[PATCH] D49188: [OpenMP] Initialize data sharing stack for SPMD case
This revision was automatically updated to reflect the committed changes.
Closed by commit rL337015: [OpenMP] Initialize data sharing stack for SPMD case
(authored by gbercea, committed by ).
Herald added a subscriber: llvm-commits.
Repository:
rL LLVM
https://reviews.llvm.org/D49188
Files:
cfe/trunk/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
cfe/trunk/test/OpenMP/nvptx_data_sharing.cpp
cfe/trunk/test/OpenMP/nvptx_target_parallel_codegen.cpp
cfe/trunk/test/OpenMP/nvptx_target_parallel_proc_bind_codegen.cpp
cfe/trunk/test/OpenMP/nvptx_target_parallel_reduction_codegen.cpp
cfe/trunk/test/OpenMP/nvptx_target_teams_codegen.cpp
cfe/trunk/test/OpenMP/nvptx_target_teams_distribute_parallel_for_codegen.cpp
cfe/trunk/test/OpenMP/nvptx_target_teams_distribute_parallel_for_simd_codegen.cpp
Index: cfe/trunk/test/OpenMP/nvptx_target_teams_distribute_parallel_for_simd_codegen.cpp
===
--- cfe/trunk/test/OpenMP/nvptx_target_teams_distribute_parallel_for_simd_codegen.cpp
+++ cfe/trunk/test/OpenMP/nvptx_target_teams_distribute_parallel_for_simd_codegen.cpp
@@ -22,7 +22,7 @@
tx a[N];
short aa[N];
tx b[10];
- tx c[M][M];
+ tx c[M][M];
tx f = n;
tx l;
int k;
@@ -47,7 +47,7 @@
for(int i = 0; i < M; i++) {
for(int j = 0; j < M; j++) {
k = M;
- c[i][j] = i+j*f+k;
+ c[i][j] = i+j*f+k;
}
}
@@ -65,6 +65,7 @@
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}(
// CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
// CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]],
+// CHECK: call void @__kmpc_data_sharing_init_stack_spmd
// CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 91,
// CHECK: {{call|invoke}} void [[OUTL1:@.+]](
// CHECK: call void @__kmpc_for_static_fini(
@@ -79,6 +80,7 @@
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}(
// CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
// CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]],
+// CHECK: call void @__kmpc_data_sharing_init_stack_spmd
// CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 92,
// CHECK: {{call|invoke}} void [[OUTL2:@.+]](
// CHECK: call void @__kmpc_for_static_fini(
@@ -93,6 +95,7 @@
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}(
// CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
// CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]],
+// CHECK: call void @__kmpc_data_sharing_init_stack_spmd
// CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 92,
// CHECK: {{call|invoke}} void [[OUTL3:@.+]](
// CHECK: call void @__kmpc_for_static_fini(
@@ -108,6 +111,7 @@
// CHECK: store {{.+}} [[F_IN]], {{.+}}* {{.+}},
// CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
// CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]],
+// CHECK: call void @__kmpc_data_sharing_init_stack_spmd
// CHECK: store {{.+}} 99, {{.+}}* [[COMB_UB:%.+]], align
// CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 92, {{.+}}, {{.+}}, {{.+}}* [[COMB_UB]],
// CHECK: {{call|invoke}} void [[OUTL4:@.+]](
Index: cfe/trunk/test/OpenMP/nvptx_target_teams_codegen.cpp
===
--- cfe/trunk/test/OpenMP/nvptx_target_teams_codegen.cpp
+++ cfe/trunk/test/OpenMP/nvptx_target_teams_codegen.cpp
@@ -228,6 +228,7 @@
// CHECK: define weak void @__omp_offloading_{{.*}}ftemplate{{.*}}_l37(
// CHECK: call void @__kmpc_spmd_kernel_init(
+// CHECK: call void @__kmpc_data_sharing_init_stack_spmd
// CHECK: call i8* @__kmpc_data_sharing_push_stack(
// CHECK-NOT: call void @__kmpc_serialized_parallel(
// CHECK: call void [[L0:@.+]](i32* %{{.+}}, i32* %{{.+}}, i16* %{{.*}})
Index: cfe/trunk/test/OpenMP/nvptx_target_parallel_proc_bind_codegen.cpp
===
--- cfe/trunk/test/OpenMP/nvptx_target_parallel_proc_bind_codegen.cpp
+++ cfe/trunk/test/OpenMP/nvptx_target_parallel_proc_bind_codegen.cpp
@@ -48,6 +48,7 @@
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+template.+l22}}(
// CHECK: call void @__kmpc_spmd_kernel_init(
+ // CHECK: call void @__kmpc_data_sharing_init_stack_spmd
// CHECK: br label {{%?}}[[EXEC:.+]]
//
// CHECK: [[EXEC]]
@@ -69,6 +70,7 @@
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+template.+l26}}(
// CHECK: call void @__kmpc_spmd_kernel_init(
+ // CHECK: call void @__kmpc_data_sharing_init_stack_spmd
// CHECK: br label {{%?}}[[EXEC:.+]]
//
// CHECK: [[EXEC]]
@@ -89,6 +91,7 @@
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+template.+l31}}(
// CHECK: call void @__kmpc_spmd_kernel_init(
+ // CHECK: call void @__kmpc_data_sharing_init_stack_spmd
// CHECK: br label {{%?}}[[EXEC:.+]]
//
// CHECK:
[PATCH] D49188: [OpenMP] Initialize data sharing stack for SPMD case
ABataev accepted this revision. ABataev added a comment. This revision is now accepted and ready to land. LG Repository: rC Clang https://reviews.llvm.org/D49188 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D49188: [OpenMP] Initialize data sharing stack for SPMD case
gtbercea updated this revision to Diff 155391.
gtbercea added a comment.
Fix tests.
Repository:
rC Clang
https://reviews.llvm.org/D49188
Files:
lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
test/OpenMP/nvptx_data_sharing.cpp
test/OpenMP/nvptx_target_parallel_codegen.cpp
test/OpenMP/nvptx_target_parallel_proc_bind_codegen.cpp
test/OpenMP/nvptx_target_parallel_reduction_codegen.cpp
test/OpenMP/nvptx_target_teams_codegen.cpp
test/OpenMP/nvptx_target_teams_distribute_parallel_for_codegen.cpp
test/OpenMP/nvptx_target_teams_distribute_parallel_for_simd_codegen.cpp
Index: test/OpenMP/nvptx_target_teams_distribute_parallel_for_simd_codegen.cpp
===
--- test/OpenMP/nvptx_target_teams_distribute_parallel_for_simd_codegen.cpp
+++ test/OpenMP/nvptx_target_teams_distribute_parallel_for_simd_codegen.cpp
@@ -22,7 +22,7 @@
tx a[N];
short aa[N];
tx b[10];
- tx c[M][M];
+ tx c[M][M];
tx f = n;
tx l;
int k;
@@ -47,7 +47,7 @@
for(int i = 0; i < M; i++) {
for(int j = 0; j < M; j++) {
k = M;
- c[i][j] = i+j*f+k;
+ c[i][j] = i+j*f+k;
}
}
@@ -65,6 +65,7 @@
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}(
// CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
// CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]],
+// CHECK: call void @__kmpc_data_sharing_init_stack_spmd
// CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 91,
// CHECK: {{call|invoke}} void [[OUTL1:@.+]](
// CHECK: call void @__kmpc_for_static_fini(
@@ -79,6 +80,7 @@
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}(
// CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
// CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]],
+// CHECK: call void @__kmpc_data_sharing_init_stack_spmd
// CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 92,
// CHECK: {{call|invoke}} void [[OUTL2:@.+]](
// CHECK: call void @__kmpc_for_static_fini(
@@ -93,6 +95,7 @@
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}(
// CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
// CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]],
+// CHECK: call void @__kmpc_data_sharing_init_stack_spmd
// CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 92,
// CHECK: {{call|invoke}} void [[OUTL3:@.+]](
// CHECK: call void @__kmpc_for_static_fini(
@@ -108,6 +111,7 @@
// CHECK: store {{.+}} [[F_IN]], {{.+}}* {{.+}},
// CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
// CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]],
+// CHECK: call void @__kmpc_data_sharing_init_stack_spmd
// CHECK: store {{.+}} 99, {{.+}}* [[COMB_UB:%.+]], align
// CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 92, {{.+}}, {{.+}}, {{.+}}* [[COMB_UB]],
// CHECK: {{call|invoke}} void [[OUTL4:@.+]](
Index: test/OpenMP/nvptx_target_teams_distribute_parallel_for_codegen.cpp
===
--- test/OpenMP/nvptx_target_teams_distribute_parallel_for_codegen.cpp
+++ test/OpenMP/nvptx_target_teams_distribute_parallel_for_codegen.cpp
@@ -70,6 +70,7 @@
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}(
// CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
// CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]],
+// CHECK: call void @__kmpc_data_sharing_init_stack_spmd
// CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 91,
// CHECK: {{call|invoke}} void [[OUTL1:@.+]](
// CHECK: call void @__kmpc_for_static_fini(
@@ -84,6 +85,7 @@
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}(
// CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
// CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]],
+// CHECK: call void @__kmpc_data_sharing_init_stack_spmd
// CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 92,
// CHECK: {{call|invoke}} void [[OUTL2:@.+]](
// CHECK: call void @__kmpc_for_static_fini(
@@ -98,6 +100,7 @@
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}(
// CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
// CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]],
+// CHECK: call void @__kmpc_data_sharing_init_stack_spmd
// CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 92,
// CHECK: {{call|invoke}} void [[OUTL3:@.+]](
// CHECK: call void @__kmpc_for_static_fini(
@@ -113,6 +116,7 @@
// CHECK: store {{.+}} [[F_IN]], {{.+}}* {{.+}},
// CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
// CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]],
+// CHECK: call void @__kmpc_data_sharing_init_stack_spmd
// CHECK: store {{.+}} 99, {{.+}}* [[COMB_UB:%.+]],
[PATCH] D49188: [OpenMP] Initialize data sharing stack for SPMD case
ABataev added inline comments. Comment at: test/OpenMP/nvptx_data_sharing.cpp:33 // CK1: call void @llvm.nvvm.barrier0() -// CK1: call void @__kmpc_data_sharing_init_stack It is better to check that this call is not emitted, like this `CK1-NOT: call void @__kmpc_data_sharing_init_stack` Comment at: test/OpenMP/nvptx_data_sharing_spmd.cpp:1-24 +// Test device global memory data sharing initialization codegen for spmd. +///==/// + +// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc +// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple nvptx64-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix CK1 + +// expected-no-diagnostics I think we already have some tests with SPMD construct, you can just modify the existing tests instead of adding another one (e.g. `nvptx_target_parallel_codegen.cpp`) Repository: rC Clang https://reviews.llvm.org/D49188 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D49188: [OpenMP] Initialize data sharing stack for SPMD case
gtbercea marked 2 inline comments as done. gtbercea added inline comments. Comment at: lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp:84-85 OMPRTL_NVPTX__kmpc_data_sharing_init_stack, + /// Call to void __kmpc_data_sharing_init_stack_spmd(); + OMPRTL_NVPTX__kmpc_data_sharing_init_stack_spmd, /// Call to void* __kmpc_data_sharing_push_stack(size_t size, Hahnfeld wrote: > In case I'm not missing something obvious this function doesn't exist (yet) > in libomptarget-nvptx? Correct. There's a libomptarget patch coming up as well. Comment at: lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp:1110-1114 - // For data sharing, we need to initialize the stack for workers. - CGF.EmitRuntimeCall( - createNVPTXRuntimeFunction( - OMPRTL_NVPTX__kmpc_data_sharing_init_stack)); - Hahnfeld wrote: > Why is this call being removed? There's no mention in the summary AFAICS Not needed on workers after libomptarget patch. I could perhaps put this into a separate patch along with the test fix. Repository: rC Clang https://reviews.llvm.org/D49188 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D49188: [OpenMP] Initialize data sharing stack for SPMD case
gtbercea updated this revision to Diff 155037.
gtbercea added a comment.
Add test for spmd stack init function.
Repository:
rC Clang
https://reviews.llvm.org/D49188
Files:
lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
test/OpenMP/nvptx_data_sharing.cpp
test/OpenMP/nvptx_data_sharing_spmd.cpp
Index: test/OpenMP/nvptx_data_sharing_spmd.cpp
===
--- /dev/null
+++ test/OpenMP/nvptx_data_sharing_spmd.cpp
@@ -0,0 +1,24 @@
+// Test device global memory data sharing initialization codegen for spmd.
+///==///
+
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown
-fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple nvptx64-unknown-unknown
-fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device
-fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix CK1
+
+// expected-no-diagnostics
+
+#ifndef HEADER
+#define HEADER
+
+void test_ds(){
+ #pragma omp target teams distribute parallel for
+ for (int i = 0; i < 100; i++ ) {
+int a = 100;
+ }
+}
+
+// CK1: {{.*}}define weak void @__omp_offloading{{.*}}test_ds{{.*}}()
+// CK1: call void @__kmpc_spmd_kernel_init
+// CK1-NEXT: call void @__kmpc_data_sharing_init_stack_spmd
+
+#endif
+
Index: test/OpenMP/nvptx_data_sharing.cpp
===
--- test/OpenMP/nvptx_data_sharing.cpp
+++ test/OpenMP/nvptx_data_sharing.cpp
@@ -30,7 +30,6 @@
/// = In the worker function = ///
// CK1: {{.*}}define internal void
@__omp_offloading{{.*}}test_ds{{.*}}_worker()
// CK1: call void @llvm.nvvm.barrier0()
-// CK1: call void @__kmpc_data_sharing_init_stack
/// = In the kernel function = ///
Index: lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
===
--- lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
+++ lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
@@ -81,6 +81,8 @@
OMPRTL_NVPTX__kmpc_end_reduce_nowait,
/// Call to void __kmpc_data_sharing_init_stack();
OMPRTL_NVPTX__kmpc_data_sharing_init_stack,
+ /// Call to void __kmpc_data_sharing_init_stack_spmd();
+ OMPRTL_NVPTX__kmpc_data_sharing_init_stack_spmd,
/// Call to void* __kmpc_data_sharing_push_stack(size_t size,
/// int16_t UseSharedMemory);
OMPRTL_NVPTX__kmpc_data_sharing_push_stack,
@@ -1025,6 +1027,12 @@
/*RequiresDataSharing=*/Bld.getInt16(1)};
CGF.EmitRuntimeCall(
createNVPTXRuntimeFunction(OMPRTL_NVPTX__kmpc_spmd_kernel_init), Args);
+
+ // For data sharing, we need to initialize the stack.
+ CGF.EmitRuntimeCall(
+ createNVPTXRuntimeFunction(
+ OMPRTL_NVPTX__kmpc_data_sharing_init_stack_spmd));
+
CGF.EmitBranch(ExecuteBB);
CGF.EmitBlock(ExecuteBB);
@@ -1107,11 +1115,6 @@
// Wait for parallel work
syncCTAThreads(CGF);
- // For data sharing, we need to initialize the stack for workers.
- CGF.EmitRuntimeCall(
- createNVPTXRuntimeFunction(
- OMPRTL_NVPTX__kmpc_data_sharing_init_stack));
-
Address WorkFn =
CGF.CreateDefaultAlignTempAlloca(CGF.Int8PtrTy, /*Name=*/"work_fn");
Address ExecStatus =
@@ -1417,6 +1420,13 @@
RTLFn = CGM.CreateRuntimeFunction(FnTy, "__kmpc_data_sharing_init_stack");
break;
}
+ case OMPRTL_NVPTX__kmpc_data_sharing_init_stack_spmd: {
+/// Build void __kmpc_data_sharing_init_stack_spmd();
+auto *FnTy =
+llvm::FunctionType::get(CGM.VoidTy, llvm::None, /*isVarArg*/ false);
+RTLFn = CGM.CreateRuntimeFunction(FnTy,
"__kmpc_data_sharing_init_stack_spmd");
+break;
+ }
case OMPRTL_NVPTX__kmpc_data_sharing_push_stack: {
// Build void *__kmpc_data_sharing_push_stack(size_t size,
// int16_t UseSharedMemory);
Index: test/OpenMP/nvptx_data_sharing_spmd.cpp
===
--- /dev/null
+++ test/OpenMP/nvptx_data_sharing_spmd.cpp
@@ -0,0 +1,24 @@
+// Test device global memory data sharing initialization codegen for spmd.
+///==///
+
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple nvptx64-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix CK1
+
+// expected-no-diagnostics
+
+#ifndef HEADER
+#define HEADER
+
+void test_ds(){
+ #pragma omp target teams distribute parallel for
+ for (int i = 0; i < 100; i++ ) {
+int a = 100;
+ }
+}
+
+// CK1: {{.*}}define weak void @__omp_offloading{{.*}}test_ds{{.*}}()
+// CK1: call void @__kmpc_spmd_kernel_init
+// CK1-NEXT: call void
[PATCH] D49188: [OpenMP] Initialize data sharing stack for SPMD case
Hahnfeld added inline comments. Comment at: lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp:84-85 OMPRTL_NVPTX__kmpc_data_sharing_init_stack, + /// Call to void __kmpc_data_sharing_init_stack_spmd(); + OMPRTL_NVPTX__kmpc_data_sharing_init_stack_spmd, /// Call to void* __kmpc_data_sharing_push_stack(size_t size, In case I'm not missing something obvious this function doesn't exist (yet) in libomptarget-nvptx? Comment at: lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp:1110-1114 - // For data sharing, we need to initialize the stack for workers. - CGF.EmitRuntimeCall( - createNVPTXRuntimeFunction( - OMPRTL_NVPTX__kmpc_data_sharing_init_stack)); - Why is this call being removed? There's no mention in the summary AFAICS Repository: rC Clang https://reviews.llvm.org/D49188 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D49188: [OpenMP] Initialize data sharing stack for SPMD case
gtbercea updated this revision to Diff 155002.
gtbercea added a comment.
Fix test.
Repository:
rC Clang
https://reviews.llvm.org/D49188
Files:
lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
test/OpenMP/nvptx_data_sharing.cpp
Index: test/OpenMP/nvptx_data_sharing.cpp
===
--- test/OpenMP/nvptx_data_sharing.cpp
+++ test/OpenMP/nvptx_data_sharing.cpp
@@ -30,7 +30,6 @@
/// = In the worker function = ///
// CK1: {{.*}}define internal void
@__omp_offloading{{.*}}test_ds{{.*}}_worker()
// CK1: call void @llvm.nvvm.barrier0()
-// CK1: call void @__kmpc_data_sharing_init_stack
/// = In the kernel function = ///
Index: lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
===
--- lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
+++ lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
@@ -81,6 +81,8 @@
OMPRTL_NVPTX__kmpc_end_reduce_nowait,
/// Call to void __kmpc_data_sharing_init_stack();
OMPRTL_NVPTX__kmpc_data_sharing_init_stack,
+ /// Call to void __kmpc_data_sharing_init_stack_spmd();
+ OMPRTL_NVPTX__kmpc_data_sharing_init_stack_spmd,
/// Call to void* __kmpc_data_sharing_push_stack(size_t size,
/// int16_t UseSharedMemory);
OMPRTL_NVPTX__kmpc_data_sharing_push_stack,
@@ -1025,6 +1027,12 @@
/*RequiresDataSharing=*/Bld.getInt16(1)};
CGF.EmitRuntimeCall(
createNVPTXRuntimeFunction(OMPRTL_NVPTX__kmpc_spmd_kernel_init), Args);
+
+ // For data sharing, we need to initialize the stack.
+ CGF.EmitRuntimeCall(
+ createNVPTXRuntimeFunction(
+ OMPRTL_NVPTX__kmpc_data_sharing_init_stack_spmd));
+
CGF.EmitBranch(ExecuteBB);
CGF.EmitBlock(ExecuteBB);
@@ -1107,11 +1115,6 @@
// Wait for parallel work
syncCTAThreads(CGF);
- // For data sharing, we need to initialize the stack for workers.
- CGF.EmitRuntimeCall(
- createNVPTXRuntimeFunction(
- OMPRTL_NVPTX__kmpc_data_sharing_init_stack));
-
Address WorkFn =
CGF.CreateDefaultAlignTempAlloca(CGF.Int8PtrTy, /*Name=*/"work_fn");
Address ExecStatus =
@@ -1417,6 +1420,13 @@
RTLFn = CGM.CreateRuntimeFunction(FnTy, "__kmpc_data_sharing_init_stack");
break;
}
+ case OMPRTL_NVPTX__kmpc_data_sharing_init_stack_spmd: {
+/// Build void __kmpc_data_sharing_init_stack_spmd();
+auto *FnTy =
+llvm::FunctionType::get(CGM.VoidTy, llvm::None, /*isVarArg*/ false);
+RTLFn = CGM.CreateRuntimeFunction(FnTy,
"__kmpc_data_sharing_init_stack_spmd");
+break;
+ }
case OMPRTL_NVPTX__kmpc_data_sharing_push_stack: {
// Build void *__kmpc_data_sharing_push_stack(size_t size,
// int16_t UseSharedMemory);
Index: test/OpenMP/nvptx_data_sharing.cpp
===
--- test/OpenMP/nvptx_data_sharing.cpp
+++ test/OpenMP/nvptx_data_sharing.cpp
@@ -30,7 +30,6 @@
/// = In the worker function = ///
// CK1: {{.*}}define internal void @__omp_offloading{{.*}}test_ds{{.*}}_worker()
// CK1: call void @llvm.nvvm.barrier0()
-// CK1: call void @__kmpc_data_sharing_init_stack
/// = In the kernel function = ///
Index: lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
===
--- lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
+++ lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
@@ -81,6 +81,8 @@
OMPRTL_NVPTX__kmpc_end_reduce_nowait,
/// Call to void __kmpc_data_sharing_init_stack();
OMPRTL_NVPTX__kmpc_data_sharing_init_stack,
+ /// Call to void __kmpc_data_sharing_init_stack_spmd();
+ OMPRTL_NVPTX__kmpc_data_sharing_init_stack_spmd,
/// Call to void* __kmpc_data_sharing_push_stack(size_t size,
/// int16_t UseSharedMemory);
OMPRTL_NVPTX__kmpc_data_sharing_push_stack,
@@ -1025,6 +1027,12 @@
/*RequiresDataSharing=*/Bld.getInt16(1)};
CGF.EmitRuntimeCall(
createNVPTXRuntimeFunction(OMPRTL_NVPTX__kmpc_spmd_kernel_init), Args);
+
+ // For data sharing, we need to initialize the stack.
+ CGF.EmitRuntimeCall(
+ createNVPTXRuntimeFunction(
+ OMPRTL_NVPTX__kmpc_data_sharing_init_stack_spmd));
+
CGF.EmitBranch(ExecuteBB);
CGF.EmitBlock(ExecuteBB);
@@ -1107,11 +1115,6 @@
// Wait for parallel work
syncCTAThreads(CGF);
- // For data sharing, we need to initialize the stack for workers.
- CGF.EmitRuntimeCall(
- createNVPTXRuntimeFunction(
- OMPRTL_NVPTX__kmpc_data_sharing_init_stack));
-
Address WorkFn =
CGF.CreateDefaultAlignTempAlloca(CGF.Int8PtrTy, /*Name=*/"work_fn");
Address ExecStatus =
@@ -1417,6 +1420,13 @@
RTLFn = CGM.CreateRuntimeFunction(FnTy, "__kmpc_data_sharing_init_stack");
break;
}
+ case OMPRTL_NVPTX__kmpc_data_sharing_init_stack_spmd: {
+/// Build void __kmpc_data_sharing_init_stack_spmd();
+auto *FnTy =
+llvm::FunctionType::get(CGM.VoidTy, llvm::None, /*isVarArg*/ false);
+
[PATCH] D49188: [OpenMP] Initialize data sharing stack for SPMD case
ABataev added a comment. What about the check for the new function call? Repository: rC Clang https://reviews.llvm.org/D49188 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D49188: [OpenMP] Initialize data sharing stack for SPMD case
ABataev added a comment. Tests? Repository: rC Clang https://reviews.llvm.org/D49188 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
