[clang] [Clang][CodeGen] Add metadata for load from reference (PR #98746)
via cfe-commits
cfe-commits at lists.llvm.org
Sat Jul 20 08:30:17 PDT 2024
llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT-->
@llvm/pr-subscribers-clang
Author: Yingwei Zheng (dtcxzyw)
<details>
<summary>Changes</summary>
This patch adds `!dereferenceable/!nonnull/!align` for load from reference to improve codegen.
---
Patch is 2.75 MiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/98746.diff
52 Files Affected:
- (modified) clang/docs/ReleaseNotes.rst (+2)
- (modified) clang/lib/CodeGen/CGExpr.cpp (+31-3)
- (modified) clang/test/CodeGenCXX/matrix-type-operators.cpp (+10-10)
- (modified) clang/test/CodeGenCXX/reference-field.cpp (+2-2)
- (modified) clang/test/OpenMP/amdgcn_target_codegen.cpp (+11-11)
- (modified) clang/test/OpenMP/cancel_codegen.cpp (+48-48)
- (modified) clang/test/OpenMP/cancellation_point_codegen.cpp (+38-38)
- (modified) clang/test/OpenMP/distribute_parallel_for_simd_firstprivate_codegen.cpp (+351-351)
- (modified) clang/test/OpenMP/distribute_parallel_for_simd_lastprivate_codegen.cpp (+331-331)
- (modified) clang/test/OpenMP/distribute_parallel_for_simd_private_codegen.cpp (+84-84)
- (modified) clang/test/OpenMP/distribute_simd_firstprivate_codegen.cpp (+243-243)
- (modified) clang/test/OpenMP/distribute_simd_lastprivate_codegen.cpp (+239-239)
- (modified) clang/test/OpenMP/distribute_simd_private_codegen.cpp (+96-96)
- (modified) clang/test/OpenMP/for_reduction_task_codegen.cpp (+23-23)
- (modified) clang/test/OpenMP/parallel_for_reduction_task_codegen.cpp (+21-21)
- (modified) clang/test/OpenMP/parallel_for_simd_codegen.cpp (+1-1)
- (modified) clang/test/OpenMP/parallel_master_reduction_task_codegen.cpp (+21-21)
- (modified) clang/test/OpenMP/parallel_master_taskloop_codegen.cpp (+161-161)
- (modified) clang/test/OpenMP/parallel_master_taskloop_firstprivate_codegen.cpp (+105-105)
- (modified) clang/test/OpenMP/parallel_master_taskloop_lastprivate_codegen.cpp (+178-178)
- (modified) clang/test/OpenMP/parallel_master_taskloop_simd_codegen.cpp (+357-357)
- (modified) clang/test/OpenMP/parallel_master_taskloop_simd_firstprivate_codegen.cpp (+118-118)
- (modified) clang/test/OpenMP/parallel_master_taskloop_simd_lastprivate_codegen.cpp (+221-221)
- (modified) clang/test/OpenMP/parallel_reduction_task_codegen.cpp (+21-21)
- (modified) clang/test/OpenMP/parallel_sections_reduction_task_codegen.cpp (+22-22)
- (modified) clang/test/OpenMP/sections_reduction_task_codegen.cpp (+24-24)
- (modified) clang/test/OpenMP/simd_private_taskloop_codegen.cpp (+260-260)
- (modified) clang/test/OpenMP/target_in_reduction_codegen.cpp (+11-11)
- (modified) clang/test/OpenMP/target_map_codegen_35.cpp (+1-1)
- (modified) clang/test/OpenMP/target_parallel_for_reduction_task_codegen.cpp (+22-22)
- (modified) clang/test/OpenMP/target_parallel_reduction_task_codegen.cpp (+22-22)
- (modified) clang/test/OpenMP/target_teams_distribute_parallel_for_simd_firstprivate_codegen.cpp (+426-426)
- (modified) clang/test/OpenMP/target_teams_distribute_parallel_for_simd_lastprivate_codegen.cpp (+326-326)
- (modified) clang/test/OpenMP/target_teams_distribute_parallel_for_simd_private_codegen.cpp (+18-18)
- (modified) clang/test/OpenMP/target_teams_distribute_simd_collapse_codegen.cpp (+66-66)
- (modified) clang/test/OpenMP/target_teams_distribute_simd_dist_schedule_codegen.cpp (+132-132)
- (modified) clang/test/OpenMP/target_teams_distribute_simd_firstprivate_codegen.cpp (+141-141)
- (modified) clang/test/OpenMP/target_teams_distribute_simd_lastprivate_codegen.cpp (+208-208)
- (modified) clang/test/OpenMP/target_teams_distribute_simd_private_codegen.cpp (+11-11)
- (modified) clang/test/OpenMP/target_teams_distribute_simd_reduction_codegen.cpp (+62-62)
- (modified) clang/test/OpenMP/task_codegen.cpp (+1843-1843)
- (modified) clang/test/OpenMP/teams_distribute_parallel_for_simd_collapse_codegen.cpp (+114-114)
- (modified) clang/test/OpenMP/teams_distribute_parallel_for_simd_dist_schedule_codegen.cpp (+322-322)
- (modified) clang/test/OpenMP/teams_distribute_parallel_for_simd_firstprivate_codegen.cpp (+236-236)
- (modified) clang/test/OpenMP/teams_distribute_parallel_for_simd_lastprivate_codegen.cpp (+325-325)
- (modified) clang/test/OpenMP/teams_distribute_parallel_for_simd_private_codegen.cpp (+33-33)
- (modified) clang/test/OpenMP/teams_distribute_parallel_for_simd_reduction_codegen.cpp (+107-107)
- (modified) clang/test/OpenMP/teams_distribute_parallel_for_simd_schedule_codegen.cpp (+980-980)
- (modified) clang/test/OpenMP/teams_distribute_simd_firstprivate_codegen.cpp (+145-145)
- (modified) clang/test/OpenMP/teams_distribute_simd_lastprivate_codegen.cpp (+229-229)
- (modified) clang/test/OpenMP/teams_distribute_simd_private_codegen.cpp (+24-24)
- (modified) clang/test/utils/update_cc_test_checks/Inputs/resolve-tmp-conflict.cpp.expected (+1-1)
``````````diff
diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index b0afadfa8d324..c3fa6d1ca7a12 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -75,6 +75,8 @@ C++ Specific Potentially Breaking Changes
To fix this, update libstdc++ to version 14.1.1 or greater.
+- Added ``!dereferenceable/!nonnull/!align`` metadata to load of references for better codegen.
+
ABI Changes in This Version
---------------------------
- Fixed Microsoft name mangling of implicitly defined variables used for thread
diff --git a/clang/lib/CodeGen/CGExpr.cpp b/clang/lib/CodeGen/CGExpr.cpp
index c5341e4b53651..51c8206071866 100644
--- a/clang/lib/CodeGen/CGExpr.cpp
+++ b/clang/lib/CodeGen/CGExpr.cpp
@@ -2799,9 +2799,37 @@ CodeGenFunction::EmitLoadOfReference(LValue RefLVal,
llvm::LoadInst *Load =
Builder.CreateLoad(RefLVal.getAddress(), RefLVal.isVolatile());
CGM.DecorateInstructionWithTBAA(Load, RefLVal.getTBAAInfo());
- return makeNaturalAddressForPointer(Load, RefLVal.getType()->getPointeeType(),
- CharUnits(), /*ForPointeeType=*/true,
- PointeeBaseInfo, PointeeTBAAInfo);
+ QualType PTy = RefLVal.getType()->getPointeeType();
+ if (!PTy->isIncompleteType() && PTy->isConstantSizeType()) {
+ llvm::LLVMContext &Ctx = getLLVMContext();
+ llvm::MDBuilder MDB(Ctx);
+ // Emit !dereferenceable metadata
+ Load->setMetadata(
+ llvm::LLVMContext::MD_dereferenceable,
+ llvm::MDNode::get(Ctx,
+ MDB.createConstant(llvm::ConstantInt::get(
+ Builder.getInt64Ty(),
+ CGM.getMinimumObjectSize(PTy).getQuantity()))));
+ // Emit !nonnull metadata
+ if (CGM.getTypes().getTargetAddressSpace(PTy) == 0 &&
+ !CGM.getCodeGenOpts().NullPointerIsValid)
+ Load->setMetadata(llvm::LLVMContext::MD_nonnull,
+ llvm::MDNode::get(Ctx, {}));
+ // Emit !align metadata
+ if (PTy->isObjectType()) {
+ auto Align =
+ CGM.getNaturalPointeeTypeAlignment(RefLVal.getType()).getQuantity();
+ if (Align > 1) {
+ Load->setMetadata(
+ llvm::LLVMContext::MD_align,
+ llvm::MDNode::get(Ctx, MDB.createConstant(llvm::ConstantInt::get(
+ Builder.getInt64Ty(), Align))));
+ }
+ }
+ }
+ return makeNaturalAddressForPointer(Load, PTy, CharUnits(),
+ /*ForPointeeType=*/true, PointeeBaseInfo,
+ PointeeTBAAInfo);
}
LValue CodeGenFunction::EmitLoadOfReferenceLValue(LValue RefLVal) {
diff --git a/clang/test/CodeGenCXX/matrix-type-operators.cpp b/clang/test/CodeGenCXX/matrix-type-operators.cpp
index 2eb832f1aca6d..1fd7fe7e3d574 100644
--- a/clang/test/CodeGenCXX/matrix-type-operators.cpp
+++ b/clang/test/CodeGenCXX/matrix-type-operators.cpp
@@ -215,11 +215,11 @@ void insert(MyMatrix<EltTy, Rows, Columns> &Mat, EltTy e, unsigned i, unsigned j
void test_insert_template1(MyMatrix<unsigned, 2, 2> &Mat, unsigned e, unsigned i, unsigned j) {
// CHECK-LABEL: @_Z21test_insert_template1R8MyMatrixIjLj2ELj2EEjjj(
- // NOOPT: [[MAT_ADDR:%.*]] = load ptr, ptr %Mat.addr, align 8{{$}}
+ // NOOPT: [[MAT_ADDR:%.*]] = load ptr, ptr %Mat.addr, align 8{{.*}}
// NOOPT-NEXT: [[E:%.*]] = load i32, ptr %e.addr, align 4{{$}}
// NOOPT-NEXT: [[I:%.*]] = load i32, ptr %i.addr, align 4{{$}}
// NOOPT-NEXT: [[J:%.*]] = load i32, ptr %j.addr, align 4{{$}}
- // OPT: [[MAT_ADDR:%.*]] = load ptr, ptr %Mat.addr, align 8, !tbaa !{{[0-9]+}}{{$}}
+ // OPT: [[MAT_ADDR:%.*]] = load ptr, ptr %Mat.addr, align 8, !tbaa !{{[0-9]+}}{{.*}}
// OPT-NEXT: [[E:%.*]] = load i32, ptr %e.addr, align 4, !tbaa !{{[0-9]+}}{{$}}
// OPT-NEXT: [[I:%.*]] = load i32, ptr %i.addr, align 4, !tbaa !{{[0-9]+}}{{$}}
// OPT-NEXT: [[J:%.*]] = load i32, ptr %j.addr, align 4, !tbaa !{{[0-9]+}}{{$}}
@@ -249,9 +249,9 @@ void test_insert_template1(MyMatrix<unsigned, 2, 2> &Mat, unsigned e, unsigned i
void test_insert_template2(MyMatrix<float, 3, 8> &Mat, float e) {
// CHECK-LABEL: @_Z21test_insert_template2R8MyMatrixIfLj3ELj8EEf(
- // NOOPT: [[MAT_ADDR:%.*]] = load ptr, ptr %Mat.addr, align 8{{$}}
+ // NOOPT: [[MAT_ADDR:%.*]] = load ptr, ptr %Mat.addr, align 8{{.*}}
// NOOPT-NEXT: [[E:%.*]] = load float, ptr %e.addr, align 4{{$}}
- // OPT: [[MAT_ADDR:%.*]] = load ptr, ptr %Mat.addr, align 8, !tbaa !{{[0-9]+}}{{$}}
+ // OPT: [[MAT_ADDR:%.*]] = load ptr, ptr %Mat.addr, align 8, !tbaa !{{[0-9]+}}{{.*}}
// OPT-NEXT: [[E:%.*]] = load float, ptr %e.addr, align 4, !tbaa !{{[0-9]+}}{{$}}
// CHECK-NEXT: call void @_Z6insertIfLj3ELj8EEvR8MyMatrixIT_XT0_EXT1_EES1_jj(ptr noundef nonnull align 4 dereferenceable(96) [[MAT_ADDR]], float noundef [[E]], i32 noundef 2, i32 noundef 5)
// CHECK-NEXT: ret void
@@ -346,11 +346,11 @@ double extract_IntWrapper_idx(double4x4 &m, IntWrapper i, UnsignedWrapper j) {
// CHECK-NEXT: [[J_SUB_EXT:%.*]] = zext i32 [[J_SUB]] to i64
// CHECK-NEXT: [[IDX1:%.*]] = mul i64 [[J_SUB_EXT]], 4
// CHECK-NEXT: [[IDX2:%.*]] = add i64 [[IDX1]], [[I_ADD_EXT]]
- // NOOPT-NEXT: [[MAT_ADDR:%.*]] = load ptr, ptr %m.addr, align 8{{$}}
+ // NOOPT-NEXT: [[MAT_ADDR:%.*]] = load ptr, ptr %m.addr, align 8{{.*}}
// NOOPT-NEXT: [[MAT:%.*]] = load <16 x double>, ptr [[MAT_ADDR]], align 8{{$}}
// OPT-NEXT: [[CMP:%.*]] = icmp ult i64 [[IDX2]], 16
// OPT-NEXT: call void @llvm.assume(i1 [[CMP]])
- // OPT-NEXT: [[MAT_ADDR:%.*]] = load ptr, ptr %m.addr, align 8, !tbaa !{{[0-9]+}}{{$}}
+ // OPT-NEXT: [[MAT_ADDR:%.*]] = load ptr, ptr %m.addr, align 8, !tbaa !{{[0-9]+}}{{.*}}
// OPT-NEXT: [[MAT:%.*]] = load <16 x double>, ptr [[MAT_ADDR]], align 8, !tbaa !{{[0-9]+}}{{$}}
// CHECK-NEXT: [[MATEXT:%.*]] = extractelement <16 x double> [[MAT]], i64 [[IDX2]]
// CHECK-NEXT: ret double [[MATEXT]]
@@ -377,8 +377,8 @@ void test_constexpr1(matrix_type<float, 4, 4> &m) {
// OPT: [[MAT:%.*]] = load <16 x float>, ptr {{.*}}, align 4, !tbaa !{{[0-9]+}}{{$}}
// CHECK-NEXT: [[IM:%.*]] = call noundef <16 x float> @_ZNK13identmatrix_tcvu11matrix_typeIXT0_EXT0_ET_EIfLj4EEEv(ptr {{[^,]*}} @_ZL11identmatrix)
// CHECK-NEXT: [[ADD:%.*]] = fadd <16 x float> [[MAT]], [[IM]]
- // NOOPT-NEXT: [[MAT_ADDR:%.*]] = load ptr, ptr %m.addr, align 8{{$}}
- // OPT-NEXT: [[MAT_ADDR:%.*]] = load ptr, ptr %m.addr, align 8, !tbaa !{{[0-9]+}}{{$}}
+ // NOOPT-NEXT: [[MAT_ADDR:%.*]] = load ptr, ptr %m.addr, align 8{{.*}}
+ // OPT-NEXT: [[MAT_ADDR:%.*]] = load ptr, ptr %m.addr, align 8, !tbaa !{{[0-9]+}}{{.*}}
// CHECK-NEXT: store <16 x float> [[ADD]], ptr [[MAT_ADDR]], align 4
// CHECK-NEXT: ret voi
@@ -408,8 +408,8 @@ void test_constexpr2(matrix_type<int, 5, 5> &m) {
// OPT: [[MAT:%.*]] = load <25 x i32>, ptr {{.*}}, align 4, !tbaa !{{[0-9]+}}{{$}}
// CHECK-NEXT: [[SUB:%.*]] = sub <25 x i32> [[IM]], [[MAT]]
// CHECK-NEXT: [[SUB2:%.*]] = add <25 x i32> [[SUB]], <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
- // NOOPT-NEXT: [[MAT_ADDR:%.*]] = load ptr, ptr %m.addr, align 8{{$}}
- // OPT-NEXT: [[MAT_ADDR:%.*]] = load ptr, ptr %m.addr, align 8, !tbaa !{{[0-9]+}}{{$}}
+ // NOOPT-NEXT: [[MAT_ADDR:%.*]] = load ptr, ptr %m.addr, align 8{{.*}}
+ // OPT-NEXT: [[MAT_ADDR:%.*]] = load ptr, ptr %m.addr, align 8, !tbaa !{{[0-9]+}}{{.*}}
// CHECK-NEXT: store <25 x i32> [[SUB2]], ptr [[MAT_ADDR]], align 4
// CHECK-NEXT: ret void
//
diff --git a/clang/test/CodeGenCXX/reference-field.cpp b/clang/test/CodeGenCXX/reference-field.cpp
index 54e914d0f3fce..8146594c3c835 100644
--- a/clang/test/CodeGenCXX/reference-field.cpp
+++ b/clang/test/CodeGenCXX/reference-field.cpp
@@ -1,8 +1,8 @@
// RUN: %clang_cc1 -emit-llvm -triple %itanium_abi_triple -o - %s -O2 | FileCheck %s
-// Make sure the call to b() doesn't get optimized out.
+// Make sure the call to b() is eliminated.
extern struct x {char& x,y;}y;
int b();
int a() { if (!&y.x) b(); }
-// CHECK: @_Z1bv
+// CHECK-NOT: @_Z1bv
diff --git a/clang/test/OpenMP/amdgcn_target_codegen.cpp b/clang/test/OpenMP/amdgcn_target_codegen.cpp
index 3ea2d107f072a..108ec3a5ca676 100644
--- a/clang/test/OpenMP/amdgcn_target_codegen.cpp
+++ b/clang/test/OpenMP/amdgcn_target_codegen.cpp
@@ -39,7 +39,7 @@ int test_amdgcn_target_tid_threads_simd() {
// CHECK-NEXT: [[I_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I]] to ptr
// CHECK-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: store ptr [[ARR]], ptr [[ARR_ADDR_ASCAST]], align 8
-// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[ARR_ADDR_ASCAST]], align 8
+// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[ARR_ADDR_ASCAST]], align 8, !nonnull [[META9:![0-9]+]], !dereferenceable [[META10:![0-9]+]], !align [[META11:![0-9]+]]
// CHECK-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr addrspacecast (ptr addrspace(1) @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30test_amdgcn_target_tid_threadsv_l14_kernel_environment to ptr), ptr [[DYN_PTR]])
// CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
@@ -60,7 +60,7 @@ int test_amdgcn_target_tid_threads_simd() {
// CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[I_ASCAST]], align 4
// CHECK-NEXT: [[INC:%.*]] = add nsw i32 [[TMP4]], 1
// CHECK-NEXT: store i32 [[INC]], ptr [[I_ASCAST]], align 4
-// CHECK-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]]
+// CHECK-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK: worker.exit:
// CHECK-NEXT: ret void
// CHECK: for.end:
@@ -83,7 +83,7 @@ int test_amdgcn_target_tid_threads_simd() {
// CHECK-NEXT: [[I_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I]] to ptr
// CHECK-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: store ptr [[ARR]], ptr [[ARR_ADDR_ASCAST]], align 8
-// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[ARR_ADDR_ASCAST]], align 8
+// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[ARR_ADDR_ASCAST]], align 8, !nonnull [[META9]], !dereferenceable [[META10]], !align [[META11]]
// CHECK-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr addrspacecast (ptr addrspace(1) @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z35test_amdgcn_target_tid_threads_simdv_l23_kernel_environment to ptr), ptr [[DYN_PTR]])
// CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
@@ -91,26 +91,26 @@ int test_amdgcn_target_tid_threads_simd() {
// CHECK-NEXT: store i32 0, ptr [[DOTOMP_IV_ASCAST]], align 4
// CHECK-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK: omp.inner.for.cond:
-// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4, !llvm.access.group [[ACC_GRP11:![0-9]+]]
+// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4, !llvm.access.group [[ACC_GRP14:![0-9]+]]
// CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 [[TMP2]], 1000
// CHECK-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK: omp.inner.for.body:
-// CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4, !llvm.access.group [[ACC_GRP11]]
+// CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4, !llvm.access.group [[ACC_GRP14]]
// CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
// CHECK-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
-// CHECK-NEXT: store i32 [[ADD]], ptr [[I_ASCAST]], align 4, !llvm.access.group [[ACC_GRP11]]
-// CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[I_ASCAST]], align 4, !llvm.access.group [[ACC_GRP11]]
+// CHECK-NEXT: store i32 [[ADD]], ptr [[I_ASCAST]], align 4, !llvm.access.group [[ACC_GRP14]]
+// CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[I_ASCAST]], align 4, !llvm.access.group [[ACC_GRP14]]
// CHECK-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP4]] to i64
// CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [1000 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
-// CHECK-NEXT: store i32 1, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP11]]
+// CHECK-NEXT: store i32 1, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP14]]
// CHECK-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK: omp.body.continue:
// CHECK-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK: omp.inner.for.inc:
-// CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4, !llvm.access.group [[ACC_GRP11]]
+// CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV_ASCAST]], align 4, !llvm.access.group [[ACC_GRP14]]
// CHECK-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], 1
-// CHECK-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV_ASCAST]], align 4, !llvm.access.group [[ACC_GRP11]]
-// CHECK-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
+// CHECK-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV_ASCAST]], align 4, !llvm.access.group [[ACC_GRP14]]
+// CHECK-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK: worker.exit:
// CHECK-NEXT: ret void
// CHECK: omp.inner.for.end:
diff --git a/clang/test/OpenMP/cancel_codegen.cpp b/clang/test/OpenMP/cancel_codegen.cpp
index 186f0ab7356d4..eb1a89fad36c7 100644
--- a/clang/test/OpenMP/cancel_codegen.cpp
+++ b/clang/test/OpenMP/cancel_codegen.cpp
@@ -131,7 +131,7 @@ for (int i = 0; i < argc; ++i) {
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_SECTIONS_IV_]], align 4
// CHECK1-NEXT: switch i32 [[TMP7]], label [[DOTOMP_SECTIONS_EXIT:%.*]] [
-// CHECK1-NEXT: i32 0, label [[DOTOMP_SECTIONS_CASE:%.*]]
+// CHECK1-NEXT: i32 0, label [[DOTOMP_SECTIONS_CASE:%.*]]
// CHECK1-NEXT: ]
// CHECK1: .omp.sections.case:
// CHECK1-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_cancel(ptr @[[GLOB1]], i32 [[TMP0]], i32 3)
@@ -173,8 +173,8 @@ for (int i = 0; i < argc; ++i) {
// CHECK1: omp.inner.for.body8:
// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_SECTIONS_IV_5]], align 4
// CHECK1-NEXT: switch i32 [[TMP17]], label [[DOTOMP_SECTIONS_EXIT15:%.*]] [
-// CHECK1-NEXT: i32 0, label [[DOTOMP_SECTIONS_CASE9:%.*]]
-// CHECK1-NEXT: i32 1, label [[DOTOMP_SECTIONS_CASE12:%.*]]
+// CHECK1-NEXT: i32 0, label [[DOTOMP_SECTIONS_CASE9:%.*]]
+// CHECK1-NEXT: i32 1, label [[DOTOMP_SECTIONS_CASE12:%.*]]
// CHECK1-NEXT: ]
// CHECK1: .omp.sections.case9:
// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__kmpc_cancel(ptr @[[GLOB1]], i32 [[TMP0]], i32 3)
@@ -310,8 +310,8 @@ for (int i = 0; i < argc; ++i) {
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8
// CHECK1-NEXT: store ptr [[ARGC]], ptr [[ARGC_ADDR]], align 8
-// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8
-// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[ARGC_ADDR]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8, !nonnull [[META3:![0-9]+]], !dereferenceable [[META4:![0-9]+]], !align [[META4]]
+// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[ARGC_ADDR]], align 8, !nonnull [[META3]], !dereferenceable [[META5:![0-9]+]], !align [[META5]]
// CHECK1-NEXT: [[TMP2:%.*]] = load float, ptr @flag, align 4
// CHECK1-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP2]], 0.000000e+00
// CHECK1-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
@@ -381,29 +381,29 @@ for (int i = 0; i < argc; ++i) {
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
-// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META5:![0-9]+]])
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]])
-// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]])
-// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META12:![0-9]+]])
-// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !14
-// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias !14
-// CHECK1-NEXT: store ptr null, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias !14
-// CHECK1-NEXT: store ptr null, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !14
-// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias !14
-// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias !14
-// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias !14
-// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !14
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META11:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META13:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META15:![0-9]+]])
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META17:![0-9]+]]
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META17]]
+// CHECK1-NEXT: store ptr null, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META17]]
+// CHECK1-NEXT: store ptr null, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META17]]
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META17]]
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META17]]
+// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META17]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META17]]
// CHECK1-NEXT: [[TMP10:%.*]] = call i32 @__kmpc_cancel(ptr @[[GLOB1]], i32 [[TMP9]], i32 4)
// CHECK1-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP10]], 0
// CHECK1-NEXT: br i1 [[TMP11]], label [[DOTCANCEL_EXIT_I:%.*]], label [[DOTCANCEL_CONTINUE_I:%.*]]
// CHECK1: .cancel.exit.i:
-// CHECK1-NEXT: store i32 1, ptr [[CLEANUP_DEST_SLOT_I]], align 4, !noalias !14
+// CHECK1-NEXT: store i32 1, ptr [[CLEANUP_DEST_SLOT_I]], align 4, !noalias [[META17]]
// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__EXIT:%.*]]
// CHECK1: .cancel.continue.i:
-// CHECK1-NEXT: store i32 0, ptr [[CLEANUP_DEST_SLOT_I]], align 4, !noalias !14
+// CHECK1-NEXT: store i32 0, ptr [[CLEANUP_DEST_SLOT_I]], align 4, !noalias [[META17]]
// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__EXIT]]
// CHECK1: .omp_outlined..exit:
-// CHECK1-NEXT: [[CLEANUP_DEST_I:%.*]] = load i32, ptr [[CLEANUP_DEST_SLOT_I]], align 4, !noalias !14
+// CHECK1-NEXT: [[CLEANUP_DEST_I:%.*]] = load i32, ptr [[CLEANUP_DEST_SLOT_I]], align 4, !noalias [[META17]]
// CHECK1-NEXT: ret i32 0
//
//
@@ -441,7 +441,7 @@ for (int i = 0; i < argc; ++i) {
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_SECTIONS_IV_]], align 4
// CHECK1-NEXT: switch i32 [[TMP8]], label [[DOTOMP_SECTIONS_EXIT:%.*]] [
-// CHECK1-NEXT: i32 0, label [[DOTOMP_SECTIONS_CASE:%.*]]
+// CHECK1-NEXT: i32 0, label [[DOTOMP_SECTIONS_CASE:%.*]]
// CHECK1-NEXT: ]
// CHECK1: .omp.sections.case:
// CHECK1-NEXT: [[TMP9:%.*]] = call i32 @__kmpc_cancel(ptr @[[GLOB1]], i32 [[TMP1]], i32 3)
@@ -502,8 +502,8 @@ for (int i = 0; i < argc; ++i) {
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_SECTIONS_IV_]], align 4
// CHECK1-NEXT: switch i32 [[TMP8]], label [[DOTOMP_SECTIO...
[truncated]
``````````
</details>
https://github.com/llvm/llvm-project/pull/98746
More information about the cfe-commits
mailing list