From 81a344e21ecb3eba36f1f47922cfd565c89f1081 Mon Sep 17 00:00:00 2001 From: Hans Wennborg Date: Tue, 28 Jul 2015 16:47:18 +0000 Subject: [PATCH] Merging r243277, r243280, r243285, and r243289 ------------------------------------------------------------------------ r243277 | sfantao | 2015-07-27 09:38:06 -0700 (Mon, 27 Jul 2015) | 4 lines [OpenMP] Add capture for threadprivate variables used in copyin clause if TLS is enabled in OpenMP code generation. ------------------------------------------------------------------------ ------------------------------------------------------------------------ r243280 | sfantao | 2015-07-27 09:59:45 -0700 (Mon, 27 Jul 2015) | 4 lines [OpenMP] Add TLS requirement for the copyin clause codegen test. This is an attempt to fix regressions triggered by r243277. ------------------------------------------------------------------------ ------------------------------------------------------------------------ r243285 | sfantao | 2015-07-27 10:30:41 -0700 (Mon, 27 Jul 2015) | 2 lines [OpenMP] Fix copyin clause codegen regression caused by r243277. ------------------------------------------------------------------------ ------------------------------------------------------------------------ r243289 | sfantao | 2015-07-27 10:49:18 -0700 (Mon, 27 Jul 2015) | 3 lines [OpenMP] Fix copyin codegen test regression in order used in compare instruction. ------------------------------------------------------------------------ git-svn-id: https://llvm.org/svn/llvm-project/cfe/branches/release_37@243443 91177308-0d34-0410-b5e6-96231b3b80d8 --- lib/CodeGen/CGStmtOpenMP.cpp | 20 +- lib/Sema/SemaOpenMP.cpp | 17 +- test/OpenMP/parallel_copyin_codegen.cpp | 259 +++++++++++++++++++++++- 3 files changed, 287 insertions(+), 9 deletions(-) diff --git a/lib/CodeGen/CGStmtOpenMP.cpp b/lib/CodeGen/CGStmtOpenMP.cpp index a1f093a8d8..55f76cea9c 100644 --- a/lib/CodeGen/CGStmtOpenMP.cpp +++ b/lib/CodeGen/CGStmtOpenMP.cpp @@ -227,10 +227,22 @@ bool CodeGenFunction::EmitOMPCopyinClause(const OMPExecutableDirective &D) { auto *VD = cast(cast(*IRef)->getDecl()); QualType Type = VD->getType(); if (CopiedVars.insert(VD->getCanonicalDecl()).second) { - // Get the address of the master variable. - auto *MasterAddr = VD->isStaticLocal() - ? CGM.getStaticLocalDeclAddress(VD) - : CGM.GetAddrOfGlobal(VD); + + // Get the address of the master variable. If we are emitting code with + // TLS support, the address is passed from the master as field in the + // captured declaration. + llvm::Value *MasterAddr; + if (getLangOpts().OpenMPUseTLS && + getContext().getTargetInfo().isTLSSupported()) { + assert(CapturedStmtInfo->lookup(VD) && + "Copyin threadprivates should have been captured!"); + DeclRefExpr DRE(const_cast(VD), true, (*IRef)->getType(), + VK_LValue, (*IRef)->getExprLoc()); + MasterAddr = EmitLValue(&DRE).getAddress(); + } else { + MasterAddr = VD->isStaticLocal() ? CGM.getStaticLocalDeclAddress(VD) + : CGM.GetAddrOfGlobal(VD); + } // Get the address of the threadprivate variable. auto *PrivateAddr = EmitLValue(*IRef).getAddress(); if (CopiedVars.size() == 1) { diff --git a/lib/Sema/SemaOpenMP.cpp b/lib/Sema/SemaOpenMP.cpp index e11ab82f88..77e120bd60 100644 --- a/lib/Sema/SemaOpenMP.cpp +++ b/lib/Sema/SemaOpenMP.cpp @@ -120,6 +120,7 @@ private: /// from current directive. OpenMPClauseKind ClauseKindMode; Sema &SemaRef; + bool ForceCapturing; typedef SmallVector::reverse_iterator reverse_iterator; @@ -130,11 +131,15 @@ private: public: explicit DSAStackTy(Sema &S) - : Stack(1), ClauseKindMode(OMPC_unknown), SemaRef(S) {} + : Stack(1), ClauseKindMode(OMPC_unknown), SemaRef(S), + ForceCapturing(false) {} bool isClauseParsingMode() const { return ClauseKindMode != OMPC_unknown; } void setClauseParsingMode(OpenMPClauseKind K) { ClauseKindMode = K; } + bool isForceVarCapturing() const { return ForceCapturing; } + void setForceVarCapturing(bool V) { ForceCapturing = V; } + void push(OpenMPDirectiveKind DKind, const DeclarationNameInfo &DirName, Scope *CurScope, SourceLocation Loc) { Stack.push_back(SharingMapTy(DKind, DirName, CurScope, Loc)); @@ -655,7 +660,8 @@ bool Sema::IsOpenMPCapturedVar(VarDecl *VD) { if (DSAStack->getCurrentDirective() != OMPD_unknown) { if (DSAStack->isLoopControlVariable(VD) || (VD->hasLocalStorage() && - isParallelOrTaskRegion(DSAStack->getCurrentDirective()))) + isParallelOrTaskRegion(DSAStack->getCurrentDirective())) || + DSAStack->isForceVarCapturing()) return true; auto DVarPrivate = DSAStack->getTopDSA(VD, DSAStack->isClauseParsingMode()); if (DVarPrivate.CKind != OMPC_unknown && isOpenMPPrivate(DVarPrivate.CKind)) @@ -1350,13 +1356,18 @@ StmtResult Sema::ActOnOpenMPRegionEnd(StmtResult S, // This is required for proper codegen. for (auto *Clause : Clauses) { if (isOpenMPPrivate(Clause->getClauseKind()) || - Clause->getClauseKind() == OMPC_copyprivate) { + Clause->getClauseKind() == OMPC_copyprivate || + (getLangOpts().OpenMPUseTLS && + getASTContext().getTargetInfo().isTLSSupported() && + Clause->getClauseKind() == OMPC_copyin)) { + DSAStack->setForceVarCapturing(Clause->getClauseKind() == OMPC_copyin); // Mark all variables in private list clauses as used in inner region. for (auto *VarRef : Clause->children()) { if (auto *E = cast_or_null(VarRef)) { MarkDeclarationsReferencedInExpr(E); } } + DSAStack->setForceVarCapturing(/*V=*/false); } else if (isParallelOrTaskRegion(DSAStack->getCurrentDirective()) && Clause->getClauseKind() == OMPC_schedule) { // Mark all variables in private list clauses as used in inner region. diff --git a/test/OpenMP/parallel_copyin_codegen.cpp b/test/OpenMP/parallel_copyin_codegen.cpp index 281d82ce97..4aacb80957 100644 --- a/test/OpenMP/parallel_copyin_codegen.cpp +++ b/test/OpenMP/parallel_copyin_codegen.cpp @@ -4,6 +4,14 @@ // RUN: %clang_cc1 -verify -fopenmp -fnoopenmp-use-tls -x c++ -std=c++11 -DLAMBDA -triple %itanium_abi_triple -emit-llvm %s -o - | FileCheck -check-prefix=LAMBDA %s // RUN: %clang_cc1 -verify -fopenmp -fnoopenmp-use-tls -x c++ -fblocks -DBLOCKS -triple %itanium_abi_triple -emit-llvm %s -o - | FileCheck -check-prefix=BLOCKS %s // RUN: %clang_cc1 -verify -fopenmp -fnoopenmp-use-tls -x c++ -std=c++11 -DARRAY -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -check-prefix=ARRAY %s + +// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple %itanium_abi_triple -emit-llvm %s -o - | FileCheck %s -check-prefix=TLS-CHECK +// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple %itanium_abi_triple -emit-pch -o %t %s +// RUN: %clang_cc1 -fopenmp -x c++ -triple %itanium_abi_triple -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s -check-prefix=TLS-CHECK +// RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -DLAMBDA -triple %itanium_abi_triple -emit-llvm %s -o - | FileCheck -check-prefix=TLS-LAMBDA %s +// RUN: %clang_cc1 -verify -fopenmp -x c++ -fblocks -DBLOCKS -triple %itanium_abi_triple -emit-llvm %s -o - | FileCheck -check-prefix=TLS-BLOCKS %s +// RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -DARRAY -triple x86_64-linux-gnu -emit-llvm %s -o - | FileCheck -check-prefix=TLS-ARRAY %s +// REQUIRES: tls // expected-no-diagnostics #ifndef ARRAY #ifndef HEADER @@ -25,7 +33,9 @@ struct S { // CHECK-DAG: [[S_FLOAT_TY:%.+]] = type { float } // CHECK-DAG: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} } // CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8* - +// TLS-CHECK-DAG: [[S_FLOAT_TY:%.+]] = type { float } +// TLS-CHECK-DAG: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} } +// TLS-CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8* // CHECK-DAG: [[T_VAR:@.+]] = internal global i{{[0-9]+}} 1122, // CHECK-DAG: [[VEC:@.+]] = internal global [2 x i{{[0-9]+}}] [i{{[0-9]+}} 1, i{{[0-9]+}} 2], @@ -35,6 +45,15 @@ struct S { // CHECK-DAG: [[TMAIN_VEC:@.+]] = linkonce_odr global [2 x i{{[0-9]+}}] [i{{[0-9]+}} 3, i{{[0-9]+}} 3], // CHECK-DAG: [[TMAIN_S_ARR:@.+]] = linkonce_odr global [2 x [[S_INT_TY]]] zeroinitializer, // CHECK-DAG: [[TMAIN_VAR:@.+]] = linkonce_odr global [[S_INT_TY]] zeroinitializer, +// TLS-CHECK-DAG: [[T_VAR:@.+]] = internal thread_local global i{{[0-9]+}} 1122, +// TLS-CHECK-DAG: [[VEC:@.+]] = internal thread_local global [2 x i{{[0-9]+}}] [i{{[0-9]+}} 1, i{{[0-9]+}} 2], +// TLS-CHECK-DAG: [[S_ARR:@.+]] = internal thread_local global [2 x [[S_FLOAT_TY]]] zeroinitializer, +// TLS-CHECK-DAG: [[VAR:@.+]] = internal thread_local global [[S_FLOAT_TY]] zeroinitializer, +// TLS-CHECK-DAG: [[TMAIN_T_VAR:@.+]] = linkonce_odr thread_local global i{{[0-9]+}} 333, +// TLS-CHECK-DAG: [[TMAIN_VEC:@.+]] = linkonce_odr thread_local global [2 x i{{[0-9]+}}] [i{{[0-9]+}} 3, i{{[0-9]+}} 3], +// TLS-CHECK-DAG: [[TMAIN_S_ARR:@.+]] = linkonce_odr thread_local global [2 x [[S_INT_TY]]] zeroinitializer, +// TLS-CHECK-DAG: [[TMAIN_VAR:@.+]] = linkonce_odr thread_local global [[S_INT_TY]] zeroinitializer, + template T tmain() { S test; @@ -59,12 +78,28 @@ int main() { // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212, // LAMBDA-LABEL: @main // LAMBDA: call{{.*}} void [[OUTER_LAMBDA:@.+]]( + // TLS-LAMBDA: [[G_CAPTURE_TY:%.+]] = type { i{{[0-9]+}}* } + // TLS-LAMBDA: [[G:@.+]] = {{.*}}thread_local {{.*}}global i{{[0-9]+}} 1212, + // TLS-LAMBDA-LABEL: @main + // TLS-LAMBDA: call{{.*}} void [[OUTER_LAMBDA:@.+]]( [&]() { // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]]( // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* + + // TLS-LAMBDA-DAG: [[G_CPY_ADDR:%.+]] = getelementptr inbounds [[G_CAPTURE_TY]], [[G_CAPTURE_TY]]* [[G_CAPTURE:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 + // TLS-LAMBDA-DAG: [[G_CPY_VAL:%.+]] = call i{{[0-9]+}}* [[G_CTOR:@.+]]() + // TLS-LAMBDA: store i{{[0-9]+}}* [[G_CPY_VAL]], i{{[0-9]+}}** [[G_CPY_ADDR]] + // TLS-LAMBDA: [[G_CAPTURE_CAST:%.+]] = bitcast [[G_CAPTURE_TY]]* [[G_CAPTURE]] to i8* + // TLS-LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* [[G_CAPTURE_CAST]] + + // TLS-LAMBDA: define {{.*}}i{{[0-9]+}}* [[G_CTOR]]() { + // TLS-LAMBDA: ret i{{[0-9]+}}* [[G]] + // TLS-LAMBDA: } + #pragma omp parallel copyin(g) { // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]]) + // TLS-LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, [[G_CAPTURE_TY]]* [[ARG:%.+]]) // threadprivate_g = g; // LAMBDA: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[G]] @@ -76,14 +111,31 @@ int main() { // LAMBDA: store volatile i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* %{{.+}}, // LAMBDA: [[DONE]] + // TLS-LAMBDA-DAG: [[G_CAPTURE_FIELD:%.+]] = getelementptr inbounds [[G_CAPTURE_TY]], [[G_CAPTURE_TY]]* {{.*}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0 + // TLS-LAMBDA-DAG: [[G_CAPTURE_SRC:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_CAPTURE_FIELD]] + // TLS-LAMBDA-DAG: [[G_CAPTURE_DST:%.+]] = call i{{[0-9]+}}* [[G_CTOR]]() + // TLS-LAMBDA-DAG: [[G_CAPTURE_SRCC:%.+]] = ptrtoint i{{[0-9]+}}* [[G_CAPTURE_SRC]] to i{{[0-9]+}} + // TLS-LAMBDA-DAG: [[G_CAPTURE_DSTC:%.+]] = ptrtoint i{{[0-9]+}}* [[G_CAPTURE_DST]] to i{{[0-9]+}} + // TLS-LAMBDA: icmp ne i{{[0-9]+}} {{%.+}}, {{%.+}} + // TLS-LAMBDA: br i1 %{{.+}}, label %[[NOT_MASTER:.+]], label %[[DONE:.+]] + // TLS-LAMBDA: [[NOT_MASTER]] + // TLS-LAMBDA: load i{{[0-9]+}}, i{{[0-9]+}}* [[G_CAPTURE_SRC]], + // TLS-LAMBDA: store volatile i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* [[G_CAPTURE_DST]], + // TLS-LAMBDA: [[DONE]] + // LAMBDA: call {{.*}}i32 @__kmpc_cancel_barrier( + // TLS-LAMBDA: call {{.*}}i32 @__kmpc_cancel_barrier( g = 1; // LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}* + // TLS-LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}* [&]() { // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]]) // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]], g = 2; // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]] + + // TLS-LAMBDA: [[G_CAPTURE_DST:%.+]] = call i{{[0-9]+}}* [[G_CTOR]]() + // TLS-LAMBDA: store volatile i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_CAPTURE_DST]] }(); } }(); @@ -92,12 +144,28 @@ int main() { // BLOCKS: [[G:@.+]] = global i{{[0-9]+}} 1212, // BLOCKS-LABEL: @main // BLOCKS: call {{.*}}void {{%.+}}(i8 + + // TLS-BLOCKS: [[G_CAPTURE_TY:%.+]] = type { i{{[0-9]+}}* } + // TLS-BLOCKS: [[G:@.+]] = {{.*}}thread_local {{.*}}global i{{[0-9]+}} 1212, + // TLS-BLOCKS-LABEL: @main + // TLS-BLOCKS: call {{.*}}void {{%.+}}(i8 ^{ // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8* // BLOCKS: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* + + // TLS-BLOCKS-DAG: [[G_CPY_ADDR:%.+]] = getelementptr inbounds [[G_CAPTURE_TY]], [[G_CAPTURE_TY]]* [[G_CAPTURE:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 + // TLS-BLOCKS-DAG: [[G_CPY_VAL:%.+]] = call i{{[0-9]+}}* [[G_CTOR:@.+]]() + // TLS-BLOCKS: store i{{[0-9]+}}* [[G_CPY_VAL]], i{{[0-9]+}}** [[G_CPY_ADDR]] + // TLS-BLOCKS: [[G_CAPTURE_CAST:%.+]] = bitcast [[G_CAPTURE_TY]]* [[G_CAPTURE]] to i8* + // TLS-BLOCKS: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* [[G_CAPTURE_CAST]] + + // TLS-BLOCKS: define {{.*}}i{{[0-9]+}}* [[G_CTOR]]() { + // TLS-BLOCKS: ret i{{[0-9]+}}* [[G]] + // TLS-BLOCKS: } #pragma omp parallel copyin(g) { // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]]) + // TLS-BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, [[G_CAPTURE_TY]]* [[ARG:%.+]]) // threadprivate_g = g; // BLOCKS: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[G]] @@ -109,19 +177,44 @@ int main() { // BLOCKS: store volatile i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* %{{.+}}, // BLOCKS: [[DONE]] + // TLS-BLOCKS-DAG: [[G_CAPTURE_FIELD:%.+]] = getelementptr inbounds [[G_CAPTURE_TY]], [[G_CAPTURE_TY]]* {{.*}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0 + // TLS-BLOCKS-DAG: [[G_CAPTURE_SRC:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_CAPTURE_FIELD]] + // TLS-BLOCKS-DAG: [[G_CAPTURE_DST:%.+]] = call i{{[0-9]+}}* [[G_CTOR]]() + // TLS-BLOCKS-DAG: [[G_CAPTURE_SRCC:%.+]] = ptrtoint i{{[0-9]+}}* [[G_CAPTURE_SRC]] to i{{[0-9]+}} + // TLS-BLOCKS-DAG: [[G_CAPTURE_DSTC:%.+]] = ptrtoint i{{[0-9]+}}* [[G_CAPTURE_DST]] to i{{[0-9]+}} + // TLS-BLOCKS: icmp ne i{{[0-9]+}} {{%.+}}, {{%.+}} + // TLS-BLOCKS: br i1 %{{.+}}, label %[[NOT_MASTER:.+]], label %[[DONE:.+]] + // TLS-BLOCKS: [[NOT_MASTER]] + // TLS-BLOCKS: load i{{[0-9]+}}, i{{[0-9]+}}* [[G_CAPTURE_SRC]], + // TLS-BLOCKS: store volatile i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* [[G_CAPTURE_DST]], + // TLS-BLOCKS: [[DONE]] + // BLOCKS: call {{.*}}i32 @__kmpc_cancel_barrier( + // TLS-BLOCKS: call {{.*}}i32 @__kmpc_cancel_barrier( g = 1; // BLOCKS: store volatile i{{[0-9]+}} 1, i{{[0-9]+}}* // BLOCKS-NOT: [[G]]{{[[^:word:]]}} // BLOCKS: call {{.*}}void {{%.+}}(i8 + + // TLS-BLOCKS: [[G_CAPTURE_DST:%.+]] = call i{{[0-9]+}}* [[G_CTOR]]() + // TLS-BLOCKS: store volatile i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_CAPTURE_DST]] + // TLS-BLOCKS-NOT: [[G]]{{[[^:word:]]}} + // TLS-BLOCKS: call {{.*}}void {{%.+}}(i8 ^{ // BLOCKS: define {{.+}} void {{@.+}}(i8* + // TLS-BLOCKS: define {{.+}} void {{@.+}}(i8* g = 2; // BLOCKS-NOT: [[G]]{{[[^:word:]]}} // BLOCKS: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[G]] // BLOCKS: store volatile i{{[0-9]+}} 2, i{{[0-9]+}}* // BLOCKS-NOT: [[G]]{{[[^:word:]]}} // BLOCKS: ret + + // TLS-BLOCKS-NOT: [[G]]{{[[^:word:]]}} + // TLS-BLOCKS: [[G_CAPTURE_DST:%.+]] = call i{{[0-9]+}}* [[G_CTOR]]() + // TLS-BLOCKS: store volatile i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_CAPTURE_DST]] + // TLS-BLOCKS-NOT: [[G]]{{[[^:word:]]}} + // TLS-BLOCKS: ret }(); } }(); @@ -153,12 +246,34 @@ int main() { // CHECK: = call {{.*}}i{{.+}} [[TMAIN_INT:@.+]]() // CHECK: call {{.*}} [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]* // CHECK: ret -// + +// TLS-CHECK-LABEL: @main +// TLS-CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]], +// TLS-CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN:@.+]]([[S_FLOAT_TY]]* [[TEST]], [[S_FLOAT_TY]]* +// TLS-CHECK-DAG: store i{{[0-9]+}}* [[T_VAR]], i{{[0-9]+}}** [[T_VAR_FIELD:%.+]], +// TLS-CHECK-DAG: store [2 x i{{[0-9]+}}]* [[VEC]], [2 x i{{[0-9]+}}]** [[VEC_FIELD:%.+]], +// TLS-CHECK-DAG: store [2 x [[S_FLOAT_TY]]]* [[S_ARR]], [2 x [[S_FLOAT_TY]]]** [[S_ARR_FIELD:%.+]], +// TLS-CHECK-DAG: store [[S_FLOAT_TY]]* [[VAR]], [[S_FLOAT_TY]]** [[VAR_FIELD:%.+]], +// TLS-CHECK-DAG: [[T_VAR_FIELD]] = getelementptr inbounds {{.*}} +// TLS-CHECK-DAG: [[VEC_FIELD]] = getelementptr inbounds {{.*}} +// TLS-CHECK-DAG: [[S_ARR_FIELD]] = getelementptr inbounds {{.*}} +// TLS-CHECK-DAG: [[VAR_FIELD]] = getelementptr inbounds {{.*}} +// TLS-CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, {{%.+}}*)* [[MAIN_MICROTASK:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}}) +// TLS-CHECK-DAG: store i{{[0-9]+}}* [[T_VAR]], i{{[0-9]+}}** [[T_VAR_FIELD:%.+]], +// TLS-CHECK-DAG: [[T_VAR_FIELD]] = getelementptr inbounds {{.*}} +// TLS-CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, {{%.+}}*)* [[MAIN_MICROTASK1:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}}) +// TLS-CHECK: = call {{.*}}i{{.+}} [[TMAIN_INT:@.+]]() +// TLS-CHECK: call {{.*}} [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]* +// TLS-CHECK: ret + // CHECK: define internal {{.*}}void [[MAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, {{%.+}}* %{{.+}}) // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]], // CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_ADDR]], // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]], +// TLS-CHECK: define internal {{.*}}void [[MAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, {{%.+}}* %{{.+}}) +// TLS-CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]], + // threadprivate_t_var = t_var; // CHECK: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[T_VAR]] // CHECK: ptrtoint i{{[0-9]+}}* %{{.+}} to i{{[0-9]+}} @@ -168,10 +283,24 @@ int main() { // CHECK: load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR]], // CHECK: store i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* %{{.+}}, +// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}} +// TLS-CHECK: [[MASTER_REF:%.+]] = load i32*, i32** [[MASTER_FIELD]] +// TLS-CHECK: [[MASTER_LONG:%.+]] = ptrtoint i32* [[MASTER_REF]] to i{{[0-9]+}} +// TLS-CHECK: icmp ne i{{[0-9]+}} [[MASTER_LONG]], ptrtoint (i{{[0-9]+}}* [[T_VAR]] to i{{[0-9]+}}) +// TLS-CHECK: br i1 %{{.+}}, label %[[NOT_MASTER:.+]], label %[[DONE:.+]] +// TLS-CHECK: [[NOT_MASTER]] +// TLS-CHECK: [[MASTER_VAL:%.+]] = load i32, i32* [[MASTER_REF]] +// TLS-CHECK: store i32 [[MASTER_VAL]], i32* [[T_VAR]] + // threadprivate_vec = vec; // CHECK: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[VEC]] // CHECK: call void @llvm.memcpy{{.*}}(i8* %{{.+}}, i8* bitcast ([2 x i{{[0-9]+}}]* [[VEC]] to i8*), +// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}} +// TLS-CHECK: [[MASTER_REF:%.+]] = load [2 x i32]*, [2 x i32]** [[MASTER_FIELD]] +// TLS-CHECK: [[MASTER_CAST:%.+]] = bitcast [2 x i32]* [[MASTER_REF]] to i8* +// TLS-CHECK: call void @llvm.memcpy{{.*}}(i8* bitcast ([2 x i{{[0-9]+}}]* [[VEC]] to i8*), i8* [[MASTER_CAST]] + // threadprivate_s_arr = s_arr; // CHECK: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[S_ARR]] // CHECK: [[S_ARR_PRIV_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* {{%.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0 @@ -182,19 +311,43 @@ int main() { // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN]]([[S_FLOAT_TY]]* {{.+}}, [[S_FLOAT_TY]]* {{.+}}) // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]] +// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}} +// TLS-CHECK: [[MASTER_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** [[MASTER_FIELD]] +// TLS-CHECK: [[MASTER_CAST:%.+]] = bitcast [2 x [[S_FLOAT_TY]]]* [[MASTER_REF]] to [[S_FLOAT_TY]]* +// TLS-CHECK-DAG: [[S_ARR_SRC_BEGIN:%.+]] = phi [[S_FLOAT_TY]]* {{.*}}[[MASTER_CAST]] +// TLS-CHECK-DAG: [[S_ARR_DST_BEGIN:%.+]] = phi [[S_FLOAT_TY]]* {{.*}}getelementptr inbounds ([2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0) +// TLS-CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN]]([[S_FLOAT_TY]]* {{.+}}, [[S_FLOAT_TY]]* {{.+}}) +// TLS-CHECK-DAG: [[S_ARR_SRC_END:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_SRC_BEGIN]], i{{[0-9]+}} 1 +// TLS-CHECK-DAG: [[S_ARR_DST_END:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_DST_BEGIN]], i{{[0-9]+}} 1 +// TLS-CHECK: icmp eq [[S_FLOAT_TY]]* [[S_ARR_DST_END]], getelementptr ([[S_FLOAT_TY]], [[S_FLOAT_TY]]* getelementptr inbounds ([2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0), i{{[0-9]+}} 2) +// TLS-CHECK: br i1 %{{.*}}, label %[[ARR_DONE:.+]], label {{.*}} +// TLS-CHECK: [[ARR_DONE]] + // threadprivate_var = var; // CHECK: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[VAR]] // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN]]([[S_FLOAT_TY]]* {{%.+}}, [[S_FLOAT_TY]]* {{.*}}[[VAR]]) // CHECK: [[DONE]] +// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}} +// TLS-CHECK: [[MASTER_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[MASTER_FIELD]] +// TLS-CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN]]([[S_FLOAT_TY]]* {{.*}}[[VAR]], [[S_FLOAT_TY]]* {{.*}}[[MASTER_REF]]) + // CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]]) // CHECK: ret void +// TLS-CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_ADDR]], +// TLS-CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]], +// TLS-CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]]) +// TLS-CHECK: ret void + // CHECK: define internal {{.*}}void [[MAIN_MICROTASK1]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, {{%.+}}* %{{.+}}) // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]], // CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_ADDR]], // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]], +// TLS-CHECK: define internal {{.*}}void [[MAIN_MICROTASK1]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, {{%.+}}* %{{.+}}) +// TLS-CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]], + // threadprivate_t_var = t_var; // CHECK: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[T_VAR]] // CHECK: ptrtoint i{{[0-9]+}}* %{{.+}} to i{{[0-9]+}} @@ -205,9 +358,24 @@ int main() { // CHECK: store i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* %{{.+}}, // CHECK: [[DONE]] +// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}} +// TLS-CHECK: [[MASTER_REF:%.+]] = load i32*, i32** [[MASTER_FIELD]] +// TLS-CHECK: [[MASTER_LONG:%.+]] = ptrtoint i32* [[MASTER_REF]] to i{{[0-9]+}} +// TLS-CHECK: icmp ne i{{[0-9]+}} [[MASTER_LONG]], ptrtoint (i{{[0-9]+}}* [[T_VAR]] to i{{[0-9]+}}) +// TLS-CHECK: br i1 %{{.+}}, label %[[NOT_MASTER:.+]], label %[[DONE:.+]] +// TLS-CHECK: [[NOT_MASTER]] +// TLS-CHECK: [[MASTER_VAL:%.+]] = load i32, i32* [[MASTER_REF]] +// TLS-CHECK: store i32 [[MASTER_VAL]], i32* [[T_VAR]] +// TLS-CHECK: [[DONE]] + // CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]]) // CHECK: ret void +// TLS-CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_ADDR]], +// TLS-CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]], +// TLS-CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]]) +// TLS-CHECK: ret void + // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]() // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]], // CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN:@.+]]([[S_INT_TY]]* [[TEST]], [[S_INT_TY]]* @@ -215,11 +383,30 @@ int main() { // CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, {{%.+}}*)* [[TMAIN_MICROTASK1:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}}) // CHECK: call {{.*}} [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]* // CHECK: ret + +// TLS-CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]() +// TLS-CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]], +// TLS-CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN:@.+]]([[S_INT_TY]]* [[TEST]], [[S_INT_TY]]* +// TLS-CHECK-DAG: store i{{[0-9]+}}* [[TMAIN_T_VAR]], i{{[0-9]+}}** [[T_VAR_FIELD:%.+]], +// TLS-CHECK-DAG: store [2 x i{{[0-9]+}}]* [[TMAIN_VEC]], [2 x i{{[0-9]+}}]** [[VEC_FIELD:%.+]], +// TLS-CHECK-DAG: store [2 x [[S_INT_TY]]]* [[TMAIN_S_ARR]], [2 x [[S_INT_TY]]]** [[S_ARR_FIELD:%.+]], +// TLS-CHECK-DAG: store [[S_INT_TY]]* [[TMAIN_VAR]], [[S_INT_TY]]** [[VAR_FIELD:%.+]], +// TLS-CHECK-DAG: [[T_VAR_FIELD]] = getelementptr inbounds {{.*}} +// TLS-CHECK-DAG: [[VEC_FIELD]] = getelementptr inbounds {{.*}} +// TLS-CHECK-DAG: [[S_ARR_FIELD]] = getelementptr inbounds {{.*}} +// TLS-CHECK-DAG: [[VAR_FIELD]] = getelementptr inbounds {{.*}} +// TLS-CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, {{%.+}}*)* [[TMAIN_MICROTASK:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}}) +// TLS-CHECK-DAG: store i{{[0-9]+}}* [[TMAIN_T_VAR]], i{{[0-9]+}}** [[T_VAR_FIELD:%.+]], +// TLS-CHECK-DAG: [[T_VAR_FIELD]] = getelementptr inbounds {{.*}} +// TLS-CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, {{%.+}}*)* [[TMAIN_MICROTASK1:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}}) // // CHECK: define internal {{.*}}void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, {{%.+}}* %{{.+}}) // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]], // CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_ADDR]], // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]], +// +// TLS-CHECK: define internal {{.*}}void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, {{%.+}}* %{{.+}}) +// TLS-CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]], // threadprivate_t_var = t_var; // CHECK: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[TMAIN_T_VAR]] @@ -230,10 +417,24 @@ int main() { // CHECK: load i{{[0-9]+}}, i{{[0-9]+}}* [[TMAIN_T_VAR]], // CHECK: store i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* %{{.+}}, +// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}} +// TLS-CHECK: [[MASTER_REF:%.+]] = load i32*, i32** [[MASTER_FIELD]] +// TLS-CHECK: [[MASTER_LONG:%.+]] = ptrtoint i32* [[MASTER_REF]] to i{{[0-9]+}} +// TLS-CHECK: icmp ne i{{[0-9]+}} [[MASTER_LONG]], ptrtoint (i{{[0-9]+}}* [[TMAIN_T_VAR]] to i{{[0-9]+}}) +// TLS-CHECK: br i1 %{{.+}}, label %[[NOT_MASTER:.+]], label %[[DONE:.+]] +// TLS-CHECK: [[NOT_MASTER]] +// TLS-CHECK: [[MASTER_VAL:%.+]] = load i32, i32* [[MASTER_REF]] +// TLS-CHECK: store i32 [[MASTER_VAL]], i32* [[TMAIN_T_VAR]] + // threadprivate_vec = vec; // CHECK: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[TMAIN_VEC]] // CHECK: call {{.*}}void @llvm.memcpy{{.*}}(i8* %{{.+}}, i8* bitcast ([2 x i{{[0-9]+}}]* [[TMAIN_VEC]] to i8*), +// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}} +// TLS-CHECK: [[MASTER_REF:%.+]] = load [2 x i32]*, [2 x i32]** [[MASTER_FIELD]] +// TLS-CHECK: [[MASTER_CAST:%.+]] = bitcast [2 x i32]* [[MASTER_REF]] to i8* +// TLS-CHECK: call void @llvm.memcpy{{.*}}(i8* bitcast ([2 x i{{[0-9]+}}]* [[TMAIN_VEC]] to i8*), i8* [[MASTER_CAST]] + // threadprivate_s_arr = s_arr; // CHECK: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[TMAIN_S_ARR]] // CHECK: [[S_ARR_PRIV_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* {{%.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0 @@ -244,19 +445,43 @@ int main() { // CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN]]([[S_INT_TY]]* {{.+}}, [[S_INT_TY]]* {{.+}}) // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]] +// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}} +// TLS-CHECK: [[MASTER_REF:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[MASTER_FIELD]] +// TLS-CHECK: [[MASTER_CAST:%.+]] = bitcast [2 x [[S_INT_TY]]]* [[MASTER_REF]] to [[S_INT_TY]]* +// TLS-CHECK-DAG: [[S_ARR_SRC_BEGIN:%.+]] = phi [[S_INT_TY]]* {{.*}}[[MASTER_CAST]] +// TLS-CHECK-DAG: [[S_ARR_DST_BEGIN:%.+]] = phi [[S_INT_TY]]* {{.*}}getelementptr inbounds ([2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[TMAIN_S_ARR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0) +// TLS-CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN]]([[S_INT_TY]]* {{.+}}, [[S_INT_TY]]* {{.+}}) +// TLS-CHECK-DAG: [[S_ARR_SRC_END:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_SRC_BEGIN]], i{{[0-9]+}} 1 +// TLS-CHECK-DAG: [[S_ARR_DST_END:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_DST_BEGIN]], i{{[0-9]+}} 1 +// TLS-CHECK: icmp eq [[S_INT_TY]]* [[S_ARR_DST_END]], getelementptr ([[S_INT_TY]], [[S_INT_TY]]* getelementptr inbounds ([2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[TMAIN_S_ARR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0), i{{[0-9]+}} 2) +// TLS-CHECK: br i1 %{{.*}}, label %[[ARR_DONE:.+]], label {{.*}} +// TLS-CHECK: [[ARR_DONE]] + // threadprivate_var = var; // CHECK: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[TMAIN_VAR]] // CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN]]([[S_INT_TY]]* {{%.+}}, [[S_INT_TY]]* {{.*}}[[TMAIN_VAR]]) // CHECK: [[DONE]] +// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}} +// TLS-CHECK: [[MASTER_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[MASTER_FIELD]] +// TLS-CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN]]([[S_INT_TY]]* {{.*}}[[TMAIN_VAR]], [[S_INT_TY]]* {{.*}}[[MASTER_REF]]) + // CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]]) // CHECK: ret void +// TLS-CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_ADDR]], +// TLS-CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]], +// TLS-CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]]) +// TLS-CHECK: ret void + // CHECK: define internal {{.*}}void [[TMAIN_MICROTASK1]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, {{%.+}}* %{{.+}}) // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]], // CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_ADDR]], // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]], +// TLS-CHECK: define internal {{.*}}void [[TMAIN_MICROTASK1]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, {{%.+}}* %{{.+}}) +// TLS-CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]], + // threadprivate_t_var = t_var; // CHECK: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[TMAIN_T_VAR]] // CHECK: ptrtoint i{{[0-9]+}}* %{{.+}} to i{{[0-9]+}} @@ -267,12 +492,30 @@ int main() { // CHECK: store i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* %{{.+}}, // CHECK: [[DONE]] +// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}} +// TLS-CHECK: [[MASTER_REF:%.+]] = load i32*, i32** [[MASTER_FIELD]] +// TLS-CHECK: [[MASTER_LONG:%.+]] = ptrtoint i32* [[MASTER_REF]] to i{{[0-9]+}} +// TLS-CHECK: icmp ne i{{[0-9]+}} [[MASTER_LONG]], ptrtoint (i{{[0-9]+}}* [[TMAIN_T_VAR]] to i{{[0-9]+}}) +// TLS-CHECK: br i1 %{{.+}}, label %[[NOT_MASTER:.+]], label %[[DONE:.+]] +// TLS-CHECK: [[NOT_MASTER]] +// TLS-CHECK: [[MASTER_VAL:%.+]] = load i32, i32* [[MASTER_REF]] +// TLS-CHECK: store i32 [[MASTER_VAL]], i32* [[TMAIN_T_VAR]] +// TLS-CHECK: [[DONE]] + // CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]]) // CHECK: ret void +// TLS-CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_ADDR]], +// TLS-CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]], +// TLS-CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]]) +// TLS-CHECK: ret void + #endif #else // ARRAY-LABEL: array_func +// TLS-ARRAY: [[ARR_CAPTURETY:%.+]] = type { [2 x {{.*}}]*, [2 x {{.*}}]* } +// TLS-ARRAY-LABEL: array_func + struct St { int a, b; St() : a(0), b(0) {} @@ -287,6 +530,18 @@ void array_func() { // ARRAY: call i8* @__kmpc_threadprivate_cached( // ARRAY: call void @llvm.memcpy.p0i8.p0i8.i64(i8* %{{.+}}, i8* bitcast ([2 x i32]* @{{.+}} to i8*), i64 8, i32 4, i1 false) // ARRAY: call dereferenceable(8) %struct.St* @{{.+}}(%struct.St* %{{.+}}, %struct.St* dereferenceable(8) %{{.+}}) + +// TLS-ARRAY-DAG: [[ARR_CAP1:%.+]] = getelementptr inbounds [[ARR_CAPTURETY]], [[ARR_CAPTURETY]]* {{.*}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0 +// TLS-ARRAY-DAG: [[ARR_CAP2:%.+]] = getelementptr inbounds [[ARR_CAPTURETY]], [[ARR_CAPTURETY]]* {{.*}}, i{{[0-9]+}} 0, i{{[0-9]+}} 1 +// TLS-ARRAY-DAG: store [2 x {{.*}}]* @{{.*}}, [2 x {{.*}}]** [[ARR_CAP1]] +// TLS-ARRAY-DAG: store [2 x {{.*}}]* @{{.*}}, [2 x {{.*}}]** [[ARR_CAP2]] +// TLS-ARRAY: @__kmpc_fork_call( +// TLS-ARRAY-DAG: call void @llvm.memcpy.p0i8.p0i8.i64(i8* bitcast ([2 x i32]* @{{.+}} to i8*), i8* [[REF:%.+]], i64 8, i32 4, i1 false) +// TLS-ARRAY-DAG: [[REF]] = bitcast [2 x i32]* [[REFT:%.+]] to i8* +// TLS-ARRAY-DAG: [[REFT]] = load [2 x i32]*, [2 x i32]** [[FIELDADDR:%.+]], +// TLS-ARRAY-DAG: [[FIELDADDR]] = getelementptr inbounds [[ARR_CAPTURETY]], [[ARR_CAPTURETY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} +// TLS-ARRAY: call dereferenceable(8) %struct.St* @{{.+}}(%struct.St* %{{.+}}, %struct.St* dereferenceable(8) %{{.+}}) + #pragma omp threadprivate(a, s) #pragma omp parallel copyin(a, s) ;