maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[OPENMP] Do not capture private variables in the target regions. 

Private variables are completely redefined in the outlined regions, so
we don't need to capture them. Patch adds this behavior to the
target-based regions.

llvm-svn: 320078
This commit is contained in:
Alexey Bataev 2017-12-07 19:49:28 +00:00
parent 6c96486962
commit 8cf35e4683
6 changed files with 28 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
clang/lib/Sema/SemaExpr.cpp
View File

@ -14619,14 +14619,16 @@ bool Sema::tryCaptureVariable(
// just break here. Similarly, global variables that are captured in a
// target region should not be captured outside the scope of the region.
if (RSI->CapRegionKind == CR_OpenMP) {
auto IsTargetCap = isOpenMPTargetCapturedDecl(Var, RSI->OpenMPLevel);
bool IsOpenMPPrivateDecl = isOpenMPPrivateDecl(Var, RSI->OpenMPLevel);
auto IsTargetCap = !IsOpenMPPrivateDecl &&
isOpenMPTargetCapturedDecl(Var, RSI->OpenMPLevel);
// When we detect target captures we are looking from inside the
// target region, therefore we need to propagate the capture from the
// enclosing region. Therefore, the capture is not initially nested.
if (IsTargetCap)
FunctionScopesIndex--;
if (IsTargetCap || isOpenMPPrivateDecl(Var, RSI->OpenMPLevel)) {
if (IsTargetCap || IsOpenMPPrivateDecl) {
Nested = !IsTargetCap;
DeclRefType = DeclRefType.getUnqualifiedType();
CaptureType = Context.getLValueReferenceType(DeclRefType);

10
clang/test/OpenMP/teams_distribute_parallel_for_private_codegen.cpp
View File

@ -72,13 +72,13 @@ int main() {
// LAMBDA: call void [[OUTER_LAMBDA:@.+]](
[&]() {
// LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
// LAMBDA: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 3, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0), i32 0, i32 0)
// LAMBDA: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 1, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0), i32 0, i32 0)
// LAMBDA: call void @[[LOFFL1:.+]](
// LAMBDA: ret
#pragma omp target
#pragma omp teams distribute parallel for private(g, g1, sivar)
for (int i = 0; i < 2; ++i) {
// LAMBDA: define{{.*}} internal{{.*}} void @[[LOFFL1]](i{{64|32}} {{%.+}}, i{{64|32}} {{%.+}})
// LAMBDA: define{{.*}} internal{{.*}} void @[[LOFFL1]](i{{64|32}} {{%.+}})
// LAMBDA: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 0, {{.+}} @[[LOUTL1:.+]] to {{.+}})
// LAMBDA: ret void
@ -164,12 +164,12 @@ int main() {
}
// CHECK: define {{.*}}i{{[0-9]+}} @main()
// CHECK: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 5, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0), i32 0, i32 0)
// CHECK: call void @[[OFFL1:.+]](i{{64|32}} %{{.+}})
// CHECK: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 0, i8** null, i8** null, i{{64|32}}* null, i64* null, i32 0, i32 0)
// CHECK: call void @[[OFFL1:.+]]()
// CHECK: {{%.+}} = call{{.*}} i32 @[[TMAIN_INT:.+]]()
// CHECK: ret
// CHECK: define{{.*}} void @[[OFFL1]]({{.+}})
// CHECK: define{{.*}} void @[[OFFL1]]()
// CHECK: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 0, {{.+}} @[[OUTL1:.+]] to {{.+}})
// CHECK: ret void

10
clang/test/OpenMP/teams_distribute_parallel_for_simd_private_codegen.cpp
View File

@ -72,13 +72,13 @@ int main() {
// LAMBDA: call void [[OUTER_LAMBDA:@.+]](
[&]() {
// LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
// LAMBDA: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 3, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0)
// LAMBDA: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 1, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0)
// LAMBDA: call void @[[LOFFL1:.+]](
// LAMBDA: ret
#pragma omp target
#pragma omp teams distribute parallel for simd private(g, g1, sivar)
for (int i = 0; i < 2; ++i) {
// LAMBDA: define{{.*}} internal{{.*}} void @[[LOFFL1]](i{{64|32}} {{%.+}}, i{{64|32}} {{%.+}})
// LAMBDA: define{{.*}} internal{{.*}} void @[[LOFFL1]](i{{64|32}} {{%.+}})
// LAMBDA: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 0, {{.+}} @[[LOUTL1:.+]] to {{.+}})
// LAMBDA: ret void
@ -167,12 +167,12 @@ int main() {
}
// CHECK: define {{.*}}i{{[0-9]+}} @main()
// CHECK: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 5, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0)
// CHECK: call void @[[OFFL1:.+]](i{{64|32}} %{{.+}})
// CHECK: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 0, i8** null, i8** null, i{{64|32}}* null, i64* null, i32 0, i32 0)
// CHECK: call void @[[OFFL1:.+]]()
// CHECK: {{%.+}} = call{{.*}} i32 @[[TMAIN_INT:.+]]()
// CHECK: ret
// CHECK: define{{.*}} void @[[OFFL1]]({{.+}})
// CHECK: define{{.*}} void @[[OFFL1]]()
// CHECK: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 0, {{.+}} @[[OUTL1:.+]] to {{.+}})
// CHECK: ret void

11
clang/test/OpenMP/teams_distribute_private_codegen.cpp
View File

@ -72,13 +72,13 @@ int main() {
// LAMBDA: call void [[OUTER_LAMBDA:@.+]](
[&]() {
// LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
// LAMBDA: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 3, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0), i32 0, i32 0)
// LAMBDA: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 1, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0), i32 0, i32 0)
// LAMBDA: call void @[[LOFFL1:.+]](
// LAMBDA: ret
#pragma omp target
#pragma omp teams distribute private(g, g1, sivar)
for (int i = 0; i < 2; ++i) {
// LAMBDA: define{{.*}} internal{{.*}} void @[[LOFFL1]](i{{64|32}} {{%.+}}, i{{64|32}} {{%.+}})
// LAMBDA: define{{.*}} internal{{.*}} void @[[LOFFL1]](i{{64|32}} {{%.+}})
// LAMBDA: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 0, {{.+}} @[[LOUTL1:.+]] to {{.+}})
// LAMBDA: ret void
@ -97,7 +97,6 @@ int main() {
// LAMBDA: [[TMP:%.+]] = alloca i{{[0-9]+}}*,
// LAMBDA: [[SIVAR_PRIV:%.+]] = alloca i{{[0-9]+}},
// LAMBDA: store{{.+}} [[G1_PRIV]], {{.+}} [[TMP]],
g = 1;
g1 = 1;
sivar = 2;
@ -142,12 +141,12 @@ int main() {
}
// CHECK: define {{.*}}i{{[0-9]+}} @main()
// CHECK: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 5, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0), i32 0, i32 0)
// CHECK: call void @[[OFFL1:.+]](i{{64|32}} %{{.+}})
// CHECK: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 0, i8** null, i8** null, i{{64|32}}* null, i64* null, i32 0, i32 0)
// CHECK: call void @[[OFFL1:.+]]()
// CHECK: {{%.+}} = call{{.*}} i32 @[[TMAIN_INT:.+]]()
// CHECK: ret
// CHECK: define{{.*}} void @[[OFFL1]]({{.+}})
// CHECK: define{{.*}} void @[[OFFL1]]()
// CHECK: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 0, {{.+}} @[[OUTL1:.+]] to {{.+}})
// CHECK: ret void

10
clang/test/OpenMP/teams_distribute_simd_private_codegen.cpp
View File

@ -72,13 +72,13 @@ int main() {
// LAMBDA: call void [[OUTER_LAMBDA:@.+]](
[&]() {
// LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
// LAMBDA: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 3, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0), i32 0, i32 0)
// LAMBDA: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 1, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0), i32 0, i32 0)
// LAMBDA: call void @[[LOFFL1:.+]](
// LAMBDA: ret
#pragma omp target
#pragma omp teams distribute simd private(g, g1, sivar)
for (int i = 0; i < 2; ++i) {
// LAMBDA: define{{.*}} internal{{.*}} void @[[LOFFL1]](i{{64|32}} {{%.+}}, i{{64|32}} {{%.+}})
// LAMBDA: define{{.*}} internal{{.*}} void @[[LOFFL1]](i{{64|32}} {{%.+}})
// LAMBDA: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 0, {{.+}} @[[LOUTL1:.+]] to {{.+}})
// LAMBDA: ret void
@ -142,12 +142,12 @@ int main() {
}
// CHECK: define {{.*}}i{{[0-9]+}} @main()
// CHECK: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 5, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0), i32 0, i32 0)
// CHECK: call void @[[OFFL1:.+]](i{{64|32}} %{{.+}})
// CHECK: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 0, i8** null, i8** null, i{{64|32}}* null, i64* null, i32 0, i32 0)
// CHECK: call void @[[OFFL1:.+]]()
// CHECK: {{%.+}} = call{{.*}} i32 @[[TMAIN_INT:.+]]()
// CHECK: ret
// CHECK: define{{.*}} void @[[OFFL1]]({{.+}})
// CHECK: define{{.*}} void @[[OFFL1]]()
// CHECK: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 0, {{.+}} @[[OUTL1:.+]] to {{.+}})
// CHECK: ret void

9
clang/test/OpenMP/teams_private_codegen.cpp
View File

@ -98,7 +98,7 @@ int main() {
// lambda and target region in main
// LAMBDA: define {{.+}} [[OUTER_LAMBDA]]([[CAP_TY]]* {{.+}})
// LAMBDA: call void @[[OMP_OFFLOADING:.+]](i{{[0-9]+}}* {{.+}}, i{{[0-9]+}} {{.+}}
// LAMBDA: call void @[[OMP_OFFLOADING:.+]]()
// target region in struct constructor
// LAMBDA: define{{.*}} void [[ST_CONSTR:@.+]]([[SS_TY]]* %this,
@ -154,7 +154,7 @@ int main() {
#pragma omp target
#pragma omp teams private(g, sivar)
{
// LAMBDA: define{{.+}} @[[OMP_OFFLOADING]](i{{[0-9]+}}* {{.+}} [[G_IN:%.+]], i{{[0-9]+}} [[SIVAR_IN:%.+]]
// LAMBDA: define{{.+}} @[[OMP_OFFLOADING]]()
// LAMBDA: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_teams(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[OMP_OUTLINED_1:@.+]] to void
// LAMBDA: define {{.+}} [[OMP_OUTLINED_1]](i{{[0-9]+}}* {{.+}}, i{{[0-9]+}}* {{.+}}
@ -196,14 +196,13 @@ int main() {
// CHECK: define{{.*}} i{{[0-9]+}} @main()
// CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
// CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
// CHECK: [[OFF_IN:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* {{%.+}},
// CHECK: call void @[[OMP_OFFLOADING:.+]](i{{[0-9]+}} [[OFF_IN]]
// CHECK: call void @[[OMP_OFFLOADING:.+]]()
// CHECK: = call{{.*}} i{{.+}} [[TMAIN_INT:@.+]]()
// CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
// CHECK: ret
// target region in main function
// CHECK: define{{.+}} @[[OMP_OFFLOADING]](i{{[0-9]+}}
// CHECK: define{{.+}} @[[OMP_OFFLOADING]]()
// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_teams(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[OMP_OUTLINED:@.+]] to void
// CHECK: define internal void [[OMP_OUTLINED]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})