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP317]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP317]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP317]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP317]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP317]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP317]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP317]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP318:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP290:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -3446,23 +3454,23 @@ 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_GRP320:![0-9]+]]
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP320]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP292:![0-9]+]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP292]]
// 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_GRP320]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP292]]
// 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_GRP320]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP292]]
// 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_GRP320]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP292]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP320]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP321:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP292]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP293:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -3537,29 +3545,29 @@ 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_GRP323:![0-9]+]]
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP323]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP295:![0-9]+]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP295]]
// 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_GRP323]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP295]]
// 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_GRP323]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP295]]
// 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_GRP323]]
+// CHECK-64-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8, !llvm.access.group [[ACC_GRP295]]
// 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_GRP323]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP323]]
+// CHECK-64-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8, !llvm.access.group [[ACC_GRP295]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP295]]
// 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_GRP323]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP323]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP295]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP295]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP15]], [[TMP16]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP323]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP324:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP295]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP296:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -3612,25 +3620,25 @@ 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_GRP326:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP298:![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_GRP326]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8, !llvm.access.group [[ACC_GRP298]]
// 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_GRP326]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP298]]
// 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_GRP326]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP298]]
// 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_GRP326]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP326]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP298]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP298]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP326]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP327:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP298]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP299:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -3724,49 +3732,49 @@ 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_GRP329:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP301:![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_GRP329]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP329]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP301]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP329]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP301]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP329]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP329]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP329]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP329]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP329]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP329]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP329]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP329]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP330:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP302:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -3824,28 +3832,29 @@ 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_GRP332:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP332]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP304:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP304]]
// 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_GRP332]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP304]]
// 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_GRP332]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP304]]
// 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_GRP332]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP304]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP332]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP333:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP304]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP305:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK-64-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -3915,49 +3924,49 @@ 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_GRP335:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP307:![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_GRP335]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP335]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP307]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP335]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP307]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP335]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP335]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP335]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP335]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP335]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP335]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP335]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP335]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP336:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP308:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -4015,28 +4024,29 @@ 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_GRP338:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP338]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP310:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP310]]
// 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_GRP338]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP310]]
// 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_GRP338]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP310]]
// 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_GRP338]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP310]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP338]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP339:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP310]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP311:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK-64-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -4106,49 +4116,49 @@ 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_GRP341:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP313:![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_GRP341]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP341]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP313]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP341]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP313]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP341]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP341]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP341]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP341]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP341]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP341]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP341]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP341]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP342:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP314:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -4206,28 +4216,29 @@ 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_GRP344:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP344]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP316:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP316]]
// 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_GRP344]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP316]]
// 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_GRP344]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP316]]
// 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_GRP344]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP316]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP344]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP345:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP316]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP317:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK-64-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -4297,49 +4308,49 @@ 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_GRP347:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP319:![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_GRP347]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
+// CHECK-64-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 8, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
-// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP347]]
+// CHECK-64-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8, !llvm.access.group [[ACC_GRP319]]
+// CHECK-64-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
-// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP347]]
+// CHECK-64-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-64-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP319]]
// CHECK-64-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
-// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP347]]
-// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP347]]
-// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP347]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-64-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-64-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
-// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP347]]
-// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP347]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-64-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-64-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
-// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP347]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-64-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
+// CHECK-64-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
-// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP347]]
-// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP347]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP348:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-64-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-64-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP320:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -4397,28 +4408,29 @@ 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_GRP350:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP350]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP322:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP322]]
// 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_GRP350]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP322]]
// 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_GRP350]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP322]]
// 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_GRP350]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP322]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP350]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP351:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP322]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP323:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK-64-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -5100,28 +5112,29 @@ 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_GRP353:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP353]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP325:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP325]]
// 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_GRP353]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP325]]
// 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_GRP353]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP325]]
// 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_GRP353]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP325]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP353]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP354:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP325]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP326:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: ret void
//
//
@@ -5277,28 +5290,29 @@ 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_GRP356:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP356]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP328:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP328]]
// 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_GRP356]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP328]]
// 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_GRP356]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP328]]
// 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_GRP356]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP328]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP356]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP357:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP328]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP329:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: ret void
//
//
@@ -5454,28 +5468,29 @@ 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_GRP359:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP359]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP331:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP331]]
// 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_GRP359]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP331]]
// 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_GRP359]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP331]]
// 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_GRP359]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP331]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP359]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP360:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP331]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP332:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: ret void
//
//
@@ -5631,28 +5646,29 @@ 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_GRP362:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP362]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP334:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP334]]
// 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_GRP362]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP334]]
// 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_GRP362]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP334]]
// 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_GRP362]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP334]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP362]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP363:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP334]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP335:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: ret void
//
//
@@ -6327,28 +6343,29 @@ 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_GRP365:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP365]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP337:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP337]]
// 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_GRP365]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP337]]
// 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_GRP365]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP337]]
// 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_GRP365]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP337]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP365]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP366:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP337]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP338:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: ret void
//
//
@@ -6504,28 +6521,29 @@ 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_GRP368:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP368]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP340:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP340]]
// 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_GRP368]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP340]]
// 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_GRP368]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP340]]
// 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_GRP368]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP340]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP368]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP369:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP340]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP341:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: ret void
//
//
@@ -6681,28 +6699,29 @@ 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_GRP371:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP371]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP343:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP343]]
// 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_GRP371]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP343]]
// 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_GRP371]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP343]]
// 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_GRP371]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP343]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP371]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP372:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP343]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP344:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: ret void
//
//
@@ -6858,28 +6877,29 @@ 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_GRP374:![0-9]+]]
-// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP374]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP346:![0-9]+]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP346]]
// 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_GRP374]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP346]]
// 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_GRP374]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP346]]
// 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_GRP374]]
+// CHECK-64-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP346]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP374]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP375:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP346]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP347:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-64-NEXT: ret void
//
//
@@ -7205,28 +7225,29 @@ 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_GRP377:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP377]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP349:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP349]]
// 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_GRP377]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP349]]
// 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_GRP377]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP349]]
// 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_GRP377]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP349]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP377]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP378:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP349]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP350:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -7279,28 +7300,29 @@ 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_GRP380:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP380]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP352:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP352]]
// 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_GRP380]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP352]]
// 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_GRP380]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP352]]
// 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_GRP380]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP352]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP380]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP381:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP352]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP353:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -7353,28 +7375,29 @@ 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_GRP383:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP383]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP355:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP355]]
// 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_GRP383]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP355]]
// 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_GRP383]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP355]]
// 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_GRP383]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP355]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP383]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP384:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP355]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP356:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -7427,28 +7450,29 @@ 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_GRP386:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP386]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP358:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP358]]
// 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_GRP386]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP358]]
// 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_GRP386]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP358]]
// 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_GRP386]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP358]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP386]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP387:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP358]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP359:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -7518,23 +7542,23 @@ 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_GRP389:![0-9]+]]
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP389]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP361:![0-9]+]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP361]]
// 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_GRP389]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP361]]
// 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_GRP389]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP361]]
// 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_GRP389]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP361]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP389]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP390:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP361]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP362:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -7614,23 +7638,23 @@ 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_GRP392:![0-9]+]]
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP392]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP364:![0-9]+]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP364]]
// 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_GRP392]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP364]]
// 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_GRP392]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP364]]
// 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_GRP392]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP364]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP392]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP393:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP364]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP365:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -7707,23 +7731,23 @@ 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_GRP395:![0-9]+]]
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP395]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP367:![0-9]+]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP367]]
// 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_GRP395]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP367]]
// 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_GRP395]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP367]]
// 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_GRP395]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP367]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP395]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP396:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP367]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP368:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -7798,28 +7822,29 @@ 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_GRP398:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP398]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP370:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP370]]
// 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_GRP398]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP370]]
// 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_GRP398]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP370]]
// 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_GRP398]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP370]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP398]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP399:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP370]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP371:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP9:%.*]] = icmp ne i32 [[TMP8]], 0
// CHECK-64-NEXT: br i1 [[TMP9]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -7880,28 +7905,29 @@ 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_GRP401:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP401]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP373:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP373]]
// 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_GRP401]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP373]]
// 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_GRP401]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP373]]
// 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_GRP401]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP373]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP401]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP402:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP373]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP374:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP9:%.*]] = icmp ne i32 [[TMP8]], 0
// CHECK-64-NEXT: br i1 [[TMP9]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -7962,28 +7988,29 @@ 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_GRP404:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP404]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP376:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP376]]
// 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_GRP404]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP376]]
// 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_GRP404]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP376]]
// 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_GRP404]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP376]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP404]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP405:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP376]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP377:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP9:%.*]] = icmp ne i32 [[TMP8]], 0
// CHECK-64-NEXT: br i1 [[TMP9]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -8044,28 +8071,29 @@ 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_GRP407:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP407]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP379:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP379]]
// 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_GRP407]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP379]]
// 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_GRP407]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP379]]
// 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_GRP407]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP379]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP407]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP408:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP379]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP380:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP9:%.*]] = icmp ne i32 [[TMP8]], 0
// CHECK-64-NEXT: br i1 [[TMP9]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -8126,29 +8154,30 @@ 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_GRP410:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP410]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP382:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP382]]
// 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_GRP410]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP382]]
// 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_GRP410]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP382]]
// 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_GRP410]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP382]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP410]]
-// CHECK-64-NEXT: call void @__kmpc_dispatch_fini_4(ptr @[[GLOB1]], i32 [[TMP1]]), !llvm.access.group [[ACC_GRP410]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP411:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP382]]
+// CHECK-64-NEXT: call void @__kmpc_dispatch_fini_4(ptr @[[GLOB1]], i32 [[TMP1]]), !llvm.access.group [[ACC_GRP382]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP383:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP9:%.*]] = icmp ne i32 [[TMP8]], 0
// CHECK-64-NEXT: br i1 [[TMP9]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -8214,23 +8243,23 @@ 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_GRP413:![0-9]+]]
-// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP413]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP385:![0-9]+]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP385]]
// 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_GRP413]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP385]]
// 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_GRP413]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP385]]
// 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_GRP413]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP385]]
// CHECK-64-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
-// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP413]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP414:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP385]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP386:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-64: omp.loop.exit:
@@ -8307,23 +8336,23 @@ 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_GRP416:![0-9]+]]
-// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP416]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP388:![0-9]+]]
+// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP388]]
// 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_GRP416]]
+// CHECK-64-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP388]]
// 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_GRP416]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP388]]
// 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_GRP416]]
+// CHECK-64-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP388]]
// CHECK-64-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP416]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP417:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP388]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP389:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
@@ -8398,28 +8427,29 @@ 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_GRP419:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP419]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP391:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP391]]
// 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_GRP419]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP391]]
// 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_GRP419]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP391]]
// 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_GRP419]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP391]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP419]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP420:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP391]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP392:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP9:%.*]] = icmp ne i32 [[TMP8]], 0
// CHECK-64-NEXT: br i1 [[TMP9]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -8480,28 +8510,29 @@ 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_GRP422:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP422]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP394:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP394]]
// 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_GRP422]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP394]]
// 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_GRP422]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP394]]
// 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_GRP422]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP394]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP422]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP423:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP394]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP395:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP9:%.*]] = icmp ne i32 [[TMP8]], 0
// CHECK-64-NEXT: br i1 [[TMP9]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -8562,28 +8593,29 @@ 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_GRP425:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP425]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP397:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP397]]
// 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_GRP425]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP397]]
// 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_GRP425]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP397]]
// 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_GRP425]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP397]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP425]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP426:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP397]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP398:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP9:%.*]] = icmp ne i32 [[TMP8]], 0
// CHECK-64-NEXT: br i1 [[TMP9]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -8644,28 +8676,29 @@ 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_GRP428:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP428]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP400:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP400]]
// 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_GRP428]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP400]]
// 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_GRP428]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP400]]
// 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_GRP428]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP400]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP428]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP429:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP400]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP401:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK-64-NEXT: [[TMP9:%.*]] = icmp ne i32 [[TMP8]], 0
// CHECK-64-NEXT: br i1 [[TMP9]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -8994,28 +9027,29 @@ 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_GRP431:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP431]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP403:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP403]]
// 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_GRP431]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP403]]
// 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_GRP431]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP403]]
// 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_GRP431]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP403]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP431]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP432:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP403]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP404:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -9068,28 +9102,29 @@ 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_GRP434:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP434]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP406:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP406]]
// 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_GRP434]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP406]]
// 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_GRP434]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP406]]
// 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_GRP434]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP406]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP434]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP435:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP406]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP407:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -9142,28 +9177,29 @@ 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_GRP437:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP437]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP409:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP409]]
// 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_GRP437]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP409]]
// 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_GRP437]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP409]]
// 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_GRP437]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP409]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP437]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP438:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP409]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP410:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -9216,28 +9252,29 @@ 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_GRP440:![0-9]+]]
-// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP440]]
+// CHECK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP412:![0-9]+]]
+// CHECK-64-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP412]]
// 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_GRP440]]
+// CHECK-64-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP412]]
// 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_GRP440]]
+// CHECK-64-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP412]]
// 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_GRP440]]
+// CHECK-64-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP412]]
// CHECK-64-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP440]]
-// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP441:![0-9]+]]
+// CHECK-64-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP412]]
+// CHECK-64-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP413:![0-9]+]]
// CHECK-64: omp.inner.for.end:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-64: omp.dispatch.inc:
// CHECK-64-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-64: omp.dispatch.end:
+// CHECK-64-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-64-NEXT: ret void
//
//
@@ -9317,58 +9354,58 @@ int a;
// 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_GRP249:![0-9]+]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221:![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_GRP249]]
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP249]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP249]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP221]]
// CHECK-32-NEXT: [[TOBOOL2:%.*]] = trunc i8 [[TMP9]] to i1
// CHECK-32-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL2]] to i8
-// CHECK-32-NEXT: store i8 [[FROMBOOL]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1, !llvm.access.group [[ACC_GRP249]]
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group [[ACC_GRP249]]
+// CHECK-32-NEXT: store i8 [[FROMBOOL]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
+// CHECK-32-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
+// CHECK-32-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP249]]
+// CHECK-32-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP221]]
// CHECK-32-NEXT: [[TOBOOL3:%.*]] = trunc i8 [[TMP17]] to i1
// CHECK-32-NEXT: [[TMP18:%.*]] = zext i1 [[TOBOOL3]] to i32
-// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP249]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP249]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP221]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP249]]
-// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP249]]
-// CHECK-32-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP249]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
-// CHECK-32-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP249]]
-// CHECK-32-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP249]]
+// CHECK-32-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
-// CHECK-32-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP249]]
+// CHECK-32-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
+// CHECK-32-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
-// CHECK-32-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP249]]
-// CHECK-32-NEXT: store i32 [[TMP27]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP249]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP250:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND10]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-NEXT: store i32 [[TMP27]], 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: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK-32: omp_if.else:
@@ -9425,7 +9462,7 @@ int a;
// 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 [[LOOP253:![0-9]+]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND11]], !llvm.loop [[LOOP225:![0-9]+]]
// CHECK-32: omp.inner.for.end28:
// CHECK-32-NEXT: br label [[OMP_IF_END]]
// CHECK-32: omp_if.end:
@@ -9481,24 +9518,24 @@ int a;
// 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_GRP255:![0-9]+]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP255]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227:![0-9]+]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP227]]
// 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_GRP255]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
// 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_GRP255]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP227]]
// 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_GRP255]]
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP255]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP227]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP255]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP256:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP228:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK-32: omp_if.else:
@@ -9526,7 +9563,7 @@ 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 [[LOOP258:![0-9]+]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND2]], !llvm.loop [[LOOP230:![0-9]+]]
// CHECK-32: omp.inner.for.end10:
// CHECK-32-NEXT: br label [[OMP_IF_END]]
// CHECK-32: omp_if.end:
@@ -9584,24 +9621,24 @@ int a;
// 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_GRP259:![0-9]+]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231:![0-9]+]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP231]]
// 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_GRP259]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231]]
// 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_GRP259]]
+// 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: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259]]
-// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP231]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP260:![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_IF_END:%.*]]
// CHECK-32: omp_if.else:
@@ -9629,7 +9666,7 @@ 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 [[LOOP262:![0-9]+]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND2]], !llvm.loop [[LOOP234:![0-9]+]]
// CHECK-32: omp.inner.for.end10:
// CHECK-32-NEXT: br label [[OMP_IF_END]]
// CHECK-32: omp_if.end:
@@ -9707,47 +9744,47 @@ 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_GRP263:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP235:![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_GRP263]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP263]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP263]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP263]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP235]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP263]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP263]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP263]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP263]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP263]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP263]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP263]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP263]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP264:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP236:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -9806,23 +9843,23 @@ 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_GRP266:![0-9]+]]
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP266]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP238:![0-9]+]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP238]]
// 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_GRP266]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP238]]
// 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_GRP266]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP238]]
// 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_GRP266]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP238]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP266]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP267:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP238]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP239:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -9896,47 +9933,47 @@ 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_GRP269:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP241:![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_GRP269]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP269]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP269]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP269]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP241]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP269]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP269]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP269]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP269]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP269]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP269]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP269]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP269]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP270:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP242:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -9984,24 +10021,24 @@ 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_GRP272:![0-9]+]]
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP272]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP244:![0-9]+]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP244]]
// 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_GRP272]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP244]]
// 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_GRP272]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP244]]
// 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_GRP272]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP272]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP244]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP244]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP272]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP273:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP244]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP245:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -10075,47 +10112,47 @@ 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_GRP275:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP247:![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_GRP275]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP275]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP275]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP275]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP247]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP275]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP275]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP275]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP275]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP275]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP275]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP275]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP275]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP276:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP248:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -10171,28 +10208,29 @@ 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_GRP278:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP278]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP250:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP250]]
// 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_GRP278]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP250]]
// 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_GRP278]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP250]]
// 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_GRP278]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP250]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP278]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP279:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP250]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP251:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// 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_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -10262,47 +10300,47 @@ 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_GRP281:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP253:![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_GRP281]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP281]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP281]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP281]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP253]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP281]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP281]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP281]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP281]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP281]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP281]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP281]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP281]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP282:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP254:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -10358,28 +10396,29 @@ 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_GRP284:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP284]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP256:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP256]]
// 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_GRP284]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP256]]
// 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_GRP284]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP256]]
// 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_GRP284]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP256]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP284]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP285:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP256]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP257:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// 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_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -10449,47 +10488,47 @@ 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_GRP287:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259:![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_GRP287]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP287]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP287]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP287]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP259]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP287]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP287]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP287]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP287]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP287]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP287]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP287]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP287]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP288:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP260:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -10545,28 +10584,29 @@ 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_GRP290:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP290]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP262:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP262]]
// 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_GRP290]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP262]]
// 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_GRP290]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP262]]
// 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_GRP290]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP262]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP290]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP291:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP262]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP263:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// 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_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -10636,47 +10676,47 @@ 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_GRP293:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP265:![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_GRP293]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP293]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP293]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP293]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP265]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP293]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP293]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP293]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP293]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP293]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP293]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP293]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP293]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP294:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP266:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -10732,28 +10772,29 @@ 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_GRP296:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP296]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP268:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP268]]
// 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_GRP296]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP268]]
// 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_GRP296]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP268]]
// 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_GRP296]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP268]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP296]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP297:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP268]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP269:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// 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_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -11455,28 +11496,29 @@ 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_GRP299:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP299]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP271:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP271]]
// 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_GRP299]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP271]]
// 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_GRP299]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP271]]
// 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_GRP299]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP271]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP299]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP300:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP271]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP272:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-NEXT: ret void
//
//
@@ -11628,28 +11670,29 @@ 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_GRP302:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP302]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP274:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP274]]
// 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_GRP302]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP274]]
// 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_GRP302]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP274]]
// 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_GRP302]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP274]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP302]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP303:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP274]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP275:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-NEXT: ret void
//
//
@@ -11801,28 +11844,29 @@ 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_GRP305:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP305]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP277:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP277]]
// 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_GRP305]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP277]]
// 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_GRP305]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP277]]
// 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_GRP305]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP277]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP305]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP306:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP277]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP278:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-NEXT: ret void
//
//
@@ -11974,28 +12018,29 @@ 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_GRP308:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP308]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP280:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP280]]
// 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_GRP308]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP280]]
// 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_GRP308]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP280]]
// 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_GRP308]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP280]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP308]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP309:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP280]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP281:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-NEXT: ret void
//
//
@@ -12059,47 +12104,47 @@ 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_GRP311:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP283:![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_GRP311]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP311]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP311]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP311]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP283]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP311]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP311]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP311]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP311]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP311]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP311]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP311]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP311]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP312:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP284:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -12147,24 +12192,24 @@ 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_GRP314:![0-9]+]]
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP314]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP286:![0-9]+]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP286]]
// 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_GRP314]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP286]]
// 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_GRP314]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP286]]
// 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_GRP314]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP314]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP286]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP286]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP314]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP315:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP286]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP287:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -12240,47 +12285,47 @@ 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_GRP317:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP289:![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_GRP317]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP317]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP317]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP317]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP289]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP317]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP317]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP317]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP317]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP317]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP317]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP317]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP317]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP318:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP290:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -12339,23 +12384,23 @@ 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_GRP320:![0-9]+]]
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP320]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP292:![0-9]+]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP292]]
// 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_GRP320]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP292]]
// 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_GRP320]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP292]]
// 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_GRP320]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP292]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP320]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP321:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP292]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP293:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -12430,27 +12475,27 @@ 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_GRP323:![0-9]+]]
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP323]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP295:![0-9]+]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP295]]
// 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_GRP323]]
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP323]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP295]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP295]]
// 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_GRP323]]
+// CHECK-32-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 4, !llvm.access.group [[ACC_GRP295]]
// 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_GRP323]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP323]]
+// CHECK-32-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4, !llvm.access.group [[ACC_GRP295]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP295]]
// 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_GRP323]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP323]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP295]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP295]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP323]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP324:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP295]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP296:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -12501,24 +12546,24 @@ 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_GRP326:![0-9]+]]
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP326]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP298:![0-9]+]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP298]]
// 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_GRP326]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP298]]
// 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_GRP326]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP298]]
// 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_GRP326]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP326]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP298]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP298]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP326]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP327:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP298]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP299:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -12612,47 +12657,47 @@ 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_GRP329:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP301:![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_GRP329]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP329]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP329]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP329]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP301]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP329]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP329]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP329]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP329]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP329]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP329]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP329]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP329]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP330:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP302:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -12708,28 +12753,29 @@ 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_GRP332:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP332]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP304:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP304]]
// 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_GRP332]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP304]]
// 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_GRP332]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP304]]
// 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_GRP332]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP304]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP332]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP333:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP304]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP305:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// 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_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -12799,47 +12845,47 @@ 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_GRP335:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP307:![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_GRP335]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP335]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP335]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP335]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP307]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP335]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP335]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP335]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP335]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP335]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP335]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP335]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP335]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP336:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP308:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -12895,28 +12941,29 @@ 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_GRP338:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP338]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP310:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP310]]
// 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_GRP338]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP310]]
// 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_GRP338]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP310]]
// 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_GRP338]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP310]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP338]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP339:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP310]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP311:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// 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_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -12986,47 +13033,47 @@ 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_GRP341:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP313:![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_GRP341]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP341]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP341]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP341]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP313]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP341]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP341]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP341]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP341]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP341]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP341]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP341]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP341]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP342:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP314:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -13082,28 +13129,29 @@ 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_GRP344:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP344]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP316:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP316]]
// 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_GRP344]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP316]]
// 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_GRP344]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP316]]
// 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_GRP344]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP316]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP344]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP345:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP316]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP317:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// 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_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -13173,47 +13221,47 @@ 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_GRP347:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP319:![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_GRP347]]
-// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP347]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
+// CHECK-32-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
-// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP347]]
+// CHECK-32-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
-// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP347]]
+// CHECK-32-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP319]]
// CHECK-32-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
-// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP347]]
-// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP347]]
-// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP347]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
-// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP347]]
-// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP347]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
-// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP347]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
+// CHECK-32-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
-// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP347]]
-// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP347]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP348:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP320:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -13269,28 +13317,29 @@ 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_GRP350:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP350]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP322:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP322]]
// 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_GRP350]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP322]]
// 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_GRP350]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP322]]
// 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_GRP350]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP322]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP350]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP351:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP322]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP323:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// 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_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -13954,28 +14003,29 @@ 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_GRP353:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP353]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP325:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP325]]
// 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_GRP353]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP325]]
// 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_GRP353]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP325]]
// 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_GRP353]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP325]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP353]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP354:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP325]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP326:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-NEXT: ret void
//
//
@@ -14127,28 +14177,29 @@ 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_GRP356:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP356]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP328:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP328]]
// 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_GRP356]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP328]]
// 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_GRP356]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP328]]
// 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_GRP356]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP328]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP356]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP357:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP328]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP329:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-NEXT: ret void
//
//
@@ -14300,28 +14351,29 @@ 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_GRP359:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP359]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP331:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP331]]
// 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_GRP359]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP331]]
// 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_GRP359]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP331]]
// 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_GRP359]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP331]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP359]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP360:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP331]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP332:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-NEXT: ret void
//
//
@@ -14473,28 +14525,29 @@ 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_GRP362:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP362]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP334:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP334]]
// 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_GRP362]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP334]]
// 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_GRP362]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP334]]
// 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_GRP362]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP334]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP362]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP363:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP334]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP335:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-NEXT: ret void
//
//
@@ -15151,28 +15204,29 @@ 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_GRP365:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP365]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP337:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP337]]
// 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_GRP365]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP337]]
// 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_GRP365]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP337]]
// 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_GRP365]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP337]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP365]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP366:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP337]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP338:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-NEXT: ret void
//
//
@@ -15324,28 +15378,29 @@ 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_GRP368:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP368]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP340:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP340]]
// 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_GRP368]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP340]]
// 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_GRP368]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP340]]
// 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_GRP368]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP340]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP368]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP369:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP340]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP341:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-NEXT: ret void
//
//
@@ -15497,28 +15552,29 @@ 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_GRP371:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP371]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP343:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP343]]
// 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_GRP371]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP343]]
// 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_GRP371]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP343]]
// 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_GRP371]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP343]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP371]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP372:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP343]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP344:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-NEXT: ret void
//
//
@@ -15670,28 +15726,29 @@ 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_GRP374:![0-9]+]]
-// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP374]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP346:![0-9]+]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP346]]
// 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_GRP374]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP346]]
// 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_GRP374]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP346]]
// 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_GRP374]]
+// CHECK-32-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP346]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP11]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP374]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP375:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP346]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP347:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-NEXT: ret void
//
//
@@ -16017,28 +16074,29 @@ 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_GRP377:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP377]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP349:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP349]]
// 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_GRP377]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP349]]
// 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_GRP377]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP349]]
// 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_GRP377]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP349]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP377]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP378:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP349]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP350:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -16091,28 +16149,29 @@ 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_GRP380:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP380]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP352:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP352]]
// 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_GRP380]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP352]]
// 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_GRP380]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP352]]
// 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_GRP380]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP352]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP380]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP381:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP352]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP353:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -16165,28 +16224,29 @@ 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_GRP383:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP383]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP355:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP355]]
// 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_GRP383]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP355]]
// 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_GRP383]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP355]]
// 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_GRP383]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP355]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP383]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP384:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP355]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP356:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -16239,28 +16299,29 @@ 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_GRP386:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP386]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP358:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP358]]
// 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_GRP386]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP358]]
// 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_GRP386]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP358]]
// 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_GRP386]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP358]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP386]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP387:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP358]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP359:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -16330,23 +16391,23 @@ 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_GRP389:![0-9]+]]
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP389]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP361:![0-9]+]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP361]]
// 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_GRP389]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP361]]
// 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_GRP389]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP361]]
// 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_GRP389]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP361]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP389]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP390:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP361]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP362:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -16426,23 +16487,23 @@ 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_GRP392:![0-9]+]]
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP392]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP364:![0-9]+]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP364]]
// 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_GRP392]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP364]]
// 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_GRP392]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP364]]
// 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_GRP392]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP364]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP392]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP393:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP364]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP365:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -16519,23 +16580,23 @@ 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_GRP395:![0-9]+]]
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP395]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP367:![0-9]+]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP367]]
// 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_GRP395]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP367]]
// 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_GRP395]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP367]]
// 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_GRP395]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP367]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP395]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP396:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP367]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP368:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -16610,28 +16671,29 @@ 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_GRP398:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP398]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP370:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP370]]
// 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_GRP398]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP370]]
// 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_GRP398]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP370]]
// 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_GRP398]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP370]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP398]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP399:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP370]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP371:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -16692,28 +16754,29 @@ 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_GRP401:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP401]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP373:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP373]]
// 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_GRP401]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP373]]
// 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_GRP401]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP373]]
// 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_GRP401]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP373]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP401]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP402:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP373]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP374:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -16774,28 +16837,29 @@ 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_GRP404:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP404]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP376:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP376]]
// 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_GRP404]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP376]]
// 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_GRP404]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP376]]
// 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_GRP404]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP376]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP404]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP405:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP376]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP377:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -16856,28 +16920,29 @@ 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_GRP407:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP407]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP379:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP379]]
// 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_GRP407]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP379]]
// 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_GRP407]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP379]]
// 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_GRP407]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP379]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP407]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP408:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP379]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP380:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -16938,29 +17003,30 @@ 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_GRP410:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP410]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP382:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP382]]
// 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_GRP410]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP382]]
// 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_GRP410]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP382]]
// 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_GRP410]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP382]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP410]]
-// CHECK-32-NEXT: call void @__kmpc_dispatch_fini_4(ptr @[[GLOB1]], i32 [[TMP1]]), !llvm.access.group [[ACC_GRP410]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP411:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP382]]
+// CHECK-32-NEXT: call void @__kmpc_dispatch_fini_4(ptr @[[GLOB1]], i32 [[TMP1]]), !llvm.access.group [[ACC_GRP382]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP383:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -17026,23 +17092,23 @@ 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_GRP413:![0-9]+]]
-// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP413]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP385:![0-9]+]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP385]]
// 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_GRP413]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP385]]
// 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_GRP413]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP385]]
// 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_GRP413]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP385]]
// CHECK-32-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
-// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP413]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP414:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP385]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP386:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32: omp.loop.exit:
@@ -17119,23 +17185,23 @@ 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_GRP416:![0-9]+]]
-// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP416]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP388:![0-9]+]]
+// CHECK-32-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP388]]
// 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_GRP416]]
+// CHECK-32-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP388]]
// 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_GRP416]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP388]]
// 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_GRP416]]
+// CHECK-32-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP388]]
// CHECK-32-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
-// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP416]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP417:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP388]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP389:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
@@ -17210,28 +17276,29 @@ 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_GRP419:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP419]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP391:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP391]]
// 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_GRP419]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP391]]
// 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_GRP419]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP391]]
// 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_GRP419]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP391]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP419]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP420:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP391]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP392:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -17292,28 +17359,29 @@ 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_GRP422:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP422]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP394:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP394]]
// 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_GRP422]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP394]]
// 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_GRP422]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP394]]
// 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_GRP422]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP394]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP422]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP423:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP394]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP395:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -17374,28 +17442,29 @@ 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_GRP425:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP425]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP397:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP397]]
// 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_GRP425]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP397]]
// 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_GRP425]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP397]]
// 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_GRP425]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP397]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP425]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP426:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP397]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP398:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -17456,28 +17525,29 @@ 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_GRP428:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP428]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP400:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP400]]
// 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_GRP428]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP400]]
// 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_GRP428]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP400]]
// 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_GRP428]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP400]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP428]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP429:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP400]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP401:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -17806,28 +17876,29 @@ 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_GRP431:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP431]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP403:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP403]]
// 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_GRP431]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP403]]
// 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_GRP431]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP403]]
// 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_GRP431]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP403]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP431]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP432:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP403]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP404:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -17880,28 +17951,29 @@ 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_GRP434:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP434]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP406:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP406]]
// 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_GRP434]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP406]]
// 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_GRP434]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP406]]
// 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_GRP434]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP406]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP434]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP435:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP406]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP407:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -17954,28 +18026,29 @@ 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_GRP437:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP437]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP409:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP409]]
// 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_GRP437]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP409]]
// 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_GRP437]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP409]]
// 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_GRP437]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP409]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP437]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP438:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP409]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP410:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -18028,28 +18101,29 @@ 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_GRP440:![0-9]+]]
-// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP440]]
+// CHECK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP412:![0-9]+]]
+// CHECK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP412]]
// 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_GRP440]]
+// CHECK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP412]]
// 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_GRP440]]
+// CHECK-32-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP412]]
// 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_GRP440]]
+// CHECK-32-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP412]]
// CHECK-32-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP7]], 1
-// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP440]]
-// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP441:![0-9]+]]
+// CHECK-32-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP412]]
+// CHECK-32-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP413:![0-9]+]]
// CHECK-32: omp.inner.for.end:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32: omp.dispatch.inc:
// CHECK-32-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32: omp.dispatch.end:
+// CHECK-32-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-32-NEXT: ret void
//
//
@@ -18129,58 +18203,58 @@ int a;
// 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_GRP249:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221:![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_GRP249]]
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP249]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP249]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP221]]
// CHECK-32-EX-NEXT: [[TOBOOL2:%.*]] = trunc i8 [[TMP9]] to i1
// CHECK-32-EX-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL2]] to i8
-// CHECK-32-EX-NEXT: store i8 [[FROMBOOL]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1, !llvm.access.group [[ACC_GRP249]]
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group [[ACC_GRP249]]
+// CHECK-32-EX-NEXT: store i8 [[FROMBOOL]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
+// CHECK-32-EX-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
+// CHECK-32-EX-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP249]]
+// CHECK-32-EX-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !llvm.access.group [[ACC_GRP221]]
// CHECK-32-EX-NEXT: [[TOBOOL3:%.*]] = trunc i8 [[TMP17]] to i1
// CHECK-32-EX-NEXT: [[TMP18:%.*]] = zext i1 [[TOBOOL3]] to i32
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP249]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP249]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
-// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP249]]
-// CHECK-32-EX-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP249]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-EX-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-EX-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
-// CHECK-32-EX-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP249]]
-// CHECK-32-EX-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP249]]
+// CHECK-32-EX-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-EX-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-EX-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
-// CHECK-32-EX-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP249]]
+// CHECK-32-EX-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-EX-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
+// CHECK-32-EX-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
// 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_GRP249]]
-// CHECK-32-EX-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP249]]
-// CHECK-32-EX-NEXT: store i32 [[TMP27]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP249]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP250:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND10]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-EX-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP221]]
+// CHECK-32-EX-NEXT: store i32 [[TMP27]], 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: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK-32-EX: omp_if.else:
@@ -18237,7 +18311,7 @@ int a;
// 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 [[LOOP253:![0-9]+]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND11]], !llvm.loop [[LOOP225:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end28:
// CHECK-32-EX-NEXT: br label [[OMP_IF_END]]
// CHECK-32-EX: omp_if.end:
@@ -18293,24 +18367,24 @@ int a;
// 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_GRP255:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP255]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP227]]
// 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_GRP255]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
// 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_GRP255]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP227]]
// 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_GRP255]]
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP255]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP227]]
// 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_GRP255]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP256:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP227]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP228:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK-32-EX: omp_if.else:
@@ -18338,7 +18412,7 @@ 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 [[LOOP258:![0-9]+]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND2]], !llvm.loop [[LOOP230:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end10:
// CHECK-32-EX-NEXT: br label [[OMP_IF_END]]
// CHECK-32-EX: omp_if.end:
@@ -18396,24 +18470,24 @@ int a;
// 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_GRP259:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP231]]
// 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_GRP259]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231]]
// 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_GRP259]]
+// 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: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259]]
-// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP231]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP231]]
// 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_GRP259]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP260:![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_IF_END:%.*]]
// CHECK-32-EX: omp_if.else:
@@ -18441,7 +18515,7 @@ 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 [[LOOP262:![0-9]+]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND2]], !llvm.loop [[LOOP234:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end10:
// CHECK-32-EX-NEXT: br label [[OMP_IF_END]]
// CHECK-32-EX: omp_if.end:
@@ -18519,47 +18593,47 @@ 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_GRP263:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP235:![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_GRP263]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP263]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP263]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP263]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP263]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP263]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP263]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP263]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP263]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
// 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_GRP263]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP263]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP263]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP264:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP235]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP236:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -18618,23 +18692,23 @@ 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_GRP266:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP266]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP238:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP238]]
// 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_GRP266]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP238]]
// 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_GRP266]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP238]]
// 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_GRP266]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP238]]
// 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_GRP266]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP267:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP238]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP239:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -18708,47 +18782,47 @@ 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_GRP269:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP241:![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_GRP269]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP269]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP269]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP269]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP269]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP269]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP269]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP269]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP269]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
// 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_GRP269]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP269]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP269]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP270:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP241]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP242:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -18796,24 +18870,24 @@ 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_GRP272:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP272]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP244:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP244]]
// 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_GRP272]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP244]]
// 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_GRP272]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP244]]
// 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_GRP272]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP272]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP244]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP244]]
// 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_GRP272]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP273:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP244]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP245:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -18887,47 +18961,47 @@ 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_GRP275:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP247:![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_GRP275]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP275]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP275]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP275]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP275]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP275]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP275]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP275]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP275]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
// 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_GRP275]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP275]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP275]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP276:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP247]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP248:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -18983,28 +19057,29 @@ 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_GRP278:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP278]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP250:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP250]]
// 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_GRP278]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP250]]
// 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_GRP278]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP250]]
// 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_GRP278]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP250]]
// 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_GRP278]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP279:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP250]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP251:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// 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_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -19074,47 +19149,47 @@ 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_GRP281:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP253:![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_GRP281]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP281]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP281]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP281]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP281]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP281]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP281]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP281]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP281]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
// 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_GRP281]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP281]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP281]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP282:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP253]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP254:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -19170,28 +19245,29 @@ 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_GRP284:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP284]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP256:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP256]]
// 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_GRP284]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP256]]
// 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_GRP284]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP256]]
// 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_GRP284]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP256]]
// 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_GRP284]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP285:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP256]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP257:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// 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_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -19261,47 +19337,47 @@ 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_GRP287:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259:![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_GRP287]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP287]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP287]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP287]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP287]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP287]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP287]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP287]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP287]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
// 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_GRP287]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP287]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP287]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP288:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP259]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP260:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -19357,28 +19433,29 @@ 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_GRP290:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP290]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP262:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP262]]
// 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_GRP290]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP262]]
// 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_GRP290]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP262]]
// 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_GRP290]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP262]]
// 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_GRP290]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP291:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP262]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP263:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// 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_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -19448,47 +19525,47 @@ 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_GRP293:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP265:![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_GRP293]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP293]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP293]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP293]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP293]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP293]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP293]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP293]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP293]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
// 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_GRP293]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP293]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP293]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP294:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP265]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP266:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -19544,28 +19621,29 @@ 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_GRP296:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP296]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP268:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP268]]
// 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_GRP296]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP268]]
// 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_GRP296]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP268]]
// 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_GRP296]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP268]]
// 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_GRP296]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP297:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP268]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP269:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// 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_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -20267,28 +20345,29 @@ 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_GRP299:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP299]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP271:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP271]]
// 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_GRP299]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP271]]
// 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_GRP299]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP271]]
// 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_GRP299]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP271]]
// 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_GRP299]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP300:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP271]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP272:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -20440,28 +20519,29 @@ 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_GRP302:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP302]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP274:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP274]]
// 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_GRP302]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP274]]
// 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_GRP302]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP274]]
// 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_GRP302]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP274]]
// 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_GRP302]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP303:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP274]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP275:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -20613,28 +20693,29 @@ 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_GRP305:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP305]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP277:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP277]]
// 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_GRP305]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP277]]
// 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_GRP305]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP277]]
// 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_GRP305]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP277]]
// 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_GRP305]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP306:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP277]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP278:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -20786,28 +20867,29 @@ 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_GRP308:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP308]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP280:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP280]]
// 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_GRP308]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP280]]
// 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_GRP308]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP280]]
// 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_GRP308]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP280]]
// 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_GRP308]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP309:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP280]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP281:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -20871,47 +20953,47 @@ 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_GRP311:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP283:![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_GRP311]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP311]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP311]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP311]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP311]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP311]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP311]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP311]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP311]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
// 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_GRP311]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP311]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP311]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP312:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP283]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP284:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -20959,24 +21041,24 @@ 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_GRP314:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP314]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP286:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP286]]
// 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_GRP314]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP286]]
// 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_GRP314]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP286]]
// 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_GRP314]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP314]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP286]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP286]]
// 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_GRP314]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP315:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP286]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP287:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -21052,47 +21134,47 @@ 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_GRP317:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP289:![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_GRP317]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP317]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP317]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP317]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP317]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP317]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP317]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP317]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP317]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
// 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_GRP317]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP317]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP317]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP318:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP289]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP290:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -21151,23 +21233,23 @@ 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_GRP320:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP320]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP292:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP292]]
// 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_GRP320]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP292]]
// 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_GRP320]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP292]]
// 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_GRP320]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP292]]
// 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_GRP320]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP321:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP292]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP293:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -21242,27 +21324,27 @@ 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_GRP323:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP323]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP295:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP295]]
// 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_GRP323]]
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP323]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP295]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP295]]
// 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_GRP323]]
+// CHECK-32-EX-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 4, !llvm.access.group [[ACC_GRP295]]
// 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_GRP323]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP323]]
+// CHECK-32-EX-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4, !llvm.access.group [[ACC_GRP295]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP295]]
// 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_GRP323]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP323]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP295]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP295]]
// 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_GRP323]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP324:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP295]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP296:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -21313,24 +21395,24 @@ 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_GRP326:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP326]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP298:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4, !llvm.access.group [[ACC_GRP298]]
// 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_GRP326]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP298]]
// 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_GRP326]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP298]]
// 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_GRP326]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP326]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP298]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP298]]
// 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_GRP326]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP327:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP298]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP299:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -21424,47 +21506,47 @@ 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_GRP329:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP301:![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_GRP329]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP329]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP329]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP329]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP329]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP329]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP329]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP329]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP329]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
// 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_GRP329]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP329]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP329]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP330:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP301]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP302:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -21520,28 +21602,29 @@ 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_GRP332:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP332]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP304:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP304]]
// 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_GRP332]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP304]]
// 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_GRP332]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP304]]
// 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_GRP332]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP304]]
// 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_GRP332]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP333:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP304]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP305:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// 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_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -21611,47 +21694,47 @@ 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_GRP335:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP307:![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_GRP335]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP335]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP335]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP335]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP335]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP335]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP335]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP335]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP335]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
// 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_GRP335]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP335]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP335]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP336:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP307]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP308:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -21707,28 +21790,29 @@ 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_GRP338:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP338]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP310:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP310]]
// 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_GRP338]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP310]]
// 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_GRP338]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP310]]
// 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_GRP338]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP310]]
// 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_GRP338]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP339:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP310]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP311:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// 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_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -21798,47 +21882,47 @@ 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_GRP341:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP313:![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_GRP341]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP341]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP341]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP341]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP341]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP341]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP341]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP341]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP341]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
// 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_GRP341]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP341]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP341]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP342:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP313]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP314:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -21894,28 +21978,29 @@ 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_GRP344:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP344]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP316:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP316]]
// 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_GRP344]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP316]]
// 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_GRP344]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP316]]
// 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_GRP344]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP316]]
// 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_GRP344]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP345:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP316]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP317:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// 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_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -21985,47 +22070,47 @@ 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_GRP347:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP319:![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_GRP347]]
-// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP347]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
+// CHECK-32-EX-NEXT: store ptr [[TMP9]], ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
-// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP347]]
+// CHECK-32-EX-NEXT: store ptr [[TMP11]], ptr [[TMP10]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-EX-NEXT: call void @__kmpc_parallel_51(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), !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
-// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP347]]
+// CHECK-32-EX-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-EX-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
-// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP347]]
-// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP347]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-EX-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-EX-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
-// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP347]]
-// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP347]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-EX-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-EX-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
-// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP347]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-EX-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
+// CHECK-32-EX-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
// 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_GRP347]]
-// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP347]]
-// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP347]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP348:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-EX-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-EX-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP319]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP320:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -22081,28 +22166,29 @@ 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_GRP350:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP350]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP322:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP322]]
// 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_GRP350]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP322]]
// 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_GRP350]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP322]]
// 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_GRP350]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP322]]
// 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_GRP350]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP351:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP322]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP323:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// 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_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -22766,28 +22852,29 @@ 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_GRP353:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP353]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP325:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP325]]
// 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_GRP353]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP325]]
// 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_GRP353]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP325]]
// 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_GRP353]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP325]]
// 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_GRP353]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP354:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP325]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP326:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -22939,28 +23026,29 @@ 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_GRP356:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP356]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP328:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP328]]
// 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_GRP356]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP328]]
// 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_GRP356]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP328]]
// 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_GRP356]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP328]]
// 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_GRP356]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP357:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP328]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP329:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -23112,28 +23200,29 @@ 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_GRP359:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP359]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP331:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP331]]
// 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_GRP359]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP331]]
// 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_GRP359]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP331]]
// 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_GRP359]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP331]]
// 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_GRP359]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP360:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP331]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP332:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -23285,28 +23374,29 @@ 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_GRP362:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP362]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP334:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP334]]
// 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_GRP362]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP334]]
// 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_GRP362]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP334]]
// 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_GRP362]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP334]]
// 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_GRP362]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP363:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP334]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP335:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -23963,28 +24053,29 @@ 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_GRP365:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP365]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP337:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP337]]
// 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_GRP365]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP337]]
// 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_GRP365]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP337]]
// 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_GRP365]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP337]]
// 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_GRP365]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP366:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP337]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP338:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -24136,28 +24227,29 @@ 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_GRP368:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP368]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP340:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP340]]
// 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_GRP368]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP340]]
// 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_GRP368]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP340]]
// 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_GRP368]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP340]]
// 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_GRP368]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP369:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP340]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP341:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -24309,28 +24401,29 @@ 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_GRP371:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP371]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP343:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP343]]
// 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_GRP371]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP343]]
// 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_GRP371]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP343]]
// 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_GRP371]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP343]]
// 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_GRP371]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP372:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP343]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP344:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -24482,28 +24575,29 @@ 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_GRP374:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP374]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP346:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP346]]
// 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_GRP374]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP346]]
// 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_GRP374]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP346]]
// 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_GRP374]]
+// CHECK-32-EX-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP346]]
// 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_GRP374]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP375:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP346]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP347:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP5]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -24829,28 +24923,29 @@ 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_GRP377:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP377]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP349:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP349]]
// 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_GRP377]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP349]]
// 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_GRP377]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP349]]
// 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_GRP377]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP349]]
// 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_GRP377]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP378:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP349]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP350:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -24903,28 +24998,29 @@ 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_GRP380:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP380]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP352:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP352]]
// 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_GRP380]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP352]]
// 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_GRP380]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP352]]
// 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_GRP380]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP352]]
// 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_GRP380]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP381:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP352]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP353:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -24977,28 +25073,29 @@ 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_GRP383:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP383]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP355:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP355]]
// 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_GRP383]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP355]]
// 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_GRP383]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP355]]
// 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_GRP383]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP355]]
// 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_GRP383]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP384:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP355]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP356:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -25051,28 +25148,29 @@ 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_GRP386:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP386]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP358:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP358]]
// 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_GRP386]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP358]]
// 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_GRP386]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP358]]
// 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_GRP386]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP358]]
// 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_GRP386]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP387:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP358]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP359:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -25142,23 +25240,23 @@ 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_GRP389:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP389]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP361:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP361]]
// 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_GRP389]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP361]]
// 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_GRP389]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP361]]
// 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_GRP389]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP361]]
// 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_GRP389]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP390:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP361]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP362:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -25238,23 +25336,23 @@ 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_GRP392:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP392]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP364:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP364]]
// 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_GRP392]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP364]]
// 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_GRP392]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP364]]
// 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_GRP392]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP364]]
// 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_GRP392]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP393:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP364]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP365:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -25331,23 +25429,23 @@ 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_GRP395:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP395]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP367:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP367]]
// 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_GRP395]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP367]]
// 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_GRP395]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP367]]
// 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_GRP395]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP367]]
// 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_GRP395]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP396:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP367]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP368:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -25422,28 +25520,29 @@ 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_GRP398:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP398]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP370:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP370]]
// 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_GRP398]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP370]]
// 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_GRP398]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP370]]
// 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_GRP398]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP370]]
// 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_GRP398]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP399:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP370]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP371:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -25504,28 +25603,29 @@ 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_GRP401:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP401]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP373:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP373]]
// 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_GRP401]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP373]]
// 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_GRP401]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP373]]
// 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_GRP401]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP373]]
// 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_GRP401]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP402:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP373]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP374:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -25586,28 +25686,29 @@ 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_GRP404:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP404]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP376:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP376]]
// 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_GRP404]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP376]]
// 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_GRP404]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP376]]
// 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_GRP404]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP376]]
// 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_GRP404]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP405:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP376]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP377:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -25668,28 +25769,29 @@ 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_GRP407:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP407]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP379:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP379]]
// 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_GRP407]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP379]]
// 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_GRP407]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP379]]
// 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_GRP407]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP379]]
// 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_GRP407]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP408:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP379]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP380:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -25750,29 +25852,30 @@ 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_GRP410:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP410]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP382:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP382]]
// 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_GRP410]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP382]]
// 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_GRP410]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP382]]
// 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_GRP410]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP382]]
// 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_GRP410]]
-// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_fini_4(ptr @[[GLOB1]], i32 [[TMP1]]), !llvm.access.group [[ACC_GRP410]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP411:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP382]]
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_fini_4(ptr @[[GLOB1]], i32 [[TMP1]]), !llvm.access.group [[ACC_GRP382]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP383:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -25838,23 +25941,23 @@ 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_GRP413:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP413]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP385:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP385]]
// 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_GRP413]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP385]]
// 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_GRP413]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP385]]
// 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_GRP413]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP385]]
// 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_GRP413]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP414:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP385]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP386:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK-32-EX: omp.loop.exit:
@@ -25931,23 +26034,23 @@ 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_GRP416:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP416]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP388:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP388]]
// 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_GRP416]]
+// CHECK-32-EX-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP388]]
// 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_GRP416]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP388]]
// 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_GRP416]]
+// CHECK-32-EX-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP388]]
// 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_GRP416]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP417:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP388]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP389:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
@@ -26022,28 +26125,29 @@ 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_GRP419:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP419]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP391:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP391]]
// 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_GRP419]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP391]]
// 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_GRP419]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP391]]
// 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_GRP419]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP391]]
// 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_GRP419]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP420:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP391]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP392:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -26104,28 +26208,29 @@ 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_GRP422:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP422]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP394:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP394]]
// 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_GRP422]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP394]]
// 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_GRP422]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP394]]
// 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_GRP422]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP394]]
// 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_GRP422]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP423:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP394]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP395:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -26186,28 +26291,29 @@ 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_GRP425:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP425]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP397:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP397]]
// 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_GRP425]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP397]]
// 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_GRP425]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP397]]
// 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_GRP425]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP397]]
// 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_GRP425]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP426:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP397]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP398:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -26268,28 +26374,29 @@ 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_GRP428:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP428]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP400:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP400]]
// 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_GRP428]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP400]]
// 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_GRP428]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP400]]
// 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_GRP428]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP400]]
// 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_GRP428]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP429:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP400]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP401:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// 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:%.*]]
@@ -26618,28 +26725,29 @@ 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_GRP431:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP431]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP403:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP403]]
// 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_GRP431]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP403]]
// 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_GRP431]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP403]]
// 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_GRP431]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP403]]
// 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_GRP431]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP432:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP403]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP404:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -26692,28 +26800,29 @@ 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_GRP434:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP434]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP406:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP406]]
// 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_GRP434]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP406]]
// 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_GRP434]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP406]]
// 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_GRP434]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP406]]
// 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_GRP434]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP435:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP406]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP407:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -26766,28 +26875,29 @@ 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_GRP437:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP437]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP409:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP409]]
// 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_GRP437]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP409]]
// 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_GRP437]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP409]]
// 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_GRP437]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP409]]
// 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_GRP437]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP438:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP409]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP410:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
//
@@ -26840,27 +26950,28 @@ 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_GRP440:![0-9]+]]
-// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP440]]
+// CHECK-32-EX-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP412:![0-9]+]]
+// CHECK-32-EX-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP412]]
// 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_GRP440]]
+// CHECK-32-EX-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP412]]
// 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_GRP440]]
+// CHECK-32-EX-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP412]]
// 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_GRP440]]
+// CHECK-32-EX-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP412]]
// 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_GRP440]]
-// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP441:![0-9]+]]
+// CHECK-32-EX-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP412]]
+// CHECK-32-EX-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP413:![0-9]+]]
// CHECK-32-EX: omp.inner.for.end:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK-32-EX: omp.dispatch.inc:
// CHECK-32-EX-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK-32-EX: omp.dispatch.end:
+// CHECK-32-EX-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK-32-EX-NEXT: ret void
//
diff --git a/clang/test/OpenMP/ordered_codegen.cpp b/clang/test/OpenMP/ordered_codegen.cpp
index 927b5e194e4e8..0a73eaefc9808 100644
--- a/clang/test/OpenMP/ordered_codegen.cpp
+++ b/clang/test/OpenMP/ordered_codegen.cpp
@@ -203,6 +203,7 @@ void foo_simd(int low, int up) {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP0]])
// CHECK1-NEXT: call void @__kmpc_barrier(ptr @[[GLOB2:[0-9]+]], i32 [[TMP0]])
// CHECK1-NEXT: ret void
//
@@ -285,6 +286,7 @@ void foo_simd(int low, int up) {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP0]])
// CHECK1-NEXT: call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK1-NEXT: ret void
//
@@ -415,6 +417,7 @@ void foo_simd(int low, int up) {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP0]])
// CHECK1-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1: omp.precond.end:
// CHECK1-NEXT: call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
@@ -516,6 +519,7 @@ void foo_simd(int low, int up) {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP0]])
// CHECK1-NEXT: call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK1-NEXT: ret void
//
@@ -671,6 +675,7 @@ void foo_simd(int low, int up) {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP0]])
// CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK1-NEXT: br i1 [[TMP35]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -805,8 +810,10 @@ void foo_simd(int low, int up) {
// CHECK1-IRBUILDER: omp.dispatch.inc:
// CHECK1-IRBUILDER-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1-IRBUILDER: omp.dispatch.end:
-// CHECK1-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM12:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
-// CHECK1-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4:[0-9]+]], i32 [[OMP_GLOBAL_THREAD_NUM12]])
+// CHECK1-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM12:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB3]])
+// CHECK1-IRBUILDER-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM12]])
+// CHECK1-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM13:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4:[0-9]+]], i32 [[OMP_GLOBAL_THREAD_NUM13]])
// CHECK1-IRBUILDER-NEXT: ret void
//
//
@@ -893,8 +900,10 @@ void foo_simd(int low, int up) {
// CHECK1-IRBUILDER: omp.dispatch.inc:
// CHECK1-IRBUILDER-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1-IRBUILDER: omp.dispatch.end:
-// CHECK1-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM11:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
-// CHECK1-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[OMP_GLOBAL_THREAD_NUM11]])
+// CHECK1-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM11:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB6]])
+// CHECK1-IRBUILDER-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM11]])
+// CHECK1-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM12:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[OMP_GLOBAL_THREAD_NUM12]])
// CHECK1-IRBUILDER-NEXT: ret void
//
//
@@ -1029,10 +1038,12 @@ void foo_simd(int low, int up) {
// CHECK1-IRBUILDER: omp.dispatch.inc:
// CHECK1-IRBUILDER-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1-IRBUILDER: omp.dispatch.end:
+// CHECK1-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM34:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB8]])
+// CHECK1-IRBUILDER-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM34]])
// CHECK1-IRBUILDER-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1-IRBUILDER: omp.precond.end:
-// CHECK1-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM34:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
-// CHECK1-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[OMP_GLOBAL_THREAD_NUM34]])
+// CHECK1-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM35:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[OMP_GLOBAL_THREAD_NUM35]])
// CHECK1-IRBUILDER-NEXT: ret void
//
//
@@ -1136,8 +1147,10 @@ void foo_simd(int low, int up) {
// CHECK1-IRBUILDER: omp.dispatch.inc:
// CHECK1-IRBUILDER-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1-IRBUILDER: omp.dispatch.end:
-// CHECK1-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM19:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
-// CHECK1-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[OMP_GLOBAL_THREAD_NUM19]])
+// CHECK1-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM19:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB10]])
+// CHECK1-IRBUILDER-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM19]])
+// CHECK1-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM20:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[OMP_GLOBAL_THREAD_NUM20]])
// CHECK1-IRBUILDER-NEXT: ret void
//
//
@@ -1298,6 +1311,8 @@ void foo_simd(int low, int up) {
// CHECK1-IRBUILDER: omp.dispatch.inc:
// CHECK1-IRBUILDER-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1-IRBUILDER: omp.dispatch.end:
+// CHECK1-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM43:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB12]])
+// CHECK1-IRBUILDER-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM43]])
// CHECK1-IRBUILDER-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-IRBUILDER-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
// CHECK1-IRBUILDER-NEXT: br i1 [[TMP34]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -1305,19 +1320,19 @@ void foo_simd(int low, int up) {
// CHECK1-IRBUILDER-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_18]], align 4
// CHECK1-IRBUILDER-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_19]], align 4
// CHECK1-IRBUILDER-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_18]], align 4
-// CHECK1-IRBUILDER-NEXT: [[SUB43:%.*]] = sub i32 [[TMP36]], [[TMP37]]
-// CHECK1-IRBUILDER-NEXT: [[SUB44:%.*]] = sub i32 [[SUB43]], 1
-// CHECK1-IRBUILDER-NEXT: [[ADD45:%.*]] = add i32 [[SUB44]], 1
-// CHECK1-IRBUILDER-NEXT: [[DIV46:%.*]] = udiv i32 [[ADD45]], 1
-// CHECK1-IRBUILDER-NEXT: [[MUL47:%.*]] = mul i32 [[DIV46]], 1
-// CHECK1-IRBUILDER-NEXT: [[ADD48:%.*]] = add i32 [[TMP35]], [[MUL47]]
-// CHECK1-IRBUILDER-NEXT: store i32 [[ADD48]], ptr [[I28]], align 4
+// CHECK1-IRBUILDER-NEXT: [[SUB44:%.*]] = sub i32 [[TMP36]], [[TMP37]]
+// CHECK1-IRBUILDER-NEXT: [[SUB45:%.*]] = sub i32 [[SUB44]], 1
+// CHECK1-IRBUILDER-NEXT: [[ADD46:%.*]] = add i32 [[SUB45]], 1
+// CHECK1-IRBUILDER-NEXT: [[DIV47:%.*]] = udiv i32 [[ADD46]], 1
+// CHECK1-IRBUILDER-NEXT: [[MUL48:%.*]] = mul i32 [[DIV47]], 1
+// CHECK1-IRBUILDER-NEXT: [[ADD49:%.*]] = add i32 [[TMP35]], [[MUL48]]
+// CHECK1-IRBUILDER-NEXT: store i32 [[ADD49]], ptr [[I28]], align 4
// CHECK1-IRBUILDER-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK1-IRBUILDER: .omp.final.done:
// CHECK1-IRBUILDER-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1-IRBUILDER: omp.precond.end:
-// CHECK1-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM49:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
-// CHECK1-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[OMP_GLOBAL_THREAD_NUM49]])
+// CHECK1-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM50:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[OMP_GLOBAL_THREAD_NUM50]])
// CHECK1-IRBUILDER-NEXT: ret void
//
//
@@ -1428,6 +1443,7 @@ void foo_simd(int low, int up) {
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP0]])
// CHECK3-NEXT: call void @__kmpc_barrier(ptr @[[GLOB2:[0-9]+]], i32 [[TMP0]])
// CHECK3-NEXT: ret void
//
@@ -1510,6 +1526,7 @@ void foo_simd(int low, int up) {
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP0]])
// CHECK3-NEXT: call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK3-NEXT: ret void
//
@@ -1640,6 +1657,7 @@ void foo_simd(int low, int up) {
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP0]])
// CHECK3-NEXT: br label [[OMP_PRECOND_END]]
// CHECK3: omp.precond.end:
// CHECK3-NEXT: call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
@@ -1741,6 +1759,7 @@ void foo_simd(int low, int up) {
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP0]])
// CHECK3-NEXT: call void @__kmpc_barrier(ptr @[[GLOB2]], i32 [[TMP0]])
// CHECK3-NEXT: ret void
//
@@ -1896,6 +1915,7 @@ void foo_simd(int low, int up) {
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[TMP0]])
// CHECK3-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK3-NEXT: br i1 [[TMP35]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -2030,8 +2050,10 @@ void foo_simd(int low, int up) {
// CHECK3-IRBUILDER: omp.dispatch.inc:
// CHECK3-IRBUILDER-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3-IRBUILDER: omp.dispatch.end:
-// CHECK3-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM12:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
-// CHECK3-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4:[0-9]+]], i32 [[OMP_GLOBAL_THREAD_NUM12]])
+// CHECK3-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM12:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB3]])
+// CHECK3-IRBUILDER-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM12]])
+// CHECK3-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM13:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4:[0-9]+]], i32 [[OMP_GLOBAL_THREAD_NUM13]])
// CHECK3-IRBUILDER-NEXT: ret void
//
//
@@ -2118,8 +2140,10 @@ void foo_simd(int low, int up) {
// CHECK3-IRBUILDER: omp.dispatch.inc:
// CHECK3-IRBUILDER-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3-IRBUILDER: omp.dispatch.end:
-// CHECK3-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM11:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
-// CHECK3-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[OMP_GLOBAL_THREAD_NUM11]])
+// CHECK3-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM11:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB6]])
+// CHECK3-IRBUILDER-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM11]])
+// CHECK3-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM12:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[OMP_GLOBAL_THREAD_NUM12]])
// CHECK3-IRBUILDER-NEXT: ret void
//
//
@@ -2254,10 +2278,12 @@ void foo_simd(int low, int up) {
// CHECK3-IRBUILDER: omp.dispatch.inc:
// CHECK3-IRBUILDER-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3-IRBUILDER: omp.dispatch.end:
+// CHECK3-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM34:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB8]])
+// CHECK3-IRBUILDER-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM34]])
// CHECK3-IRBUILDER-NEXT: br label [[OMP_PRECOND_END]]
// CHECK3-IRBUILDER: omp.precond.end:
-// CHECK3-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM34:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
-// CHECK3-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[OMP_GLOBAL_THREAD_NUM34]])
+// CHECK3-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM35:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[OMP_GLOBAL_THREAD_NUM35]])
// CHECK3-IRBUILDER-NEXT: ret void
//
//
@@ -2361,8 +2387,10 @@ void foo_simd(int low, int up) {
// CHECK3-IRBUILDER: omp.dispatch.inc:
// CHECK3-IRBUILDER-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3-IRBUILDER: omp.dispatch.end:
-// CHECK3-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM19:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
-// CHECK3-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[OMP_GLOBAL_THREAD_NUM19]])
+// CHECK3-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM19:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB10]])
+// CHECK3-IRBUILDER-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM19]])
+// CHECK3-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM20:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[OMP_GLOBAL_THREAD_NUM20]])
// CHECK3-IRBUILDER-NEXT: ret void
//
//
@@ -2523,6 +2551,8 @@ void foo_simd(int low, int up) {
// CHECK3-IRBUILDER: omp.dispatch.inc:
// CHECK3-IRBUILDER-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3-IRBUILDER: omp.dispatch.end:
+// CHECK3-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM43:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB12]])
+// CHECK3-IRBUILDER-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM43]])
// CHECK3-IRBUILDER-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-IRBUILDER-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
// CHECK3-IRBUILDER-NEXT: br i1 [[TMP34]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -2530,19 +2560,19 @@ void foo_simd(int low, int up) {
// CHECK3-IRBUILDER-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_18]], align 4
// CHECK3-IRBUILDER-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_19]], align 4
// CHECK3-IRBUILDER-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_18]], align 4
-// CHECK3-IRBUILDER-NEXT: [[SUB43:%.*]] = sub i32 [[TMP36]], [[TMP37]]
-// CHECK3-IRBUILDER-NEXT: [[SUB44:%.*]] = sub i32 [[SUB43]], 1
-// CHECK3-IRBUILDER-NEXT: [[ADD45:%.*]] = add i32 [[SUB44]], 1
-// CHECK3-IRBUILDER-NEXT: [[DIV46:%.*]] = udiv i32 [[ADD45]], 1
-// CHECK3-IRBUILDER-NEXT: [[MUL47:%.*]] = mul i32 [[DIV46]], 1
-// CHECK3-IRBUILDER-NEXT: [[ADD48:%.*]] = add i32 [[TMP35]], [[MUL47]]
-// CHECK3-IRBUILDER-NEXT: store i32 [[ADD48]], ptr [[I28]], align 4
+// CHECK3-IRBUILDER-NEXT: [[SUB44:%.*]] = sub i32 [[TMP36]], [[TMP37]]
+// CHECK3-IRBUILDER-NEXT: [[SUB45:%.*]] = sub i32 [[SUB44]], 1
+// CHECK3-IRBUILDER-NEXT: [[ADD46:%.*]] = add i32 [[SUB45]], 1
+// CHECK3-IRBUILDER-NEXT: [[DIV47:%.*]] = udiv i32 [[ADD46]], 1
+// CHECK3-IRBUILDER-NEXT: [[MUL48:%.*]] = mul i32 [[DIV47]], 1
+// CHECK3-IRBUILDER-NEXT: [[ADD49:%.*]] = add i32 [[TMP35]], [[MUL48]]
+// CHECK3-IRBUILDER-NEXT: store i32 [[ADD49]], ptr [[I28]], align 4
// CHECK3-IRBUILDER-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK3-IRBUILDER: .omp.final.done:
// CHECK3-IRBUILDER-NEXT: br label [[OMP_PRECOND_END]]
// CHECK3-IRBUILDER: omp.precond.end:
-// CHECK3-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM49:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
-// CHECK3-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[OMP_GLOBAL_THREAD_NUM49]])
+// CHECK3-IRBUILDER-NEXT: [[OMP_GLOBAL_THREAD_NUM50:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-IRBUILDER-NEXT: call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[OMP_GLOBAL_THREAD_NUM50]])
// CHECK3-IRBUILDER-NEXT: ret void
//
//
diff --git a/clang/test/OpenMP/parallel_for_codegen.cpp b/clang/test/OpenMP/parallel_for_codegen.cpp
index dcd4ccfa5dcb3..2dec32d71b91a 100644
--- a/clang/test/OpenMP/parallel_for_codegen.cpp
+++ b/clang/test/OpenMP/parallel_for_codegen.cpp
@@ -808,6 +808,7 @@ void range_for_collapsed() {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP5]])
// CHECK1-NEXT: ret void
//
//
@@ -910,6 +911,7 @@ void range_for_collapsed() {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP5]])
// CHECK1-NEXT: ret void
//
//
@@ -1065,6 +1067,9 @@ void range_for_collapsed() {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: [[TMP35:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP36]])
// CHECK1-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1: omp.precond.end:
// CHECK1-NEXT: ret void
@@ -1186,6 +1191,7 @@ void range_for_collapsed() {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP5]])
// CHECK1-NEXT: ret void
//
//
@@ -1292,7 +1298,7 @@ void range_for_collapsed() {
// CHECK1-NEXT: [[ADD:%.*]] = add i32 131071, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT: [[CALL:%.*]] = invoke noundef i32 @_Z3foov()
-// CHECK1-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
+// CHECK1-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK1: invoke.cont:
// CHECK1-NEXT: [[CONV:%.*]] = sitofp i32 [[CALL]] to float
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4
@@ -1337,7 +1343,7 @@ void range_for_collapsed() {
// CHECK1-NEXT: ret void
// CHECK1: terminate.lpad:
// CHECK1-NEXT: [[TMP25:%.*]] = landingpad { ptr, i32 }
-// CHECK1-NEXT: catch ptr null
+// CHECK1-NEXT: catch ptr null
// CHECK1-NEXT: [[TMP26:%.*]] = extractvalue { ptr, i32 } [[TMP25]], 0
// CHECK1-NEXT: call void @__clang_call_terminate(ptr [[TMP26]]) #[[ATTR7:[0-9]+]]
// CHECK1-NEXT: unreachable
@@ -1918,6 +1924,7 @@ void range_for_collapsed() {
// CHECK2: omp.dispatch.inc:
// CHECK2-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK2: omp.dispatch.end:
+// CHECK2-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP5]])
// CHECK2-NEXT: ret void
//
//
@@ -2020,6 +2027,7 @@ void range_for_collapsed() {
// CHECK2: omp.dispatch.inc:
// CHECK2-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK2: omp.dispatch.end:
+// CHECK2-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP5]])
// CHECK2-NEXT: ret void
//
//
@@ -2175,6 +2183,9 @@ void range_for_collapsed() {
// CHECK2: omp.dispatch.inc:
// CHECK2-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK2: omp.dispatch.end:
+// CHECK2-NEXT: [[TMP35:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK2-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
+// CHECK2-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP36]])
// CHECK2-NEXT: br label [[OMP_PRECOND_END]]
// CHECK2: omp.precond.end:
// CHECK2-NEXT: ret void
@@ -2296,6 +2307,7 @@ void range_for_collapsed() {
// CHECK2: omp.dispatch.inc:
// CHECK2-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK2: omp.dispatch.end:
+// CHECK2-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP5]])
// CHECK2-NEXT: ret void
//
//
@@ -2402,7 +2414,7 @@ void range_for_collapsed() {
// CHECK2-NEXT: [[ADD:%.*]] = add i32 131071, [[MUL]]
// CHECK2-NEXT: store i32 [[ADD]], ptr [[I]], align 4
// CHECK2-NEXT: [[CALL:%.*]] = invoke noundef i32 @_Z3foov()
-// CHECK2-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
+// CHECK2-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK2: invoke.cont:
// CHECK2-NEXT: [[CONV:%.*]] = sitofp i32 [[CALL]] to float
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4
@@ -2447,7 +2459,7 @@ void range_for_collapsed() {
// CHECK2-NEXT: ret void
// CHECK2: terminate.lpad:
// CHECK2-NEXT: [[TMP25:%.*]] = landingpad { ptr, i32 }
-// CHECK2-NEXT: catch ptr null
+// CHECK2-NEXT: catch ptr null
// CHECK2-NEXT: [[TMP26:%.*]] = extractvalue { ptr, i32 } [[TMP25]], 0
// CHECK2-NEXT: call void @__clang_call_terminate(ptr [[TMP26]]) #[[ATTR7:[0-9]+]]
// CHECK2-NEXT: unreachable
@@ -3028,6 +3040,7 @@ void range_for_collapsed() {
// CHECK5: omp.dispatch.inc:
// CHECK5-NEXT: br label [[OMP_DISPATCH_COND]], !dbg [[DBG53]], !llvm.loop [[LOOP60:![0-9]+]]
// CHECK5: omp.dispatch.end:
+// CHECK5-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB21]], i32 [[TMP5]]), !dbg [[DBG53]]
// CHECK5-NEXT: ret void, !dbg [[DBG57]]
//
//
@@ -3130,6 +3143,7 @@ void range_for_collapsed() {
// CHECK5: omp.dispatch.inc:
// CHECK5-NEXT: br label [[OMP_DISPATCH_COND]], !dbg [[DBG65]], !llvm.loop [[LOOP72:![0-9]+]]
// CHECK5: omp.dispatch.end:
+// CHECK5-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB23]], i32 [[TMP5]]), !dbg [[DBG65]]
// CHECK5-NEXT: ret void, !dbg [[DBG69]]
//
//
@@ -3285,6 +3299,9 @@ void range_for_collapsed() {
// CHECK5: omp.dispatch.inc:
// CHECK5-NEXT: br label [[OMP_DISPATCH_COND]], !dbg [[DBG79]], !llvm.loop [[LOOP87:![0-9]+]]
// CHECK5: omp.dispatch.end:
+// CHECK5-NEXT: [[TMP35:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8, !dbg [[DBG79]]
+// CHECK5-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4, !dbg [[DBG79]]
+// CHECK5-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB25]], i32 [[TMP36]]), !dbg [[DBG79]]
// CHECK5-NEXT: br label [[OMP_PRECOND_END]], !dbg [[DBG79]]
// CHECK5: omp.precond.end:
// CHECK5-NEXT: ret void, !dbg [[DBG84]]
@@ -3406,6 +3423,7 @@ void range_for_collapsed() {
// CHECK5: omp.dispatch.inc:
// CHECK5-NEXT: br label [[OMP_DISPATCH_COND]], !dbg [[DBG93]], !llvm.loop [[LOOP101:![0-9]+]]
// CHECK5: omp.dispatch.end:
+// CHECK5-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB27]], i32 [[TMP5]]), !dbg [[DBG93]]
// CHECK5-NEXT: ret void, !dbg [[DBG98]]
//
//
@@ -3512,7 +3530,7 @@ void range_for_collapsed() {
// CHECK5-NEXT: [[ADD:%.*]] = add i32 131071, [[MUL]], !dbg [[DBG110]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !dbg [[DBG110]]
// CHECK5-NEXT: [[CALL:%.*]] = invoke noundef i32 @_Z3foov()
-// CHECK5-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]], !dbg [[DBG111:![0-9]+]]
+// CHECK5-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]], !dbg [[DBG111:![0-9]+]]
// CHECK5: invoke.cont:
// CHECK5-NEXT: [[CONV:%.*]] = sitofp i32 [[CALL]] to float, !dbg [[DBG111]]
// CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4, !dbg [[DBG111]]
@@ -3557,7 +3575,7 @@ void range_for_collapsed() {
// CHECK5-NEXT: ret void, !dbg [[DBG111]]
// CHECK5: terminate.lpad:
// CHECK5-NEXT: [[TMP25:%.*]] = landingpad { ptr, i32 }
-// CHECK5-NEXT: catch ptr null, !dbg [[DBG111]]
+// CHECK5-NEXT: catch ptr null, !dbg [[DBG111]]
// CHECK5-NEXT: [[TMP26:%.*]] = extractvalue { ptr, i32 } [[TMP25]], 0, !dbg [[DBG111]]
// CHECK5-NEXT: call void @__clang_call_terminate(ptr [[TMP26]]) #[[ATTR7:[0-9]+]], !dbg [[DBG111]]
// CHECK5-NEXT: unreachable, !dbg [[DBG111]]
@@ -4138,6 +4156,7 @@ void range_for_collapsed() {
// CHECK6: omp.dispatch.inc:
// CHECK6-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK6: omp.dispatch.end:
+// CHECK6-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP5]])
// CHECK6-NEXT: ret void
//
//
@@ -4240,6 +4259,7 @@ void range_for_collapsed() {
// CHECK6: omp.dispatch.inc:
// CHECK6-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK6: omp.dispatch.end:
+// CHECK6-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP5]])
// CHECK6-NEXT: ret void
//
//
@@ -4395,6 +4415,9 @@ void range_for_collapsed() {
// CHECK6: omp.dispatch.inc:
// CHECK6-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK6: omp.dispatch.end:
+// CHECK6-NEXT: [[TMP35:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK6-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
+// CHECK6-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP36]])
// CHECK6-NEXT: br label [[OMP_PRECOND_END]]
// CHECK6: omp.precond.end:
// CHECK6-NEXT: ret void
@@ -4516,6 +4539,7 @@ void range_for_collapsed() {
// CHECK6: omp.dispatch.inc:
// CHECK6-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK6: omp.dispatch.end:
+// CHECK6-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP5]])
// CHECK6-NEXT: ret void
//
//
diff --git a/clang/test/OpenMP/target_parallel_for_codegen.cpp b/clang/test/OpenMP/target_parallel_for_codegen.cpp
index 73504cf7a8024..ebdff9e8cf8a5 100644
--- a/clang/test/OpenMP/target_parallel_for_codegen.cpp
+++ b/clang/test/OpenMP/target_parallel_for_codegen.cpp
@@ -825,6 +825,7 @@ int bar(int n){
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP2]])
// 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_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
@@ -2488,6 +2489,7 @@ int bar(int n){
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
// CHECK3-NEXT: br i1 [[TMP14]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
@@ -5715,6 +5717,7 @@ int bar(int n){
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP2]])
// CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK17-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK17-NEXT: br i1 [[TMP13]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
@@ -7378,6 +7381,7 @@ int bar(int n){
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK19-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK19-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
// CHECK19-NEXT: br i1 [[TMP14]], label [[DOTOMP_LINEAR_PU:%.*]], label [[DOTOMP_LINEAR_PU_DONE:%.*]]
diff --git a/clang/test/OpenMP/target_parallel_for_simd_codegen.cpp b/clang/test/OpenMP/target_parallel_for_simd_codegen.cpp
index 8009e06666c82..bc8dcbc21eba0 100644
--- a/clang/test/OpenMP/target_parallel_for_simd_codegen.cpp
+++ b/clang/test/OpenMP/target_parallel_for_simd_codegen.cpp
@@ -846,6 +846,7 @@ int bar(int n){
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP2]])
// 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:%.*]]
@@ -2464,6 +2465,7 @@ int bar(int n){
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
// CHECK3-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -4088,6 +4090,7 @@ int bar(int n){
// CHECK5: omp.dispatch.inc:
// CHECK5-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK5: omp.dispatch.end:
+// CHECK5-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP2]])
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK5-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -5813,6 +5816,7 @@ int bar(int n){
// CHECK7: omp.dispatch.inc:
// CHECK7-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK7: omp.dispatch.end:
+// CHECK7-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB2]], i32 [[TMP3]])
// CHECK7-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK7-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
// CHECK7-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
diff --git a/clang/test/OpenMP/target_teams_distribute_parallel_for_schedule_codegen.cpp b/clang/test/OpenMP/target_teams_distribute_parallel_for_schedule_codegen.cpp
index fb0b2893d9529..5ecabd022b79d 100644
--- a/clang/test/OpenMP/target_teams_distribute_parallel_for_schedule_codegen.cpp
+++ b/clang/test/OpenMP/target_teams_distribute_parallel_for_schedule_codegen.cpp
@@ -1069,6 +1069,7 @@ int main (int argc, char **argv) {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK1-NEXT: ret void
//
//
@@ -1218,6 +1219,7 @@ int main (int argc, char **argv) {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK1-NEXT: ret void
//
//
@@ -2074,6 +2076,7 @@ int main (int argc, char **argv) {
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK3-NEXT: ret void
//
//
@@ -2218,6 +2221,7 @@ int main (int argc, char **argv) {
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK3-NEXT: ret void
//
//
@@ -3096,6 +3100,7 @@ int main (int argc, char **argv) {
// CHECK5: omp.dispatch.inc:
// CHECK5-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK5: omp.dispatch.end:
+// CHECK5-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK5-NEXT: ret void
//
//
@@ -3245,6 +3250,7 @@ int main (int argc, char **argv) {
// CHECK5: omp.dispatch.inc:
// CHECK5-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK5: omp.dispatch.end:
+// CHECK5-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK5-NEXT: ret void
//
//
@@ -4101,6 +4107,7 @@ int main (int argc, char **argv) {
// CHECK7: omp.dispatch.inc:
// CHECK7-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK7: omp.dispatch.end:
+// CHECK7-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK7-NEXT: ret void
//
//
@@ -4245,6 +4252,7 @@ int main (int argc, char **argv) {
// CHECK7: omp.dispatch.inc:
// CHECK7-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK7: omp.dispatch.end:
+// CHECK7-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK7-NEXT: ret void
//
//
@@ -5602,6 +5610,9 @@ int main (int argc, char **argv) {
// CHECK13: omp.dispatch.inc:
// CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK13: omp.dispatch.end:
+// CHECK13-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK13-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
+// CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP22]])
// CHECK13-NEXT: br label [[OMP_PRECOND_END]]
// CHECK13: omp.precond.end:
// CHECK13-NEXT: ret void
@@ -5830,6 +5841,9 @@ int main (int argc, char **argv) {
// CHECK13: omp.dispatch.inc:
// CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK13: omp.dispatch.end:
+// CHECK13-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK13-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
+// CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
// CHECK13-NEXT: br label [[OMP_PRECOND_END]]
// CHECK13: omp.precond.end:
// CHECK13-NEXT: ret void
@@ -6733,6 +6747,7 @@ int main (int argc, char **argv) {
// CHECK13: omp.dispatch.inc:
// CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK13: omp.dispatch.end:
+// CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK13-NEXT: ret void
//
//
@@ -6896,6 +6911,7 @@ int main (int argc, char **argv) {
// CHECK13: omp.dispatch.inc:
// CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK13: omp.dispatch.end:
+// CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
// CHECK13-NEXT: ret void
//
//
@@ -8237,6 +8253,9 @@ int main (int argc, char **argv) {
// CHECK15: omp.dispatch.inc:
// CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK15: omp.dispatch.end:
+// CHECK15-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK15-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
+// CHECK15-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP22]])
// CHECK15-NEXT: br label [[OMP_PRECOND_END]]
// CHECK15: omp.precond.end:
// CHECK15-NEXT: ret void
@@ -8460,6 +8479,9 @@ int main (int argc, char **argv) {
// CHECK15: omp.dispatch.inc:
// CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK15: omp.dispatch.end:
+// CHECK15-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK15-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
+// CHECK15-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
// CHECK15-NEXT: br label [[OMP_PRECOND_END]]
// CHECK15: omp.precond.end:
// CHECK15-NEXT: ret void
@@ -9341,6 +9363,7 @@ int main (int argc, char **argv) {
// CHECK15: omp.dispatch.inc:
// CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK15: omp.dispatch.end:
+// CHECK15-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK15-NEXT: ret void
//
//
@@ -9499,6 +9522,7 @@ int main (int argc, char **argv) {
// CHECK15: omp.dispatch.inc:
// CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK15: omp.dispatch.end:
+// CHECK15-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
// CHECK15-NEXT: ret void
//
//
@@ -10856,6 +10880,9 @@ int main (int argc, char **argv) {
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
+// CHECK17-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK17-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP22]])
// CHECK17-NEXT: br label [[OMP_PRECOND_END]]
// CHECK17: omp.precond.end:
// CHECK17-NEXT: ret void
@@ -11084,6 +11111,9 @@ int main (int argc, char **argv) {
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
+// CHECK17-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK17-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
// CHECK17-NEXT: br label [[OMP_PRECOND_END]]
// CHECK17: omp.precond.end:
// CHECK17-NEXT: ret void
@@ -11987,6 +12017,7 @@ int main (int argc, char **argv) {
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK17-NEXT: ret void
//
//
@@ -12150,6 +12181,7 @@ int main (int argc, char **argv) {
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
// CHECK17-NEXT: ret void
//
//
@@ -13491,6 +13523,9 @@ int main (int argc, char **argv) {
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
+// CHECK19-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK19-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP22]])
// CHECK19-NEXT: br label [[OMP_PRECOND_END]]
// CHECK19: omp.precond.end:
// CHECK19-NEXT: ret void
@@ -13714,6 +13749,9 @@ int main (int argc, char **argv) {
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
+// CHECK19-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK19-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
// CHECK19-NEXT: br label [[OMP_PRECOND_END]]
// CHECK19: omp.precond.end:
// CHECK19-NEXT: ret void
@@ -14595,6 +14633,7 @@ int main (int argc, char **argv) {
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK19-NEXT: ret void
//
//
@@ -14753,5 +14792,6 @@ int main (int argc, char **argv) {
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
// CHECK19-NEXT: ret void
//
diff --git a/clang/test/OpenMP/target_teams_distribute_parallel_for_simd_schedule_codegen.cpp b/clang/test/OpenMP/target_teams_distribute_parallel_for_simd_schedule_codegen.cpp
index b92dd14de108d..35f236722add8 100644
--- a/clang/test/OpenMP/target_teams_distribute_parallel_for_simd_schedule_codegen.cpp
+++ b/clang/test/OpenMP/target_teams_distribute_parallel_for_simd_schedule_codegen.cpp
@@ -1118,6 +1118,7 @@ int main (int argc, char **argv) {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -1281,6 +1282,7 @@ int main (int argc, char **argv) {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -2193,6 +2195,7 @@ int main (int argc, char **argv) {
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -2351,6 +2354,7 @@ int main (int argc, char **argv) {
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -3285,6 +3289,7 @@ int main (int argc, char **argv) {
// CHECK5: omp.dispatch.inc:
// CHECK5-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK5: omp.dispatch.end:
+// CHECK5-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK5-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -3448,6 +3453,7 @@ int main (int argc, char **argv) {
// CHECK5: omp.dispatch.inc:
// CHECK5-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK5: omp.dispatch.end:
+// CHECK5-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK5-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -4360,6 +4366,7 @@ int main (int argc, char **argv) {
// CHECK7: omp.dispatch.inc:
// CHECK7-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK7: omp.dispatch.end:
+// CHECK7-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK7-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK7-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK7-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -4518,6 +4525,7 @@ int main (int argc, char **argv) {
// CHECK7: omp.dispatch.inc:
// CHECK7-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK7: omp.dispatch.end:
+// CHECK7-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK7-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK7-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK7-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -6351,12 +6359,15 @@ int main (int argc, char **argv) {
// CHECK13: omp.dispatch.inc:
// CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK13: omp.dispatch.end:
-// CHECK13-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK13-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
-// CHECK13-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK13-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK13-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
+// CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP22]])
+// CHECK13-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK13-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
+// CHECK13-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK13: .omp.final.then:
-// CHECK13-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
-// CHECK13-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP23]], 0
+// CHECK13-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK13-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP25]], 0
// CHECK13-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1
// CHECK13-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1
// CHECK13-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]]
@@ -6603,12 +6614,15 @@ int main (int argc, char **argv) {
// CHECK13: omp.dispatch.inc:
// CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK13: omp.dispatch.end:
-// CHECK13-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK13-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
-// CHECK13-NEXT: br i1 [[TMP23]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK13-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK13-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
+// CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
+// CHECK13-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK13-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
+// CHECK13-NEXT: br i1 [[TMP25]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK13: .omp.final.then:
-// CHECK13-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK13-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP24]], 0
+// CHECK13-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK13-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK13-NEXT: [[DIV9:%.*]] = sdiv i32 [[SUB8]], 1
// CHECK13-NEXT: [[MUL10:%.*]] = mul nsw i32 [[DIV9]], 1
// CHECK13-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
@@ -7567,6 +7581,7 @@ int main (int argc, char **argv) {
// CHECK13: omp.dispatch.inc:
// CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK13: omp.dispatch.end:
+// CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK13-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK13-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK13-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -7744,6 +7759,7 @@ int main (int argc, char **argv) {
// CHECK13: omp.dispatch.inc:
// CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK13: omp.dispatch.end:
+// CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
// CHECK13-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK13-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK13-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -9176,12 +9192,15 @@ int main (int argc, char **argv) {
// CHECK15: omp.dispatch.inc:
// CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK15: omp.dispatch.end:
-// CHECK15-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK15-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
-// CHECK15-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK15-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK15-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
+// CHECK15-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP22]])
+// CHECK15-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK15-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
+// CHECK15-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK15: .omp.final.then:
-// CHECK15-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
-// CHECK15-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP23]], 0
+// CHECK15-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK15-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP25]], 0
// CHECK15-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK15-NEXT: [[MUL8:%.*]] = mul nsw i32 [[DIV7]], 1
// CHECK15-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL8]]
@@ -9423,12 +9442,15 @@ int main (int argc, char **argv) {
// CHECK15: omp.dispatch.inc:
// CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK15: omp.dispatch.end:
-// CHECK15-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK15-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
-// CHECK15-NEXT: br i1 [[TMP23]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK15-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK15-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
+// CHECK15-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
+// CHECK15-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK15-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
+// CHECK15-NEXT: br i1 [[TMP25]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK15: .omp.final.then:
-// CHECK15-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK15-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP24]], 0
+// CHECK15-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK15-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK15-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1
// CHECK15-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1
// CHECK15-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]]
@@ -10365,6 +10387,7 @@ int main (int argc, char **argv) {
// CHECK15: omp.dispatch.inc:
// CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK15: omp.dispatch.end:
+// CHECK15-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK15-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK15-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK15-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -10537,6 +10560,7 @@ int main (int argc, char **argv) {
// CHECK15: omp.dispatch.inc:
// CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK15: omp.dispatch.end:
+// CHECK15-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
// CHECK15-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK15-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK15-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -11985,12 +12009,15 @@ int main (int argc, char **argv) {
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
-// CHECK17-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK17-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
-// CHECK17-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK17-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK17-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP22]])
+// CHECK17-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK17-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
+// CHECK17-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK17: .omp.final.then:
-// CHECK17-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
-// CHECK17-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP23]], 0
+// CHECK17-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK17-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP25]], 0
// CHECK17-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1
// CHECK17-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1
// CHECK17-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]]
@@ -12237,12 +12264,15 @@ int main (int argc, char **argv) {
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
-// CHECK17-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK17-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
-// CHECK17-NEXT: br i1 [[TMP23]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK17-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK17-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
+// CHECK17-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK17-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
+// CHECK17-NEXT: br i1 [[TMP25]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK17: .omp.final.then:
-// CHECK17-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK17-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP24]], 0
+// CHECK17-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK17-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK17-NEXT: [[DIV9:%.*]] = sdiv i32 [[SUB8]], 1
// CHECK17-NEXT: [[MUL10:%.*]] = mul nsw i32 [[DIV9]], 1
// CHECK17-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
@@ -13201,6 +13231,7 @@ int main (int argc, char **argv) {
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK17-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK17-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK17-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -13378,6 +13409,7 @@ int main (int argc, char **argv) {
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
// CHECK17-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK17-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK17-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -14810,12 +14842,15 @@ int main (int argc, char **argv) {
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
-// CHECK19-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK19-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
-// CHECK19-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK19-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK19-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP22]])
+// CHECK19-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK19-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
+// CHECK19-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK19: .omp.final.then:
-// CHECK19-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
-// CHECK19-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP23]], 0
+// CHECK19-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK19-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP25]], 0
// CHECK19-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK19-NEXT: [[MUL8:%.*]] = mul nsw i32 [[DIV7]], 1
// CHECK19-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL8]]
@@ -15057,12 +15092,15 @@ int main (int argc, char **argv) {
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
-// CHECK19-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK19-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
-// CHECK19-NEXT: br i1 [[TMP23]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK19-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK19-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
+// CHECK19-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK19-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
+// CHECK19-NEXT: br i1 [[TMP25]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK19: .omp.final.then:
-// CHECK19-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK19-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP24]], 0
+// CHECK19-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK19-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK19-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1
// CHECK19-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1
// CHECK19-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]]
@@ -15999,6 +16037,7 @@ int main (int argc, char **argv) {
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK19-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK19-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK19-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -16171,6 +16210,7 @@ int main (int argc, char **argv) {
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
// CHECK19-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK19-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK19-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
diff --git a/clang/test/OpenMP/teams_distribute_parallel_for_schedule_codegen.cpp b/clang/test/OpenMP/teams_distribute_parallel_for_schedule_codegen.cpp
index 2dd2302c0f2cf..88769d69e1a18 100644
--- a/clang/test/OpenMP/teams_distribute_parallel_for_schedule_codegen.cpp
+++ b/clang/test/OpenMP/teams_distribute_parallel_for_schedule_codegen.cpp
@@ -1083,6 +1083,7 @@ int main (int argc, char **argv) {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK1-NEXT: ret void
//
//
@@ -1232,6 +1233,7 @@ int main (int argc, char **argv) {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK1-NEXT: ret void
//
//
@@ -2088,6 +2090,7 @@ int main (int argc, char **argv) {
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK3-NEXT: ret void
//
//
@@ -2232,6 +2235,7 @@ int main (int argc, char **argv) {
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
+// CHECK3-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK3-NEXT: ret void
//
//
@@ -3110,6 +3114,7 @@ int main (int argc, char **argv) {
// CHECK5: omp.dispatch.inc:
// CHECK5-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK5: omp.dispatch.end:
+// CHECK5-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK5-NEXT: ret void
//
//
@@ -3259,6 +3264,7 @@ int main (int argc, char **argv) {
// CHECK5: omp.dispatch.inc:
// CHECK5-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK5: omp.dispatch.end:
+// CHECK5-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK5-NEXT: ret void
//
//
@@ -4115,6 +4121,7 @@ int main (int argc, char **argv) {
// CHECK7: omp.dispatch.inc:
// CHECK7-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK7: omp.dispatch.end:
+// CHECK7-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK7-NEXT: ret void
//
//
@@ -4259,6 +4266,7 @@ int main (int argc, char **argv) {
// CHECK7: omp.dispatch.inc:
// CHECK7-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK7: omp.dispatch.end:
+// CHECK7-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK7-NEXT: ret void
//
//
@@ -5589,6 +5597,9 @@ int main (int argc, char **argv) {
// CHECK13: omp.dispatch.inc:
// CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK13: omp.dispatch.end:
+// CHECK13-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK13-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
+// CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
// CHECK13-NEXT: br label [[OMP_PRECOND_END]]
// CHECK13: omp.precond.end:
// CHECK13-NEXT: ret void
@@ -5814,6 +5825,9 @@ int main (int argc, char **argv) {
// CHECK13: omp.dispatch.inc:
// CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK13: omp.dispatch.end:
+// CHECK13-NEXT: [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK13-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
+// CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP24]])
// CHECK13-NEXT: br label [[OMP_PRECOND_END]]
// CHECK13: omp.precond.end:
// CHECK13-NEXT: ret void
@@ -6714,6 +6728,7 @@ int main (int argc, char **argv) {
// CHECK13: omp.dispatch.inc:
// CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK13: omp.dispatch.end:
+// CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK13-NEXT: ret void
//
//
@@ -6880,6 +6895,7 @@ int main (int argc, char **argv) {
// CHECK13: omp.dispatch.inc:
// CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK13: omp.dispatch.end:
+// CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
// CHECK13-NEXT: ret void
//
//
@@ -8194,6 +8210,9 @@ int main (int argc, char **argv) {
// CHECK15: omp.dispatch.inc:
// CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK15: omp.dispatch.end:
+// CHECK15-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK15-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
+// CHECK15-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
// CHECK15-NEXT: br label [[OMP_PRECOND_END]]
// CHECK15: omp.precond.end:
// CHECK15-NEXT: ret void
@@ -8414,6 +8433,9 @@ int main (int argc, char **argv) {
// CHECK15: omp.dispatch.inc:
// CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK15: omp.dispatch.end:
+// CHECK15-NEXT: [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK15-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
+// CHECK15-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP24]])
// CHECK15-NEXT: br label [[OMP_PRECOND_END]]
// CHECK15: omp.precond.end:
// CHECK15-NEXT: ret void
@@ -9292,6 +9314,7 @@ int main (int argc, char **argv) {
// CHECK15: omp.dispatch.inc:
// CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK15: omp.dispatch.end:
+// CHECK15-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK15-NEXT: ret void
//
//
@@ -9453,6 +9476,7 @@ int main (int argc, char **argv) {
// CHECK15: omp.dispatch.inc:
// CHECK15-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK15: omp.dispatch.end:
+// CHECK15-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
// CHECK15-NEXT: ret void
//
//
@@ -10783,6 +10807,9 @@ int main (int argc, char **argv) {
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
+// CHECK17-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK17-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
// CHECK17-NEXT: br label [[OMP_PRECOND_END]]
// CHECK17: omp.precond.end:
// CHECK17-NEXT: ret void
@@ -11008,6 +11035,9 @@ int main (int argc, char **argv) {
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
+// CHECK17-NEXT: [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK17-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP24]])
// CHECK17-NEXT: br label [[OMP_PRECOND_END]]
// CHECK17: omp.precond.end:
// CHECK17-NEXT: ret void
@@ -11908,6 +11938,7 @@ int main (int argc, char **argv) {
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK17-NEXT: ret void
//
//
@@ -12074,6 +12105,7 @@ int main (int argc, char **argv) {
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
// CHECK17-NEXT: ret void
//
//
@@ -13388,6 +13420,9 @@ int main (int argc, char **argv) {
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
+// CHECK19-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK19-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
// CHECK19-NEXT: br label [[OMP_PRECOND_END]]
// CHECK19: omp.precond.end:
// CHECK19-NEXT: ret void
@@ -13608,6 +13643,9 @@ int main (int argc, char **argv) {
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
+// CHECK19-NEXT: [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK19-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP24]])
// CHECK19-NEXT: br label [[OMP_PRECOND_END]]
// CHECK19: omp.precond.end:
// CHECK19-NEXT: ret void
@@ -14486,6 +14524,7 @@ int main (int argc, char **argv) {
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK19-NEXT: ret void
//
//
@@ -14647,5 +14686,6 @@ int main (int argc, char **argv) {
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
// CHECK19-NEXT: ret void
//
diff --git a/clang/test/OpenMP/teams_distribute_parallel_for_simd_schedule_codegen.cpp b/clang/test/OpenMP/teams_distribute_parallel_for_simd_schedule_codegen.cpp
index 3d40ac12e5ec6..e863dea3ebd8f 100644
--- a/clang/test/OpenMP/teams_distribute_parallel_for_simd_schedule_codegen.cpp
+++ b/clang/test/OpenMP/teams_distribute_parallel_for_simd_schedule_codegen.cpp
@@ -1142,6 +1142,7 @@ int main (int argc, char **argv) {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -1305,6 +1306,7 @@ int main (int argc, char **argv) {
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
+// CHECK1-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -2239,6 +2241,7 @@ int main (int argc, char **argv) {
// CHECK2: omp.dispatch.inc:
// CHECK2-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK2: omp.dispatch.end:
+// CHECK2-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// 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:%.*]]
@@ -2402,6 +2405,7 @@ int main (int argc, char **argv) {
// CHECK2: omp.dispatch.inc:
// CHECK2-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK2: omp.dispatch.end:
+// CHECK2-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// 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:%.*]]
@@ -3314,6 +3318,7 @@ int main (int argc, char **argv) {
// CHECK5: omp.dispatch.inc:
// CHECK5-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK5: omp.dispatch.end:
+// CHECK5-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK5-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -3472,6 +3477,7 @@ int main (int argc, char **argv) {
// CHECK5: omp.dispatch.inc:
// CHECK5-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK5: omp.dispatch.end:
+// CHECK5-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK5-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -4384,6 +4390,7 @@ int main (int argc, char **argv) {
// CHECK6: omp.dispatch.inc:
// CHECK6-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK6: omp.dispatch.end:
+// CHECK6-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK6-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK6-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK6-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -4542,6 +4549,7 @@ int main (int argc, char **argv) {
// CHECK6: omp.dispatch.inc:
// CHECK6-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK6: omp.dispatch.end:
+// CHECK6-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK6-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK6-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK6-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -6348,12 +6356,15 @@ int main (int argc, char **argv) {
// CHECK13: omp.dispatch.inc:
// CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK13: omp.dispatch.end:
-// CHECK13-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK13-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
-// CHECK13-NEXT: br i1 [[TMP23]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK13-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK13-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
+// CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
+// CHECK13-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK13-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
+// CHECK13-NEXT: br i1 [[TMP25]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK13: .omp.final.then:
-// CHECK13-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
-// CHECK13-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP24]], 0
+// CHECK13-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK13-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK13-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1
// CHECK13-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1
// CHECK13-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]]
@@ -6597,12 +6608,15 @@ int main (int argc, char **argv) {
// CHECK13: omp.dispatch.inc:
// CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK13: omp.dispatch.end:
-// CHECK13-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK13-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
-// CHECK13-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK13-NEXT: [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK13-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
+// CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP24]])
+// CHECK13-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK13-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
+// CHECK13-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK13: .omp.final.then:
-// CHECK13-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK13-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP25]], 0
+// CHECK13-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK13-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP27]], 0
// CHECK13-NEXT: [[DIV9:%.*]] = sdiv i32 [[SUB8]], 1
// CHECK13-NEXT: [[MUL10:%.*]] = mul nsw i32 [[DIV9]], 1
// CHECK13-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
@@ -7558,6 +7572,7 @@ int main (int argc, char **argv) {
// CHECK13: omp.dispatch.inc:
// CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK13: omp.dispatch.end:
+// CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK13-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK13-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK13-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -7738,6 +7753,7 @@ int main (int argc, char **argv) {
// CHECK13: omp.dispatch.inc:
// CHECK13-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK13: omp.dispatch.end:
+// CHECK13-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
// CHECK13-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK13-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK13-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -9159,12 +9175,15 @@ int main (int argc, char **argv) {
// CHECK14: omp.dispatch.inc:
// CHECK14-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK14: omp.dispatch.end:
-// CHECK14-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK14-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
-// CHECK14-NEXT: br i1 [[TMP23]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK14-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK14-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
+// CHECK14-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
+// CHECK14-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK14-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
+// CHECK14-NEXT: br i1 [[TMP25]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK14: .omp.final.then:
-// CHECK14-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
-// CHECK14-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP24]], 0
+// CHECK14-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK14-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK14-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1
// CHECK14-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1
// CHECK14-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]]
@@ -9408,12 +9427,15 @@ int main (int argc, char **argv) {
// CHECK14: omp.dispatch.inc:
// CHECK14-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK14: omp.dispatch.end:
-// CHECK14-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK14-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
-// CHECK14-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK14-NEXT: [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK14-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
+// CHECK14-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP24]])
+// CHECK14-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK14-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
+// CHECK14-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK14: .omp.final.then:
-// CHECK14-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK14-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP25]], 0
+// CHECK14-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK14-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP27]], 0
// CHECK14-NEXT: [[DIV9:%.*]] = sdiv i32 [[SUB8]], 1
// CHECK14-NEXT: [[MUL10:%.*]] = mul nsw i32 [[DIV9]], 1
// CHECK14-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
@@ -10369,6 +10391,7 @@ int main (int argc, char **argv) {
// CHECK14: omp.dispatch.inc:
// CHECK14-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK14: omp.dispatch.end:
+// CHECK14-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK14-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK14-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK14-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -10549,6 +10572,7 @@ int main (int argc, char **argv) {
// CHECK14: omp.dispatch.inc:
// CHECK14-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK14: omp.dispatch.end:
+// CHECK14-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
// CHECK14-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK14-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK14-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -11954,12 +11978,15 @@ int main (int argc, char **argv) {
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
-// CHECK17-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK17-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
-// CHECK17-NEXT: br i1 [[TMP23]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK17-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK17-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
+// CHECK17-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK17-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
+// CHECK17-NEXT: br i1 [[TMP25]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK17: .omp.final.then:
-// CHECK17-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
-// CHECK17-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP24]], 0
+// CHECK17-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK17-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK17-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK17-NEXT: [[MUL8:%.*]] = mul nsw i32 [[DIV7]], 1
// CHECK17-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL8]]
@@ -12198,12 +12225,15 @@ int main (int argc, char **argv) {
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
-// CHECK17-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK17-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
-// CHECK17-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK17-NEXT: [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK17-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP24]])
+// CHECK17-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK17-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
+// CHECK17-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK17: .omp.final.then:
-// CHECK17-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK17-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP25]], 0
+// CHECK17-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK17-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP27]], 0
// CHECK17-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1
// CHECK17-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1
// CHECK17-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]]
@@ -13137,6 +13167,7 @@ int main (int argc, char **argv) {
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK17-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK17-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK17-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -13312,6 +13343,7 @@ int main (int argc, char **argv) {
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
+// CHECK17-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
// CHECK17-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK17-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK17-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -14717,12 +14749,15 @@ int main (int argc, char **argv) {
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
-// CHECK19-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK19-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
-// CHECK19-NEXT: br i1 [[TMP23]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK19-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK19-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP23]])
+// CHECK19-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK19-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
+// CHECK19-NEXT: br i1 [[TMP25]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK19: .omp.final.then:
-// CHECK19-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
-// CHECK19-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP24]], 0
+// CHECK19-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK19-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK19-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK19-NEXT: [[MUL8:%.*]] = mul nsw i32 [[DIV7]], 1
// CHECK19-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL8]]
@@ -14961,12 +14996,15 @@ int main (int argc, char **argv) {
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
-// CHECK19-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
-// CHECK19-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
-// CHECK19-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK19-NEXT: [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK19-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP24]])
+// CHECK19-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK19-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
+// CHECK19-NEXT: br i1 [[TMP26]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK19: .omp.final.then:
-// CHECK19-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
-// CHECK19-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP25]], 0
+// CHECK19-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK19-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP27]], 0
// CHECK19-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1
// CHECK19-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1
// CHECK19-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]]
@@ -15900,6 +15938,7 @@ int main (int argc, char **argv) {
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP6]])
// CHECK19-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK19-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK19-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
@@ -16075,6 +16114,7 @@ int main (int argc, char **argv) {
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
+// CHECK19-NEXT: call void @__kmpc_dispatch_deinit(ptr @[[GLOB3]], i32 [[TMP7]])
// CHECK19-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK19-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK19-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
diff --git a/llvm/include/llvm/Frontend/OpenMP/OMPIRBuilder.h b/llvm/include/llvm/Frontend/OpenMP/OMPIRBuilder.h
index 883699de182e0..5c8b63f478782 100644
--- a/llvm/include/llvm/Frontend/OpenMP/OMPIRBuilder.h
+++ b/llvm/include/llvm/Frontend/OpenMP/OMPIRBuilder.h
@@ -2832,6 +2832,9 @@ class OpenMPIRBuilder {
/// size \a IVSize and sign \a IVSigned.
FunctionCallee createDispatchFiniFunction(unsigned IVSize, bool IVSigned);
+ /// Returns __kmpc_dispatch_deinit runtime function.
+ FunctionCallee createDispatchDeinitFunction();
+
/// Declarations for LLVM-IR types (simple, array, function and structure) are
/// generated below. Their names are defined and used in OpenMPKinds.def. Here
/// we provide the declarations, the initializeTypes function will provide the
diff --git a/llvm/include/llvm/Frontend/OpenMP/OMPKinds.def b/llvm/include/llvm/Frontend/OpenMP/OMPKinds.def
index fe09bb8177c28..535a3a74d7eaa 100644
--- a/llvm/include/llvm/Frontend/OpenMP/OMPKinds.def
+++ b/llvm/include/llvm/Frontend/OpenMP/OMPKinds.def
@@ -327,6 +327,7 @@ __OMP_RTL(__kmpc_dispatch_fini_4, false, Void, IdentPtr, Int32)
__OMP_RTL(__kmpc_dispatch_fini_4u, false, Void, IdentPtr, Int32)
__OMP_RTL(__kmpc_dispatch_fini_8, false, Void, IdentPtr, Int32)
__OMP_RTL(__kmpc_dispatch_fini_8u, false, Void, IdentPtr, Int32)
+__OMP_RTL(__kmpc_dispatch_deinit, false, Void, IdentPtr, Int32)
__OMP_RTL(__kmpc_team_static_init_4, false, Void, IdentPtr, Int32, Int32Ptr,
Int32Ptr, Int32Ptr, Int32Ptr, Int32, Int32)
__OMP_RTL(__kmpc_team_static_init_4u, false, Void, IdentPtr, Int32, Int32Ptr,
@@ -861,6 +862,8 @@ __OMP_RTL_ATTRS(__kmpc_dispatch_fini_8, InaccessibleArgOnlyAttrs,
AttributeSet(), ParamAttrs(ReadOnlyPtrAttrs, SExt))
__OMP_RTL_ATTRS(__kmpc_dispatch_fini_8u, InaccessibleArgOnlyAttrs,
AttributeSet(), ParamAttrs(ReadOnlyPtrAttrs, SExt))
+__OMP_RTL_ATTRS(__kmpc_dispatch_deinit, InaccessibleArgOnlyAttrs,
+ AttributeSet(), ParamAttrs(ReadOnlyPtrAttrs, SExt))
__OMP_RTL_ATTRS(__kmpc_team_static_init_4, GetterArgWriteAttrs, AttributeSet(),
ParamAttrs(ReadOnlyPtrAttrs, SExt, ArgPtrAttrs, ArgPtrAttrs,
ArgPtrAttrs, ArgPtrAttrs, SExt, SExt))
diff --git a/llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp b/llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp
index ed06d4e2658e5..a06e81dd280e0 100644
--- a/llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp
+++ b/llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp
@@ -6471,6 +6471,10 @@ FunctionCallee OpenMPIRBuilder::createDispatchFiniFunction(unsigned IVSize,
return getOrCreateRuntimeFunction(M, Name);
}
+FunctionCallee OpenMPIRBuilder::createDispatchDeinitFunction() {
+ return getOrCreateRuntimeFunction(M, omp::OMPRTL___kmpc_dispatch_deinit);
+}
+
static Function *createOutlinedFunction(
OpenMPIRBuilder &OMPBuilder, IRBuilderBase &Builder, StringRef FuncName,
SmallVectorImpl<Value *> &Inputs,
diff --git a/offload/DeviceRTL/CMakeLists.txt b/offload/DeviceRTL/CMakeLists.txt
index d88430a52b8b7..7818c8d752599 100644
--- a/offload/DeviceRTL/CMakeLists.txt
+++ b/offload/DeviceRTL/CMakeLists.txt
@@ -81,6 +81,7 @@ set(include_files
${include_directory}/Synchronization.h
${include_directory}/Types.h
${include_directory}/Utils.h
+ ${include_directory}/Workshare.h
)
set(src_files
diff --git a/offload/DeviceRTL/include/Interface.h b/offload/DeviceRTL/include/Interface.h
index f4854ed3d1678..d36d4227091ef 100644
--- a/offload/DeviceRTL/include/Interface.h
+++ b/offload/DeviceRTL/include/Interface.h
@@ -359,6 +359,7 @@ int32_t __kmpc_cancel(IdentTy *Loc, int32_t TId, int32_t CancelVal);
///{
int32_t __kmpc_shuffle_int32(int32_t val, int16_t delta, int16_t size);
int64_t __kmpc_shuffle_int64(int64_t val, int16_t delta, int16_t size);
+
///}
}
diff --git a/offload/DeviceRTL/include/Workshare.h b/offload/DeviceRTL/include/Workshare.h
new file mode 100644
index 0000000000000..809adc4bdb552
--- /dev/null
+++ b/offload/DeviceRTL/include/Workshare.h
@@ -0,0 +1,30 @@
+//===-------- Workshare.h - OpenMP Workshare interface ------------ C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef OMPTARGET_WORKSHARE_H
+#define OMPTARGET_WORKSHARE_H
+
+#pragma omp begin declare target device_type(nohost)
+
+namespace ompx {
+
+namespace workshare {
+
+/// Initialize the worksharing machinery.
+void init();
+
+} // namespace workshare
+
+} // namespace ompx
+
+#pragma omp end declare target
+
+#endif
diff --git a/offload/DeviceRTL/src/Kernel.cpp b/offload/DeviceRTL/src/Kernel.cpp
index 95d4c728016d2..e99ce01cc6223 100644
--- a/offload/DeviceRTL/src/Kernel.cpp
+++ b/offload/DeviceRTL/src/Kernel.cpp
@@ -19,6 +19,7 @@
#include "State.h"
#include "Synchronization.h"
#include "Types.h"
+#include "Workshare.h"
#include "llvm/Frontend/OpenMP/OMPDeviceConstants.h"
@@ -34,6 +35,7 @@ inititializeRuntime(bool IsSPMD, KernelEnvironmentTy &KernelEnvironment,
mapping::init(IsSPMD);
state::init(IsSPMD, KernelEnvironment, KernelLaunchEnvironment);
allocator::init(IsSPMD, KernelEnvironment);
+ workshare::init();
}
/// Simple generic state machine for worker threads.
diff --git a/offload/DeviceRTL/src/Workshare.cpp b/offload/DeviceRTL/src/Workshare.cpp
index bcb7c5ad50a18..e197628266af1 100644
--- a/offload/DeviceRTL/src/Workshare.cpp
+++ b/offload/DeviceRTL/src/Workshare.cpp
@@ -12,6 +12,7 @@
//
//===----------------------------------------------------------------------===//
+#include "Workshare.h"
#include "Debug.h"
#include "Interface.h"
#include "Mapping.h"
@@ -444,28 +445,77 @@ template <typename T, typename ST> struct omptarget_nvptx_LoopSupport {
// KMP interface implementation (dyn loops)
////////////////////////////////////////////////////////////////////////////////
-// TODO: This is a stopgap. We probably want to expand the dispatch API to take
-// an DST pointer which can then be allocated properly without malloc.
-static DynamicScheduleTracker *THREAD_LOCAL(ThreadDSTPtr);
+// TODO: Expand the dispatch API to take a DST pointer which can then be
+// allocated properly without malloc.
+// For now, each team will contain an LDS pointer (ThreadDST) to a global array
+// of references to the DST structs allocated (in global memory) for each thread
+// in the team. The global memory array is allocated during the init phase if it
+// was not allocated already and will be deallocated when the dispatch phase
+// ends:
+//
+// __kmpc_dispatch_init
+//
+// ** Dispatch loop **
+//
+// __kmpc_dispatch_deinit
+//
+static DynamicScheduleTracker **SHARED(ThreadDST);
// Create a new DST, link the current one, and define the new as current.
static DynamicScheduleTracker *pushDST() {
+ int32_t ThreadIndex = mapping::getThreadIdInBlock();
+ // Each block will allocate an array of pointers to DST structs. The array is
+ // equal in length to the number of threads in that block.
+ if (!ThreadDST) {
+ // Allocate global memory array of pointers to DST structs:
+ if (mapping::isMainThreadInGenericMode() || ThreadIndex == 0)
+ ThreadDST = static_cast<DynamicScheduleTracker **>(
+ memory::allocGlobal(mapping::getNumberOfThreadsInBlock() *
+ sizeof(DynamicScheduleTracker *),
+ "new ThreadDST array"));
+ synchronize::threads(atomic::seq_cst);
+
+ // Initialize the array pointers:
+ ThreadDST[ThreadIndex] = nullptr;
+ }
+
+ // Create a DST struct for the current thread:
DynamicScheduleTracker *NewDST = static_cast<DynamicScheduleTracker *>(
memory::allocGlobal(sizeof(DynamicScheduleTracker), "new DST"));
*NewDST = DynamicScheduleTracker({0});
- NewDST->NextDST = ThreadDSTPtr;
- ThreadDSTPtr = NewDST;
- return ThreadDSTPtr;
+
+ // Add the new DST struct to the array of DST structs:
+ NewDST->NextDST = ThreadDST[ThreadIndex];
+ ThreadDST[ThreadIndex] = NewDST;
+ return NewDST;
}
// Return the current DST.
-static DynamicScheduleTracker *peekDST() { return ThreadDSTPtr; }
+static DynamicScheduleTracker *peekDST() {
+ return ThreadDST[mapping::getThreadIdInBlock()];
+}
// Pop the current DST and restore the last one.
static void popDST() {
- DynamicScheduleTracker *OldDST = ThreadDSTPtr->NextDST;
- memory::freeGlobal(ThreadDSTPtr, "remove DST");
- ThreadDSTPtr = OldDST;
+ int32_t ThreadIndex = mapping::getThreadIdInBlock();
+ DynamicScheduleTracker *CurrentDST = ThreadDST[ThreadIndex];
+ DynamicScheduleTracker *OldDST = CurrentDST->NextDST;
+ memory::freeGlobal(CurrentDST, "remove DST");
+ ThreadDST[ThreadIndex] = OldDST;
+
+ // Check if we need to deallocate the global array. Ensure all threads
+ // in the block have finished deallocating the individual DSTs.
+ synchronize::threads(atomic::seq_cst);
+ if (ThreadDST[ThreadIndex] == 0 && ThreadIndex == 0)
+ memory::freeGlobal(ThreadDST, "remove ThreadDST array");
+ synchronize::threads(atomic::seq_cst);
+}
+
+void workshare::init() {
+ if (mapping::isMainThreadInGenericMode() ||
+ mapping::getThreadIdInBlock() == 0)
+ ThreadDST = nullptr;
+ synchronize::threads(atomic::seq_cst);
}
extern "C" {
@@ -533,24 +583,23 @@ int __kmpc_dispatch_next_8u(IdentTy *loc, int32_t tid, int32_t *p_last,
// fini
void __kmpc_dispatch_fini_4(IdentTy *loc, int32_t tid) {
omptarget_nvptx_LoopSupport<int32_t, int32_t>::dispatch_fini();
- popDST();
}
void __kmpc_dispatch_fini_4u(IdentTy *loc, int32_t tid) {
omptarget_nvptx_LoopSupport<uint32_t, int32_t>::dispatch_fini();
- popDST();
}
void __kmpc_dispatch_fini_8(IdentTy *loc, int32_t tid) {
omptarget_nvptx_LoopSupport<int64_t, int64_t>::dispatch_fini();
- popDST();
}
void __kmpc_dispatch_fini_8u(IdentTy *loc, int32_t tid) {
omptarget_nvptx_LoopSupport<uint64_t, int64_t>::dispatch_fini();
- popDST();
}
+// deinit
+void __kmpc_dispatch_deinit(IdentTy *loc, int32_t tid) { popDST(); }
+
////////////////////////////////////////////////////////////////////////////////
// KMP interface implementation (static loops)
////////////////////////////////////////////////////////////////////////////////
diff --git a/offload/test/offloading/dynamic-schedule.cpp b/offload/test/offloading/dynamic-schedule.cpp
new file mode 100644
index 0000000000000..500528f5ccb67
--- /dev/null
+++ b/offload/test/offloading/dynamic-schedule.cpp
@@ -0,0 +1,57 @@
+// clang-format off
+// RUN: %libomptarget-compilexx-generic && %libomptarget-run-generic 2>&1 | %fcheck-generic
+// clang-format on
+
+// UNSUPPORTED: aarch64-unknown-linux-gnu
+// UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+// UNSUPPORTED: x86_64-pc-linux-gnu
+// UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+// UNSUPPORTED: s390x-ibm-linux-gnu
+// UNSUPPORTED: s390x-ibm-linux-gnu-LTO
+
+// REQUIRES: amdgcn-amd-amdhsa
+
+#include <omp.h>
+#include <stdio.h>
+
+#define N 100
+
+bool schedule(int lb, int ub, int stride, int chunk) {
+ int i;
+
+ int result[N];
+ for (i = 0; i < N; i++) {
+ result[i] = 0;
+ }
+
+#pragma omp target parallel for schedule(dynamic, chunk) \
+ map(tofrom : result[ : N])
+ for (i = lb; i < ub; i += stride) {
+ result[i] += i;
+ }
+
+ int value = 0;
+ bool success = true;
+ for (i = 0; i < N; i += stride) {
+ if (value != result[i]) {
+ printf("ERROR: result[%d] = %d instead of %d\n", i, result[i], value);
+ success = false;
+ break;
+ }
+ value += stride;
+ }
+
+ return success;
+}
+
+int main() {
+ // CHECK: SUCCESS CHUNK SIZE 1
+ if (schedule(0, N, 5, 1))
+ printf("SUCCESS CHUNK SIZE 1\n");
+
+ // CHECK: SUCCESS CHUNK SIZE 3
+ if (schedule(0, N, 5, 3))
+ printf("SUCCESS CHUNK SIZE 3\n");
+
+ return 0;
+}
diff --git a/offload/test/offloading/schedule.c b/offload/test/offloading/schedule.c
new file mode 100644
index 0000000000000..aed97073c898c
--- /dev/null
+++ b/offload/test/offloading/schedule.c
@@ -0,0 +1,84 @@
+// clang-format off
+// RUN: %libomptarget-compile-generic && %libomptarget-run-generic 2>&1 | %fcheck-generic
+// clang-format on
+
+// UNSUPPORTED: aarch64-unknown-linux-gnu
+// UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+// UNSUPPORTED: x86_64-pc-linux-gnu
+// UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+// UNSUPPORTED: s390x-ibm-linux-gnu
+// UNSUPPORTED: s390x-ibm-linux-gnu-LTO
+
+// REQUIRES: amdgcn-amd-amdhsa
+
+#include <omp.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+int ordered_example(int lb, int ub, int stride, int nteams) {
+ int i;
+ int size = (ub - lb) / stride;
+ double *output = (double *)malloc(size * sizeof(double));
+
+#pragma omp target teams map(from : output[0 : size]) num_teams(nteams) \
+ thread_limit(128)
+#pragma omp parallel for ordered schedule(dynamic)
+ for (i = lb; i < ub; i += stride) {
+#pragma omp ordered
+ { output[(i - lb) / stride] = omp_get_wtime(); }
+ }
+
+ // verification
+ for (int j = 0; j < size; j++) {
+ for (int jj = j + 1; jj < size; jj++) {
+ if (output[j] > output[jj]) {
+ printf("Fail to schedule in order.\n");
+ free(output);
+ return 1;
+ }
+ }
+ }
+
+ free(output);
+
+ printf("test ordered OK\n");
+
+ return 0;
+}
+
+int NO_order_example(int lb, int ub, int stride, int nteams) {
+ int i;
+ int size = (ub - lb) / stride;
+ double *output = (double *)malloc(size * sizeof(double));
+
+#pragma omp target teams map(from : output[0 : size]) num_teams(nteams) \
+ thread_limit(128)
+#pragma omp parallel for schedule(dynamic)
+ for (i = lb; i < ub; i += stride) {
+ output[(i - lb) / stride] = omp_get_wtime();
+ }
+
+ // verification
+ for (int j = 0; j < size; j++) {
+ for (int jj = j + 1; jj < size; jj++) {
+ if (output[j] > output[jj]) {
+ printf("Fail to schedule in order.\n");
+ free(output);
+ return 1;
+ }
+ }
+ }
+
+ free(output);
+
+ printf("test no order OK\n");
+
+ return 0;
+}
+
+int main() {
+ // CHECK: test no order OK
+ NO_order_example(0, 10, 1, 8);
+ // CHECK: test ordered OK
+ return ordered_example(0, 10, 1, 8);
+}
diff --git a/openmp/runtime/src/kmp.h b/openmp/runtime/src/kmp.h
index c8d821b12ff0c..916c1dc257008 100644
--- a/openmp/runtime/src/kmp.h
+++ b/openmp/runtime/src/kmp.h
@@ -3840,6 +3840,8 @@ extern void __kmpc_dispatch_fini_8(ident_t *loc, kmp_int32 gtid);
extern void __kmpc_dispatch_fini_4u(ident_t *loc, kmp_int32 gtid);
extern void __kmpc_dispatch_fini_8u(ident_t *loc, kmp_int32 gtid);
+extern void __kmpc_dispatch_deinit(ident_t *loc, kmp_int32 gtid);
+
#ifdef KMP_GOMP_COMPAT
extern void __kmp_aux_dispatch_init_4(ident_t *loc, kmp_int32 gtid,
diff --git a/openmp/runtime/src/kmp_dispatch.cpp b/openmp/runtime/src/kmp_dispatch.cpp
index fc33376511817..8475e80f0b02d 100644
--- a/openmp/runtime/src/kmp_dispatch.cpp
+++ b/openmp/runtime/src/kmp_dispatch.cpp
@@ -3009,6 +3009,11 @@ See @ref __kmpc_dispatch_fini_4
void __kmpc_dispatch_fini_8u(ident_t *loc, kmp_int32 gtid) {
__kmp_dispatch_finish<kmp_uint64>(gtid, loc);
}
+
+/*!
+See @ref __kmpc_dispatch_deinit
+*/
+void __kmpc_dispatch_deinit(ident_t *loc, kmp_int32 gtid) {}
/*! @} */
//-----------------------------------------------------------------------------
More information about the cfe-commits
mailing list