// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _ // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1 // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple x86_64-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK2 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple x86_64-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" // RUN: %clang_cc1 -verify -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK5 // RUN: %clang_cc1 -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple x86_64-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK6 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" // RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple x86_64-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK9 // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple aarch64-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK10 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple aarch64-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" // RUN: %clang_cc1 -verify -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK13 // RUN: %clang_cc1 -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple aarch64-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK14 // RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" // RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple aarch64-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" // expected-no-diagnostics #ifndef HEADER #define HEADER void fn1(); void fn2(); void fn3(); void fn4(); void fn5(); void fn6(); int Arg; void gtid_test() { #pragma omp target #pragma omp teams #pragma omp distribute parallel for for(int i = 0 ; i < 100; i++) {} #pragma omp target #pragma omp teams #pragma omp distribute parallel for if (parallel: false) for(int i = 0 ; i < 100; i++) { gtid_test(); } } template int tmain(T Arg) { #pragma omp target #pragma omp teams #pragma omp distribute parallel for if (true) for(int i = 0 ; i < 100; i++) { fn1(); } #pragma omp target #pragma omp teams #pragma omp distribute parallel for if (false) for(int i = 0 ; i < 100; i++) { fn2(); } #pragma omp target #pragma omp teams #pragma omp distribute parallel for if (parallel: Arg) for(int i = 0 ; i < 100; i++) { fn3(); } return 0; } int main() { #pragma omp target #pragma omp teams #pragma omp distribute parallel for if (true) for(int i = 0 ; i < 100; i++) { fn4(); } #pragma omp target #pragma omp teams #pragma omp distribute parallel for if (false) for(int i = 0 ; i < 100; i++) { fn5(); } #pragma omp target #pragma omp teams #pragma omp distribute parallel for if (Arg) for(int i = 0 ; i < 100; i++) { fn6(); } return tmain(Arg); } // call void [[T_OUTLINE_FUN_3:@.+]]( #endif // CHECK1-LABEL: define {{[^@]+}}@_Z9gtid_testv // CHECK1-SAME: () #[[ATTR0:[0-9]+]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK1-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK1-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1: omp_offload.failed: // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47() #[[ATTR2:[0-9]+]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1: omp_offload.cont: // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK1-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK1-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK1: omp_offload.failed2: // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52() #[[ATTR2]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK1: omp_offload.cont3: // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47 // CHECK1-SAME: () #[[ATTR1:[0-9]+]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined. // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1: cond.true: // CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1: cond.false: // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: br label [[COND_END]] // CHECK1: cond.end: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1: omp.inner.for.cond: // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1: omp.inner.for.body: // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1: omp.inner.for.inc: // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1: omp.inner.for.end: // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1: omp.loop.exit: // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1: cond.true: // CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1: cond.false: // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: br label [[COND_END]] // CHECK1: cond.end: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1: omp.inner.for.cond: // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1: omp.inner.for.body: // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK1: omp.body.continue: // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1: omp.inner.for.inc: // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1: omp.inner.for.end: // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1: omp.loop.exit: // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52 // CHECK1-SAME: () #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*)) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..2 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1: cond.true: // CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1: cond.false: // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: br label [[COND_END]] // CHECK1: cond.end: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1: omp.inner.for.cond: // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1: omp.inner.for.body: // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK1-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK1-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: call void @.omp_outlined..3(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK1-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1: omp.inner.for.inc: // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1: omp.inner.for.end: // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1: omp.loop.exit: // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1: cond.true: // CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1: cond.false: // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: br label [[COND_END]] // CHECK1: cond.end: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1: omp.inner.for.cond: // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1: omp.inner.for.body: // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK1-NEXT: call void @_Z9gtid_testv() // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK1: omp.body.continue: // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1: omp.inner.for.inc: // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1: omp.inner.for.end: // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1: omp.loop.exit: // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@main // CHECK1-SAME: () #[[ATTR3:[0-9]+]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK1-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK1-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK1-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1: omp_offload.failed: // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85() #[[ATTR2]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1: omp_offload.cont: // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK1-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK1-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK1: omp_offload.failed2: // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94() #[[ATTR2]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK1: omp_offload.cont3: // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* @Arg, align 4 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32* // CHECK1-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: store i8* null, i8** [[TMP10]], align 8 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK1-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK1-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK1-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK1: omp_offload.failed5: // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103(i64 [[TMP5]]) #[[ATTR2]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK1: omp_offload.cont6: // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* @Arg, align 4 // CHECK1-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP15]]) // CHECK1-NEXT: ret i32 [[CALL]] // // // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85 // CHECK1-SAME: () #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*)) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..4 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1: cond.true: // CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1: cond.false: // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: br label [[COND_END]] // CHECK1: cond.end: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1: omp.inner.for.cond: // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1: omp.inner.for.body: // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1: omp.inner.for.inc: // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1: omp.inner.for.end: // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1: omp.loop.exit: // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1: cond.true: // CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1: cond.false: // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: br label [[COND_END]] // CHECK1: cond.end: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1: omp.inner.for.cond: // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1: omp.inner.for.body: // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK1-NEXT: call void @_Z3fn4v() // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK1: omp.body.continue: // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1: omp.inner.for.inc: // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1: omp.inner.for.end: // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1: omp.loop.exit: // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94 // CHECK1-SAME: () #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*)) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1: cond.true: // CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1: cond.false: // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: br label [[COND_END]] // CHECK1: cond.end: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1: omp.inner.for.cond: // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1: omp.inner.for.body: // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK1-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK1-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: call void @.omp_outlined..7(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK1-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1: omp.inner.for.inc: // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1: omp.inner.for.end: // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1: omp.loop.exit: // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1: cond.true: // CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1: cond.false: // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: br label [[COND_END]] // CHECK1: cond.end: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1: omp.inner.for.cond: // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1: omp.inner.for.body: // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK1-NEXT: call void @_Z3fn5v() // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK1: omp.body.continue: // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1: omp.inner.for.inc: // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1: omp.inner.for.end: // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1: omp.loop.exit: // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103 // CHECK1-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32* // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i32* [[CONV]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[ARG_ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[ARG]], i32** [[ARG_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARG_ADDR]], align 8 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1: cond.true: // CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1: cond.false: // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: br label [[COND_END]] // CHECK1: cond.end: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1: omp.inner.for.cond: // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1: omp.inner.for.body: // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK1-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK1-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK1: omp_if.then: // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]]) // CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1: omp_if.else: // CHECK1-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK1-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: call void @.omp_outlined..9(i32* [[TMP13]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]]) #[[ATTR2]] // CHECK1-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK1-NEXT: br label [[OMP_IF_END]] // CHECK1: omp_if.end: // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1: omp.inner.for.inc: // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1: omp.inner.for.end: // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1: omp.loop.exit: // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1: cond.true: // CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1: cond.false: // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: br label [[COND_END]] // CHECK1: cond.end: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1: omp.inner.for.cond: // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1: omp.inner.for.body: // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK1-NEXT: call void @_Z3fn6v() // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK1: omp.body.continue: // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1: omp.inner.for.inc: // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1: omp.inner.for.end: // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1: omp.loop.exit: // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_ // CHECK1-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK1-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK1-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK1-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1: omp_offload.failed: // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63() #[[ATTR2]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1: omp_offload.cont: // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK1-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK1-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK1: omp_offload.failed2: // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69() #[[ATTR2]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK1: omp_offload.cont3: // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[ARG_ADDR]], align 4 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32* // CHECK1-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: store i8* null, i8** [[TMP10]], align 8 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK1-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK1-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK1-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK1: omp_offload.failed5: // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75(i64 [[TMP5]]) #[[ATTR2]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK1: omp_offload.cont6: // CHECK1-NEXT: ret i32 0 // // // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63 // CHECK1-SAME: () #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*)) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1: cond.true: // CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1: cond.false: // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: br label [[COND_END]] // CHECK1: cond.end: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1: omp.inner.for.cond: // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1: omp.inner.for.body: // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1: omp.inner.for.inc: // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1: omp.inner.for.end: // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1: omp.loop.exit: // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1: cond.true: // CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1: cond.false: // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: br label [[COND_END]] // CHECK1: cond.end: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1: omp.inner.for.cond: // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1: omp.inner.for.body: // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK1-NEXT: call void @_Z3fn1v() // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK1: omp.body.continue: // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1: omp.inner.for.inc: // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1: omp.inner.for.end: // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1: omp.loop.exit: // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69 // CHECK1-SAME: () #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..12 to void (i32*, i32*, ...)*)) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1: cond.true: // CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1: cond.false: // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: br label [[COND_END]] // CHECK1: cond.end: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1: omp.inner.for.cond: // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1: omp.inner.for.body: // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK1-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK1-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: call void @.omp_outlined..13(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK1-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1: omp.inner.for.inc: // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1: omp.inner.for.end: // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1: omp.loop.exit: // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1: cond.true: // CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1: cond.false: // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: br label [[COND_END]] // CHECK1: cond.end: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1: omp.inner.for.cond: // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1: omp.inner.for.body: // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK1-NEXT: call void @_Z3fn2v() // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK1: omp.body.continue: // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1: omp.inner.for.inc: // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1: omp.inner.for.end: // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1: omp.loop.exit: // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75 // CHECK1-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32* // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32* [[CONV]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[ARG_ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[ARG]], i32** [[ARG_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARG_ADDR]], align 8 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1: cond.true: // CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1: cond.false: // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: br label [[COND_END]] // CHECK1: cond.end: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1: omp.inner.for.cond: // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1: omp.inner.for.body: // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK1-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK1-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK1: omp_if.then: // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]]) // CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1: omp_if.else: // CHECK1-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK1-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: call void @.omp_outlined..15(i32* [[TMP13]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]]) #[[ATTR2]] // CHECK1-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK1-NEXT: br label [[OMP_IF_END]] // CHECK1: omp_if.end: // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1: omp.inner.for.inc: // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1: omp.inner.for.end: // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1: omp.loop.exit: // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1: cond.true: // CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1: cond.false: // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: br label [[COND_END]] // CHECK1: cond.end: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1: omp.inner.for.cond: // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1: omp.inner.for.body: // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK1-NEXT: call void @_Z3fn3v() // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK1: omp.body.continue: // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1: omp.inner.for.inc: // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1: omp.inner.for.end: // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1: omp.loop.exit: // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK1-NEXT: ret void // // // CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK1-SAME: () #[[ATTR5:[0-9]+]] { // CHECK1-NEXT: entry: // CHECK1-NEXT: call void @__tgt_register_requires(i64 1) // CHECK1-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@_Z9gtid_testv // CHECK2-SAME: () #[[ATTR0:[0-9]+]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK2-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK2-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK2: omp_offload.failed: // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47() #[[ATTR2:[0-9]+]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2: omp_offload.cont: // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK2-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK2-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK2: omp_offload.failed2: // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52() #[[ATTR2]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK2: omp_offload.cont3: // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47 // CHECK2-SAME: () #[[ATTR1:[0-9]+]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined. // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK2: cond.true: // CHECK2-NEXT: br label [[COND_END:%.*]] // CHECK2: cond.false: // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: br label [[COND_END]] // CHECK2: cond.end: // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2: omp.inner.for.cond: // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2: omp.inner.for.body: // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2: omp.inner.for.inc: // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2: omp.inner.for.end: // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2: omp.loop.exit: // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..1 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK2: cond.true: // CHECK2-NEXT: br label [[COND_END:%.*]] // CHECK2: cond.false: // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: br label [[COND_END]] // CHECK2: cond.end: // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2: omp.inner.for.cond: // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2: omp.inner.for.body: // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK2: omp.body.continue: // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2: omp.inner.for.inc: // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2: omp.inner.for.end: // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2: omp.loop.exit: // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52 // CHECK2-SAME: () #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*)) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..2 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK2: cond.true: // CHECK2-NEXT: br label [[COND_END:%.*]] // CHECK2: cond.false: // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: br label [[COND_END]] // CHECK2: cond.end: // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2: omp.inner.for.cond: // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2: omp.inner.for.body: // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK2-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK2-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: call void @.omp_outlined..3(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK2-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2: omp.inner.for.inc: // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2: omp.inner.for.end: // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2: omp.loop.exit: // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK2: cond.true: // CHECK2-NEXT: br label [[COND_END:%.*]] // CHECK2: cond.false: // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: br label [[COND_END]] // CHECK2: cond.end: // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2: omp.inner.for.cond: // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2: omp.inner.for.body: // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK2-NEXT: call void @_Z9gtid_testv() // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK2: omp.body.continue: // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2: omp.inner.for.inc: // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2: omp.inner.for.end: // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2: omp.loop.exit: // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@main // CHECK2-SAME: () #[[ATTR3:[0-9]+]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK2-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK2-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK2-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK2: omp_offload.failed: // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85() #[[ATTR2]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2: omp_offload.cont: // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK2-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK2-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK2: omp_offload.failed2: // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94() #[[ATTR2]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK2: omp_offload.cont3: // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* @Arg, align 4 // CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32* // CHECK2-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4 // CHECK2-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8 // CHECK2-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8 // CHECK2-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK2-NEXT: store i8* null, i8** [[TMP10]], align 8 // CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK2-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK2-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK2-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK2: omp_offload.failed5: // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103(i64 [[TMP5]]) #[[ATTR2]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK2: omp_offload.cont6: // CHECK2-NEXT: [[TMP15:%.*]] = load i32, i32* @Arg, align 4 // CHECK2-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP15]]) // CHECK2-NEXT: ret i32 [[CALL]] // // // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85 // CHECK2-SAME: () #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*)) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..4 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK2: cond.true: // CHECK2-NEXT: br label [[COND_END:%.*]] // CHECK2: cond.false: // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: br label [[COND_END]] // CHECK2: cond.end: // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2: omp.inner.for.cond: // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2: omp.inner.for.body: // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2: omp.inner.for.inc: // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2: omp.inner.for.end: // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2: omp.loop.exit: // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK2: cond.true: // CHECK2-NEXT: br label [[COND_END:%.*]] // CHECK2: cond.false: // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: br label [[COND_END]] // CHECK2: cond.end: // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2: omp.inner.for.cond: // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2: omp.inner.for.body: // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK2-NEXT: call void @_Z3fn4v() // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK2: omp.body.continue: // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2: omp.inner.for.inc: // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2: omp.inner.for.end: // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2: omp.loop.exit: // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94 // CHECK2-SAME: () #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*)) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK2: cond.true: // CHECK2-NEXT: br label [[COND_END:%.*]] // CHECK2: cond.false: // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: br label [[COND_END]] // CHECK2: cond.end: // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2: omp.inner.for.cond: // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2: omp.inner.for.body: // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK2-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK2-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: call void @.omp_outlined..7(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK2-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2: omp.inner.for.inc: // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2: omp.inner.for.end: // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2: omp.loop.exit: // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK2: cond.true: // CHECK2-NEXT: br label [[COND_END:%.*]] // CHECK2: cond.false: // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: br label [[COND_END]] // CHECK2: cond.end: // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2: omp.inner.for.cond: // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2: omp.inner.for.body: // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK2-NEXT: call void @_Z3fn5v() // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK2: omp.body.continue: // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2: omp.inner.for.inc: // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2: omp.inner.for.end: // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2: omp.loop.exit: // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103 // CHECK2-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8 // CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32* // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i32* [[CONV]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[ARG_ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[ARG]], i32** [[ARG_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARG_ADDR]], align 8 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK2: cond.true: // CHECK2-NEXT: br label [[COND_END:%.*]] // CHECK2: cond.false: // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: br label [[COND_END]] // CHECK2: cond.end: // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2: omp.inner.for.cond: // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2: omp.inner.for.body: // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK2-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK2-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK2: omp_if.then: // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]]) // CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2: omp_if.else: // CHECK2-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK2-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: call void @.omp_outlined..9(i32* [[TMP13]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]]) #[[ATTR2]] // CHECK2-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK2-NEXT: br label [[OMP_IF_END]] // CHECK2: omp_if.end: // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2: omp.inner.for.inc: // CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2: omp.inner.for.end: // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2: omp.loop.exit: // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK2: cond.true: // CHECK2-NEXT: br label [[COND_END:%.*]] // CHECK2: cond.false: // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: br label [[COND_END]] // CHECK2: cond.end: // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2: omp.inner.for.cond: // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2: omp.inner.for.body: // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK2-NEXT: call void @_Z3fn6v() // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK2: omp.body.continue: // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2: omp.inner.for.inc: // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2: omp.inner.for.end: // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2: omp.loop.exit: // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_ // CHECK2-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK2-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4 // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK2-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK2-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK2: omp_offload.failed: // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63() #[[ATTR2]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2: omp_offload.cont: // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK2-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK2-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK2: omp_offload.failed2: // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69() #[[ATTR2]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK2: omp_offload.cont3: // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[ARG_ADDR]], align 4 // CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32* // CHECK2-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4 // CHECK2-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8 // CHECK2-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8 // CHECK2-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK2-NEXT: store i8* null, i8** [[TMP10]], align 8 // CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK2-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK2-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK2-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK2: omp_offload.failed5: // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75(i64 [[TMP5]]) #[[ATTR2]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK2: omp_offload.cont6: // CHECK2-NEXT: ret i32 0 // // // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63 // CHECK2-SAME: () #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*)) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK2: cond.true: // CHECK2-NEXT: br label [[COND_END:%.*]] // CHECK2: cond.false: // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: br label [[COND_END]] // CHECK2: cond.end: // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2: omp.inner.for.cond: // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2: omp.inner.for.body: // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2: omp.inner.for.inc: // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2: omp.inner.for.end: // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2: omp.loop.exit: // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK2: cond.true: // CHECK2-NEXT: br label [[COND_END:%.*]] // CHECK2: cond.false: // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: br label [[COND_END]] // CHECK2: cond.end: // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2: omp.inner.for.cond: // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2: omp.inner.for.body: // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK2-NEXT: call void @_Z3fn1v() // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK2: omp.body.continue: // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2: omp.inner.for.inc: // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2: omp.inner.for.end: // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2: omp.loop.exit: // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69 // CHECK2-SAME: () #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..12 to void (i32*, i32*, ...)*)) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK2: cond.true: // CHECK2-NEXT: br label [[COND_END:%.*]] // CHECK2: cond.false: // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: br label [[COND_END]] // CHECK2: cond.end: // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2: omp.inner.for.cond: // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2: omp.inner.for.body: // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK2-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK2-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: call void @.omp_outlined..13(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK2-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2: omp.inner.for.inc: // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2: omp.inner.for.end: // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2: omp.loop.exit: // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK2: cond.true: // CHECK2-NEXT: br label [[COND_END:%.*]] // CHECK2: cond.false: // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: br label [[COND_END]] // CHECK2: cond.end: // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2: omp.inner.for.cond: // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2: omp.inner.for.body: // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK2-NEXT: call void @_Z3fn2v() // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK2: omp.body.continue: // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2: omp.inner.for.inc: // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2: omp.inner.for.end: // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2: omp.loop.exit: // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75 // CHECK2-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8 // CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32* // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32* [[CONV]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[ARG_ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[ARG]], i32** [[ARG_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARG_ADDR]], align 8 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK2: cond.true: // CHECK2-NEXT: br label [[COND_END:%.*]] // CHECK2: cond.false: // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: br label [[COND_END]] // CHECK2: cond.end: // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2: omp.inner.for.cond: // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2: omp.inner.for.body: // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK2-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK2-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK2: omp_if.then: // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]]) // CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2: omp_if.else: // CHECK2-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK2-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: call void @.omp_outlined..15(i32* [[TMP13]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]]) #[[ATTR2]] // CHECK2-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK2-NEXT: br label [[OMP_IF_END]] // CHECK2: omp_if.end: // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2: omp.inner.for.inc: // CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2: omp.inner.for.end: // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2: omp.loop.exit: // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK2: cond.true: // CHECK2-NEXT: br label [[COND_END:%.*]] // CHECK2: cond.false: // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: br label [[COND_END]] // CHECK2: cond.end: // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2: omp.inner.for.cond: // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2: omp.inner.for.body: // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK2-NEXT: call void @_Z3fn3v() // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK2: omp.body.continue: // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2: omp.inner.for.inc: // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK2-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2: omp.inner.for.end: // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2: omp.loop.exit: // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK2-NEXT: ret void // // // CHECK2-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK2-SAME: () #[[ATTR5:[0-9]+]] { // CHECK2-NEXT: entry: // CHECK2-NEXT: call void @__tgt_register_requires(i64 1) // CHECK2-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@_Z9gtid_testv // CHECK5-SAME: () #[[ATTR0:[0-9]+]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100) // CHECK5-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK5-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK5-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK5: omp_offload.failed: // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47() #[[ATTR2:[0-9]+]] // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK5: omp_offload.cont: // CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK5-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK5-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK5-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK5: omp_offload.failed2: // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52() #[[ATTR2]] // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK5: omp_offload.cont3: // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47 // CHECK5-SAME: () #[[ATTR1:[0-9]+]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined. // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5: cond.true: // CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5: cond.false: // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: br label [[COND_END]] // CHECK5: cond.end: // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5: omp.inner.for.cond: // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5: omp.inner.for.body: // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5: omp.inner.for.inc: // CHECK5-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK5-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5: omp.inner.for.end: // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5: omp.loop.exit: // CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..1 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK5-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK5-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5: cond.true: // CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5: cond.false: // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: br label [[COND_END]] // CHECK5: cond.end: // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5: omp.inner.for.cond: // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5: omp.inner.for.body: // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK5: omp.body.continue: // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5: omp.inner.for.inc: // CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK5-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5: omp.inner.for.end: // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5: omp.loop.exit: // CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52 // CHECK5-SAME: () #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*)) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..2 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5: cond.true: // CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5: cond.false: // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: br label [[COND_END]] // CHECK5: cond.end: // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5: omp.inner.for.cond: // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5: omp.inner.for.body: // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK5-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK5-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: call void @.omp_outlined..3(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK5-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5: omp.inner.for.inc: // CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK5-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5: omp.inner.for.end: // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5: omp.loop.exit: // CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK5-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK5-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5: cond.true: // CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5: cond.false: // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: br label [[COND_END]] // CHECK5: cond.end: // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5: omp.inner.for.cond: // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5: omp.inner.for.body: // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK5-NEXT: call void @_Z9gtid_testv() // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK5: omp.body.continue: // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5: omp.inner.for.inc: // CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK5-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5: omp.inner.for.end: // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5: omp.loop.exit: // CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@main // CHECK5-SAME: () #[[ATTR3:[0-9]+]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK5-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK5-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK5-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK5-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK5: omp_offload.failed: // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85() #[[ATTR2]] // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK5: omp_offload.cont: // CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK5-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK5-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK5-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK5: omp_offload.failed2: // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94() #[[ATTR2]] // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK5: omp_offload.cont3: // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* @Arg, align 4 // CHECK5-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32* // CHECK5-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4 // CHECK5-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8 // CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK5-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8 // CHECK5-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK5-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8 // CHECK5-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK5-NEXT: store i8* null, i8** [[TMP10]], align 8 // CHECK5-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK5-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK5-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK5-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK5: omp_offload.failed5: // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103(i64 [[TMP5]]) #[[ATTR2]] // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK5: omp_offload.cont6: // CHECK5-NEXT: [[TMP15:%.*]] = load i32, i32* @Arg, align 4 // CHECK5-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP15]]) // CHECK5-NEXT: ret i32 [[CALL]] // // // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85 // CHECK5-SAME: () #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*)) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..4 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5: cond.true: // CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5: cond.false: // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: br label [[COND_END]] // CHECK5: cond.end: // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5: omp.inner.for.cond: // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5: omp.inner.for.body: // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5: omp.inner.for.inc: // CHECK5-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK5-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5: omp.inner.for.end: // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5: omp.loop.exit: // CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK5-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK5-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5: cond.true: // CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5: cond.false: // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: br label [[COND_END]] // CHECK5: cond.end: // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5: omp.inner.for.cond: // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5: omp.inner.for.body: // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK5-NEXT: call void @_Z3fn4v() // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK5: omp.body.continue: // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5: omp.inner.for.inc: // CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK5-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5: omp.inner.for.end: // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5: omp.loop.exit: // CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94 // CHECK5-SAME: () #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*)) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5: cond.true: // CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5: cond.false: // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: br label [[COND_END]] // CHECK5: cond.end: // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5: omp.inner.for.cond: // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5: omp.inner.for.body: // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK5-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK5-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: call void @.omp_outlined..7(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK5-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5: omp.inner.for.inc: // CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK5-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5: omp.inner.for.end: // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5: omp.loop.exit: // CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK5-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK5-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5: cond.true: // CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5: cond.false: // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: br label [[COND_END]] // CHECK5: cond.end: // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5: omp.inner.for.cond: // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5: omp.inner.for.body: // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK5-NEXT: call void @_Z3fn5v() // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK5: omp.body.continue: // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5: omp.inner.for.inc: // CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK5-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5: omp.inner.for.end: // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5: omp.loop.exit: // CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103 // CHECK5-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8 // CHECK5-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32* // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i32* [[CONV]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[ARG_ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[ARG]], i32** [[ARG_ADDR]], align 8 // CHECK5-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARG_ADDR]], align 8 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5: cond.true: // CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5: cond.false: // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: br label [[COND_END]] // CHECK5: cond.end: // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5: omp.inner.for.cond: // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5: omp.inner.for.body: // CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK5-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK5-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK5: omp_if.then: // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]]) // CHECK5-NEXT: br label [[OMP_IF_END:%.*]] // CHECK5: omp_if.else: // CHECK5-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK5-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: call void @.omp_outlined..9(i32* [[TMP13]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]]) #[[ATTR2]] // CHECK5-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK5-NEXT: br label [[OMP_IF_END]] // CHECK5: omp_if.end: // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5: omp.inner.for.inc: // CHECK5-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK5-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5: omp.inner.for.end: // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5: omp.loop.exit: // CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK5-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK5-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5: cond.true: // CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5: cond.false: // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: br label [[COND_END]] // CHECK5: cond.end: // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5: omp.inner.for.cond: // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5: omp.inner.for.body: // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK5-NEXT: call void @_Z3fn6v() // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK5: omp.body.continue: // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5: omp.inner.for.inc: // CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK5-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5: omp.inner.for.end: // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5: omp.loop.exit: // CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_ // CHECK5-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK5-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4 // CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK5-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK5-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK5-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK5: omp_offload.failed: // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63() #[[ATTR2]] // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK5: omp_offload.cont: // CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK5-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK5-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK5-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK5: omp_offload.failed2: // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69() #[[ATTR2]] // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK5: omp_offload.cont3: // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[ARG_ADDR]], align 4 // CHECK5-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32* // CHECK5-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4 // CHECK5-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8 // CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK5-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8 // CHECK5-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK5-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8 // CHECK5-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK5-NEXT: store i8* null, i8** [[TMP10]], align 8 // CHECK5-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK5-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK5-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK5-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK5: omp_offload.failed5: // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75(i64 [[TMP5]]) #[[ATTR2]] // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK5: omp_offload.cont6: // CHECK5-NEXT: ret i32 0 // // // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63 // CHECK5-SAME: () #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*)) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5: cond.true: // CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5: cond.false: // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: br label [[COND_END]] // CHECK5: cond.end: // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5: omp.inner.for.cond: // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5: omp.inner.for.body: // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5: omp.inner.for.inc: // CHECK5-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK5-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5: omp.inner.for.end: // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5: omp.loop.exit: // CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK5-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK5-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5: cond.true: // CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5: cond.false: // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: br label [[COND_END]] // CHECK5: cond.end: // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5: omp.inner.for.cond: // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5: omp.inner.for.body: // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK5-NEXT: call void @_Z3fn1v() // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK5: omp.body.continue: // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5: omp.inner.for.inc: // CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK5-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5: omp.inner.for.end: // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5: omp.loop.exit: // CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69 // CHECK5-SAME: () #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..12 to void (i32*, i32*, ...)*)) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5: cond.true: // CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5: cond.false: // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: br label [[COND_END]] // CHECK5: cond.end: // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5: omp.inner.for.cond: // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5: omp.inner.for.body: // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK5-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK5-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: call void @.omp_outlined..13(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK5-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5: omp.inner.for.inc: // CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK5-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5: omp.inner.for.end: // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5: omp.loop.exit: // CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK5-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK5-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5: cond.true: // CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5: cond.false: // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: br label [[COND_END]] // CHECK5: cond.end: // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5: omp.inner.for.cond: // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5: omp.inner.for.body: // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK5-NEXT: call void @_Z3fn2v() // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK5: omp.body.continue: // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5: omp.inner.for.inc: // CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK5-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5: omp.inner.for.end: // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5: omp.loop.exit: // CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75 // CHECK5-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8 // CHECK5-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32* // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32* [[CONV]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[ARG_ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[ARG]], i32** [[ARG_ADDR]], align 8 // CHECK5-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARG_ADDR]], align 8 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5: cond.true: // CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5: cond.false: // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: br label [[COND_END]] // CHECK5: cond.end: // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5: omp.inner.for.cond: // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5: omp.inner.for.body: // CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK5-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK5-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK5: omp_if.then: // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]]) // CHECK5-NEXT: br label [[OMP_IF_END:%.*]] // CHECK5: omp_if.else: // CHECK5-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK5-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: call void @.omp_outlined..15(i32* [[TMP13]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]]) #[[ATTR2]] // CHECK5-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK5-NEXT: br label [[OMP_IF_END]] // CHECK5: omp_if.end: // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5: omp.inner.for.inc: // CHECK5-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK5-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5: omp.inner.for.end: // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5: omp.loop.exit: // CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK5-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK5-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5: cond.true: // CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5: cond.false: // CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: br label [[COND_END]] // CHECK5: cond.end: // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5: omp.inner.for.cond: // CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5: omp.inner.for.body: // CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK5-NEXT: call void @_Z3fn3v() // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK5: omp.body.continue: // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5: omp.inner.for.inc: // CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK5-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5: omp.inner.for.end: // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5: omp.loop.exit: // CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK5-NEXT: ret void // // // CHECK5-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK5-SAME: () #[[ATTR5:[0-9]+]] { // CHECK5-NEXT: entry: // CHECK5-NEXT: call void @__tgt_register_requires(i64 1) // CHECK5-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@_Z9gtid_testv // CHECK6-SAME: () #[[ATTR0:[0-9]+]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100) // CHECK6-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK6-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK6-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK6: omp_offload.failed: // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47() #[[ATTR2:[0-9]+]] // CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK6: omp_offload.cont: // CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK6-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK6-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK6-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK6: omp_offload.failed2: // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52() #[[ATTR2]] // CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK6: omp_offload.cont3: // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47 // CHECK6-SAME: () #[[ATTR1:[0-9]+]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined. // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK6-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK6: cond.true: // CHECK6-NEXT: br label [[COND_END:%.*]] // CHECK6: cond.false: // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: br label [[COND_END]] // CHECK6: cond.end: // CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK6: omp.inner.for.cond: // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK6-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK6: omp.inner.for.body: // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK6: omp.inner.for.inc: // CHECK6-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK6-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK6: omp.inner.for.end: // CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK6: omp.loop.exit: // CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..1 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK6-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK6-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK6-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK6-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK6-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK6-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK6-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK6: cond.true: // CHECK6-NEXT: br label [[COND_END:%.*]] // CHECK6: cond.false: // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: br label [[COND_END]] // CHECK6: cond.end: // CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK6: omp.inner.for.cond: // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK6-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK6: omp.inner.for.body: // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK6: omp.body.continue: // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK6: omp.inner.for.inc: // CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK6-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK6: omp.inner.for.end: // CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK6: omp.loop.exit: // CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52 // CHECK6-SAME: () #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*)) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..2 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK6-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK6: cond.true: // CHECK6-NEXT: br label [[COND_END:%.*]] // CHECK6: cond.false: // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: br label [[COND_END]] // CHECK6: cond.end: // CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK6: omp.inner.for.cond: // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK6-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK6: omp.inner.for.body: // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK6-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK6-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: call void @.omp_outlined..3(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK6-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK6: omp.inner.for.inc: // CHECK6-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK6-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK6: omp.inner.for.end: // CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK6: omp.loop.exit: // CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK6-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK6-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK6-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK6-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK6-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK6-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK6-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK6: cond.true: // CHECK6-NEXT: br label [[COND_END:%.*]] // CHECK6: cond.false: // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: br label [[COND_END]] // CHECK6: cond.end: // CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK6: omp.inner.for.cond: // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK6-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK6: omp.inner.for.body: // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK6-NEXT: call void @_Z9gtid_testv() // CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK6: omp.body.continue: // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK6: omp.inner.for.inc: // CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK6-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK6: omp.inner.for.end: // CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK6: omp.loop.exit: // CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@main // CHECK6-SAME: () #[[ATTR3:[0-9]+]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK6-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK6-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK6-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK6-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK6: omp_offload.failed: // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85() #[[ATTR2]] // CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK6: omp_offload.cont: // CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK6-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK6-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK6-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK6: omp_offload.failed2: // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94() #[[ATTR2]] // CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK6: omp_offload.cont3: // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* @Arg, align 4 // CHECK6-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32* // CHECK6-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4 // CHECK6-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8 // CHECK6-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK6-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK6-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8 // CHECK6-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK6-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK6-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8 // CHECK6-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK6-NEXT: store i8* null, i8** [[TMP10]], align 8 // CHECK6-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK6-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK6-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK6-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK6-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK6: omp_offload.failed5: // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103(i64 [[TMP5]]) #[[ATTR2]] // CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK6: omp_offload.cont6: // CHECK6-NEXT: [[TMP15:%.*]] = load i32, i32* @Arg, align 4 // CHECK6-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP15]]) // CHECK6-NEXT: ret i32 [[CALL]] // // // CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85 // CHECK6-SAME: () #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*)) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..4 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK6-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK6: cond.true: // CHECK6-NEXT: br label [[COND_END:%.*]] // CHECK6: cond.false: // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: br label [[COND_END]] // CHECK6: cond.end: // CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK6: omp.inner.for.cond: // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK6-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK6: omp.inner.for.body: // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK6: omp.inner.for.inc: // CHECK6-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK6-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK6: omp.inner.for.end: // CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK6: omp.loop.exit: // CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK6-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK6-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK6-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK6-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK6-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK6-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK6-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK6: cond.true: // CHECK6-NEXT: br label [[COND_END:%.*]] // CHECK6: cond.false: // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: br label [[COND_END]] // CHECK6: cond.end: // CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK6: omp.inner.for.cond: // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK6-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK6: omp.inner.for.body: // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK6-NEXT: call void @_Z3fn4v() // CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK6: omp.body.continue: // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK6: omp.inner.for.inc: // CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK6-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK6: omp.inner.for.end: // CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK6: omp.loop.exit: // CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94 // CHECK6-SAME: () #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*)) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK6-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK6: cond.true: // CHECK6-NEXT: br label [[COND_END:%.*]] // CHECK6: cond.false: // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: br label [[COND_END]] // CHECK6: cond.end: // CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK6: omp.inner.for.cond: // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK6-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK6: omp.inner.for.body: // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK6-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK6-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: call void @.omp_outlined..7(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK6-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK6: omp.inner.for.inc: // CHECK6-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK6-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK6: omp.inner.for.end: // CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK6: omp.loop.exit: // CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK6-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK6-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK6-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK6-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK6-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK6-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK6-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK6: cond.true: // CHECK6-NEXT: br label [[COND_END:%.*]] // CHECK6: cond.false: // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: br label [[COND_END]] // CHECK6: cond.end: // CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK6: omp.inner.for.cond: // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK6-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK6: omp.inner.for.body: // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK6-NEXT: call void @_Z3fn5v() // CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK6: omp.body.continue: // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK6: omp.inner.for.inc: // CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK6-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK6: omp.inner.for.end: // CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK6: omp.loop.exit: // CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103 // CHECK6-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8 // CHECK6-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32* // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i32* [[CONV]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[ARG_ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[ARG]], i32** [[ARG_ADDR]], align 8 // CHECK6-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARG_ADDR]], align 8 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK6-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 // CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99 // CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK6: cond.true: // CHECK6-NEXT: br label [[COND_END:%.*]] // CHECK6: cond.false: // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: br label [[COND_END]] // CHECK6: cond.end: // CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK6: omp.inner.for.cond: // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK6-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK6: omp.inner.for.body: // CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK6-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK6-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK6-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK6: omp_if.then: // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]]) // CHECK6-NEXT: br label [[OMP_IF_END:%.*]] // CHECK6: omp_if.else: // CHECK6-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK6-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: call void @.omp_outlined..9(i32* [[TMP13]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]]) #[[ATTR2]] // CHECK6-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK6-NEXT: br label [[OMP_IF_END]] // CHECK6: omp_if.end: // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK6: omp.inner.for.inc: // CHECK6-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK6-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK6: omp.inner.for.end: // CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK6: omp.loop.exit: // CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK6-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK6-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK6-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK6-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK6-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK6-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK6-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK6: cond.true: // CHECK6-NEXT: br label [[COND_END:%.*]] // CHECK6: cond.false: // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: br label [[COND_END]] // CHECK6: cond.end: // CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK6: omp.inner.for.cond: // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK6-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK6: omp.inner.for.body: // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK6-NEXT: call void @_Z3fn6v() // CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK6: omp.body.continue: // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK6: omp.inner.for.inc: // CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK6-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK6: omp.inner.for.end: // CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK6: omp.loop.exit: // CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_ // CHECK6-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK6-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4 // CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK6-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK6-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK6-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK6: omp_offload.failed: // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63() #[[ATTR2]] // CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK6: omp_offload.cont: // CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK6-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK6-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK6-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK6: omp_offload.failed2: // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69() #[[ATTR2]] // CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK6: omp_offload.cont3: // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[ARG_ADDR]], align 4 // CHECK6-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32* // CHECK6-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4 // CHECK6-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8 // CHECK6-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK6-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK6-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8 // CHECK6-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK6-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK6-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8 // CHECK6-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK6-NEXT: store i8* null, i8** [[TMP10]], align 8 // CHECK6-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK6-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK6-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK6-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK6-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK6: omp_offload.failed5: // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75(i64 [[TMP5]]) #[[ATTR2]] // CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK6: omp_offload.cont6: // CHECK6-NEXT: ret i32 0 // // // CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63 // CHECK6-SAME: () #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*)) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK6-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK6: cond.true: // CHECK6-NEXT: br label [[COND_END:%.*]] // CHECK6: cond.false: // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: br label [[COND_END]] // CHECK6: cond.end: // CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK6: omp.inner.for.cond: // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK6-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK6: omp.inner.for.body: // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK6: omp.inner.for.inc: // CHECK6-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK6-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK6: omp.inner.for.end: // CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK6: omp.loop.exit: // CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK6-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK6-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK6-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK6-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK6-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK6-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK6-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK6: cond.true: // CHECK6-NEXT: br label [[COND_END:%.*]] // CHECK6: cond.false: // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: br label [[COND_END]] // CHECK6: cond.end: // CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK6: omp.inner.for.cond: // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK6-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK6: omp.inner.for.body: // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK6-NEXT: call void @_Z3fn1v() // CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK6: omp.body.continue: // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK6: omp.inner.for.inc: // CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK6-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK6: omp.inner.for.end: // CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK6: omp.loop.exit: // CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69 // CHECK6-SAME: () #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..12 to void (i32*, i32*, ...)*)) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK6-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK6: cond.true: // CHECK6-NEXT: br label [[COND_END:%.*]] // CHECK6: cond.false: // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: br label [[COND_END]] // CHECK6: cond.end: // CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK6: omp.inner.for.cond: // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK6-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK6: omp.inner.for.body: // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK6-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK6-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: call void @.omp_outlined..13(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK6-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK6: omp.inner.for.inc: // CHECK6-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK6-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK6: omp.inner.for.end: // CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK6: omp.loop.exit: // CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK6-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK6-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK6-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK6-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK6-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK6-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK6-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK6: cond.true: // CHECK6-NEXT: br label [[COND_END:%.*]] // CHECK6: cond.false: // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: br label [[COND_END]] // CHECK6: cond.end: // CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK6: omp.inner.for.cond: // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK6-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK6: omp.inner.for.body: // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK6-NEXT: call void @_Z3fn2v() // CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK6: omp.body.continue: // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK6: omp.inner.for.inc: // CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK6-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK6: omp.inner.for.end: // CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK6: omp.loop.exit: // CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75 // CHECK6-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8 // CHECK6-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32* // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32* [[CONV]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[ARG_ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[ARG]], i32** [[ARG_ADDR]], align 8 // CHECK6-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARG_ADDR]], align 8 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK6-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 // CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99 // CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK6: cond.true: // CHECK6-NEXT: br label [[COND_END:%.*]] // CHECK6: cond.false: // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: br label [[COND_END]] // CHECK6: cond.end: // CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK6: omp.inner.for.cond: // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK6-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK6: omp.inner.for.body: // CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK6-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK6-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK6-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK6-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK6-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK6: omp_if.then: // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]]) // CHECK6-NEXT: br label [[OMP_IF_END:%.*]] // CHECK6: omp_if.else: // CHECK6-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK6-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: call void @.omp_outlined..15(i32* [[TMP13]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]]) #[[ATTR2]] // CHECK6-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK6-NEXT: br label [[OMP_IF_END]] // CHECK6: omp_if.end: // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK6: omp.inner.for.inc: // CHECK6-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK6-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK6: omp.inner.for.end: // CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK6: omp.loop.exit: // CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK6-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK6-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK6-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK6-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK6-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK6-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK6-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK6: cond.true: // CHECK6-NEXT: br label [[COND_END:%.*]] // CHECK6: cond.false: // CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: br label [[COND_END]] // CHECK6: cond.end: // CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK6-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK6: omp.inner.for.cond: // CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK6-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK6-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK6: omp.inner.for.body: // CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK6-NEXT: call void @_Z3fn3v() // CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK6: omp.body.continue: // CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK6: omp.inner.for.inc: // CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK6-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK6: omp.inner.for.end: // CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK6: omp.loop.exit: // CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK6-NEXT: ret void // // // CHECK6-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK6-SAME: () #[[ATTR5:[0-9]+]] { // CHECK6-NEXT: entry: // CHECK6-NEXT: call void @__tgt_register_requires(i64 1) // CHECK6-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@_Z9gtid_testv // CHECK9-SAME: () #[[ATTR0:[0-9]+]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100) // CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK9-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9: omp_offload.failed: // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47() #[[ATTR2:[0-9]+]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9: omp_offload.cont: // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK9-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK9-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK9: omp_offload.failed2: // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52() #[[ATTR2]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK9: omp_offload.cont3: // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47 // CHECK9-SAME: () #[[ATTR1:[0-9]+]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined. // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9: cond.true: // CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9: cond.false: // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: br label [[COND_END]] // CHECK9: cond.end: // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9: omp.inner.for.cond: // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9: omp.inner.for.body: // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9: omp.inner.for.inc: // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9: omp.inner.for.end: // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9: omp.loop.exit: // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..1 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9: cond.true: // CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9: cond.false: // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: br label [[COND_END]] // CHECK9: cond.end: // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9: omp.inner.for.cond: // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9: omp.inner.for.body: // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK9: omp.body.continue: // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9: omp.inner.for.inc: // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9: omp.inner.for.end: // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9: omp.loop.exit: // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52 // CHECK9-SAME: () #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*)) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..2 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9: cond.true: // CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9: cond.false: // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: br label [[COND_END]] // CHECK9: cond.end: // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9: omp.inner.for.cond: // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9: omp.inner.for.body: // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK9-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK9-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: call void @.omp_outlined..3(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK9-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9: omp.inner.for.inc: // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9: omp.inner.for.end: // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9: omp.loop.exit: // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9: cond.true: // CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9: cond.false: // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: br label [[COND_END]] // CHECK9: cond.end: // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9: omp.inner.for.cond: // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9: omp.inner.for.body: // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK9-NEXT: call void @_Z9gtid_testv() // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK9: omp.body.continue: // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9: omp.inner.for.inc: // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9: omp.inner.for.end: // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9: omp.loop.exit: // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@main // CHECK9-SAME: () #[[ATTR3:[0-9]+]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK9-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK9-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9: omp_offload.failed: // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85() #[[ATTR2]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9: omp_offload.cont: // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK9-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK9-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK9: omp_offload.failed2: // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94() #[[ATTR2]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK9: omp_offload.cont3: // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* @Arg, align 4 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32* // CHECK9-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: store i8* null, i8** [[TMP10]], align 8 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK9-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK9-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK9: omp_offload.failed5: // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103(i64 [[TMP5]]) #[[ATTR2]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK9: omp_offload.cont6: // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* @Arg, align 4 // CHECK9-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP15]]) // CHECK9-NEXT: ret i32 [[CALL]] // // // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85 // CHECK9-SAME: () #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*)) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..4 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9: cond.true: // CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9: cond.false: // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: br label [[COND_END]] // CHECK9: cond.end: // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9: omp.inner.for.cond: // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9: omp.inner.for.body: // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9: omp.inner.for.inc: // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9: omp.inner.for.end: // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9: omp.loop.exit: // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9: cond.true: // CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9: cond.false: // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: br label [[COND_END]] // CHECK9: cond.end: // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9: omp.inner.for.cond: // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9: omp.inner.for.body: // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK9-NEXT: call void @_Z3fn4v() // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK9: omp.body.continue: // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9: omp.inner.for.inc: // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9: omp.inner.for.end: // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9: omp.loop.exit: // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94 // CHECK9-SAME: () #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*)) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9: cond.true: // CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9: cond.false: // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: br label [[COND_END]] // CHECK9: cond.end: // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9: omp.inner.for.cond: // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9: omp.inner.for.body: // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK9-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK9-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: call void @.omp_outlined..7(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK9-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9: omp.inner.for.inc: // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9: omp.inner.for.end: // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9: omp.loop.exit: // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9: cond.true: // CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9: cond.false: // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: br label [[COND_END]] // CHECK9: cond.end: // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9: omp.inner.for.cond: // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9: omp.inner.for.body: // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK9-NEXT: call void @_Z3fn5v() // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK9: omp.body.continue: // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9: omp.inner.for.inc: // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9: omp.inner.for.end: // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9: omp.loop.exit: // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103 // CHECK9-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32* // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i32* [[CONV]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[ARG_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[ARG]], i32** [[ARG_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARG_ADDR]], align 8 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9: cond.true: // CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9: cond.false: // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: br label [[COND_END]] // CHECK9: cond.end: // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9: omp.inner.for.cond: // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9: omp.inner.for.body: // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK9-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK9-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK9: omp_if.then: // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]]) // CHECK9-NEXT: br label [[OMP_IF_END:%.*]] // CHECK9: omp_if.else: // CHECK9-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK9-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: call void @.omp_outlined..9(i32* [[TMP13]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]]) #[[ATTR2]] // CHECK9-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK9-NEXT: br label [[OMP_IF_END]] // CHECK9: omp_if.end: // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9: omp.inner.for.inc: // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9: omp.inner.for.end: // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9: omp.loop.exit: // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9: cond.true: // CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9: cond.false: // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: br label [[COND_END]] // CHECK9: cond.end: // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9: omp.inner.for.cond: // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9: omp.inner.for.body: // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK9-NEXT: call void @_Z3fn6v() // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK9: omp.body.continue: // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9: omp.inner.for.inc: // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9: omp.inner.for.end: // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9: omp.loop.exit: // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_ // CHECK9-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK9-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK9-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9: omp_offload.failed: // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63() #[[ATTR2]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9: omp_offload.cont: // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK9-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK9-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK9: omp_offload.failed2: // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69() #[[ATTR2]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK9: omp_offload.cont3: // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[ARG_ADDR]], align 4 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32* // CHECK9-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: store i8* null, i8** [[TMP10]], align 8 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK9-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK9-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK9: omp_offload.failed5: // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75(i64 [[TMP5]]) #[[ATTR2]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK9: omp_offload.cont6: // CHECK9-NEXT: ret i32 0 // // // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63 // CHECK9-SAME: () #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*)) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9: cond.true: // CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9: cond.false: // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: br label [[COND_END]] // CHECK9: cond.end: // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9: omp.inner.for.cond: // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9: omp.inner.for.body: // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9: omp.inner.for.inc: // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9: omp.inner.for.end: // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9: omp.loop.exit: // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9: cond.true: // CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9: cond.false: // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: br label [[COND_END]] // CHECK9: cond.end: // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9: omp.inner.for.cond: // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9: omp.inner.for.body: // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK9-NEXT: call void @_Z3fn1v() // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK9: omp.body.continue: // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9: omp.inner.for.inc: // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9: omp.inner.for.end: // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9: omp.loop.exit: // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69 // CHECK9-SAME: () #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..12 to void (i32*, i32*, ...)*)) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9: cond.true: // CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9: cond.false: // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: br label [[COND_END]] // CHECK9: cond.end: // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9: omp.inner.for.cond: // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9: omp.inner.for.body: // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK9-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK9-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: call void @.omp_outlined..13(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK9-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9: omp.inner.for.inc: // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9: omp.inner.for.end: // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9: omp.loop.exit: // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9: cond.true: // CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9: cond.false: // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: br label [[COND_END]] // CHECK9: cond.end: // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9: omp.inner.for.cond: // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9: omp.inner.for.body: // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK9-NEXT: call void @_Z3fn2v() // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK9: omp.body.continue: // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9: omp.inner.for.inc: // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9: omp.inner.for.end: // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9: omp.loop.exit: // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75 // CHECK9-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32* // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32* [[CONV]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[ARG_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[ARG]], i32** [[ARG_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARG_ADDR]], align 8 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9: cond.true: // CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9: cond.false: // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: br label [[COND_END]] // CHECK9: cond.end: // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9: omp.inner.for.cond: // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9: omp.inner.for.body: // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK9-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK9-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK9: omp_if.then: // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]]) // CHECK9-NEXT: br label [[OMP_IF_END:%.*]] // CHECK9: omp_if.else: // CHECK9-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK9-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: call void @.omp_outlined..15(i32* [[TMP13]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]]) #[[ATTR2]] // CHECK9-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK9-NEXT: br label [[OMP_IF_END]] // CHECK9: omp_if.end: // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9: omp.inner.for.inc: // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9: omp.inner.for.end: // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9: omp.loop.exit: // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9: cond.true: // CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9: cond.false: // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: br label [[COND_END]] // CHECK9: cond.end: // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9: omp.inner.for.cond: // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9: omp.inner.for.body: // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK9-NEXT: call void @_Z3fn3v() // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK9: omp.body.continue: // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9: omp.inner.for.inc: // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9: omp.inner.for.end: // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9: omp.loop.exit: // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK9-NEXT: ret void // // // CHECK9-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK9-SAME: () #[[ATTR5:[0-9]+]] { // CHECK9-NEXT: entry: // CHECK9-NEXT: call void @__tgt_register_requires(i64 1) // CHECK9-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@_Z9gtid_testv // CHECK10-SAME: () #[[ATTR0:[0-9]+]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100) // CHECK10-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK10-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10: omp_offload.failed: // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47() #[[ATTR2:[0-9]+]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10: omp_offload.cont: // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK10-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK10-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK10: omp_offload.failed2: // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52() #[[ATTR2]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK10: omp_offload.cont3: // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47 // CHECK10-SAME: () #[[ATTR1:[0-9]+]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined. // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK10: cond.true: // CHECK10-NEXT: br label [[COND_END:%.*]] // CHECK10: cond.false: // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: br label [[COND_END]] // CHECK10: cond.end: // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10: omp.inner.for.cond: // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10: omp.inner.for.body: // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10: omp.inner.for.inc: // CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10: omp.inner.for.end: // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10: omp.loop.exit: // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..1 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK10: cond.true: // CHECK10-NEXT: br label [[COND_END:%.*]] // CHECK10: cond.false: // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: br label [[COND_END]] // CHECK10: cond.end: // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10: omp.inner.for.cond: // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10: omp.inner.for.body: // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK10: omp.body.continue: // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10: omp.inner.for.inc: // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10: omp.inner.for.end: // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10: omp.loop.exit: // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52 // CHECK10-SAME: () #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*)) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..2 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK10: cond.true: // CHECK10-NEXT: br label [[COND_END:%.*]] // CHECK10: cond.false: // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: br label [[COND_END]] // CHECK10: cond.end: // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10: omp.inner.for.cond: // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10: omp.inner.for.body: // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK10-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK10-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: call void @.omp_outlined..3(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK10-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10: omp.inner.for.inc: // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10: omp.inner.for.end: // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10: omp.loop.exit: // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK10: cond.true: // CHECK10-NEXT: br label [[COND_END:%.*]] // CHECK10: cond.false: // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: br label [[COND_END]] // CHECK10: cond.end: // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10: omp.inner.for.cond: // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10: omp.inner.for.body: // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK10-NEXT: call void @_Z9gtid_testv() // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK10: omp.body.continue: // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10: omp.inner.for.inc: // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10: omp.inner.for.end: // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10: omp.loop.exit: // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@main // CHECK10-SAME: () #[[ATTR3:[0-9]+]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK10-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK10-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK10-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10: omp_offload.failed: // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85() #[[ATTR2]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10: omp_offload.cont: // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK10-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK10-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK10: omp_offload.failed2: // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94() #[[ATTR2]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK10: omp_offload.cont3: // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* @Arg, align 4 // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32* // CHECK10-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8 // CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8 // CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: store i8* null, i8** [[TMP10]], align 8 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK10-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK10-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK10: omp_offload.failed5: // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103(i64 [[TMP5]]) #[[ATTR2]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK10: omp_offload.cont6: // CHECK10-NEXT: [[TMP15:%.*]] = load i32, i32* @Arg, align 4 // CHECK10-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP15]]) // CHECK10-NEXT: ret i32 [[CALL]] // // // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85 // CHECK10-SAME: () #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*)) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..4 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK10: cond.true: // CHECK10-NEXT: br label [[COND_END:%.*]] // CHECK10: cond.false: // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: br label [[COND_END]] // CHECK10: cond.end: // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10: omp.inner.for.cond: // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10: omp.inner.for.body: // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10: omp.inner.for.inc: // CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10: omp.inner.for.end: // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10: omp.loop.exit: // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK10: cond.true: // CHECK10-NEXT: br label [[COND_END:%.*]] // CHECK10: cond.false: // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: br label [[COND_END]] // CHECK10: cond.end: // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10: omp.inner.for.cond: // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10: omp.inner.for.body: // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK10-NEXT: call void @_Z3fn4v() // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK10: omp.body.continue: // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10: omp.inner.for.inc: // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10: omp.inner.for.end: // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10: omp.loop.exit: // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94 // CHECK10-SAME: () #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*)) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK10: cond.true: // CHECK10-NEXT: br label [[COND_END:%.*]] // CHECK10: cond.false: // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: br label [[COND_END]] // CHECK10: cond.end: // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10: omp.inner.for.cond: // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10: omp.inner.for.body: // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK10-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK10-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: call void @.omp_outlined..7(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK10-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10: omp.inner.for.inc: // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10: omp.inner.for.end: // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10: omp.loop.exit: // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK10: cond.true: // CHECK10-NEXT: br label [[COND_END:%.*]] // CHECK10: cond.false: // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: br label [[COND_END]] // CHECK10: cond.end: // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10: omp.inner.for.cond: // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10: omp.inner.for.body: // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK10-NEXT: call void @_Z3fn5v() // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK10: omp.body.continue: // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10: omp.inner.for.inc: // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10: omp.inner.for.end: // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10: omp.loop.exit: // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103 // CHECK10-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8 // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32* // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i32* [[CONV]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[ARG_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[ARG]], i32** [[ARG_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARG_ADDR]], align 8 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK10: cond.true: // CHECK10-NEXT: br label [[COND_END:%.*]] // CHECK10: cond.false: // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: br label [[COND_END]] // CHECK10: cond.end: // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10: omp.inner.for.cond: // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10: omp.inner.for.body: // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK10-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK10-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK10: omp_if.then: // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]]) // CHECK10-NEXT: br label [[OMP_IF_END:%.*]] // CHECK10: omp_if.else: // CHECK10-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK10-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: call void @.omp_outlined..9(i32* [[TMP13]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]]) #[[ATTR2]] // CHECK10-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK10-NEXT: br label [[OMP_IF_END]] // CHECK10: omp_if.end: // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10: omp.inner.for.inc: // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10: omp.inner.for.end: // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10: omp.loop.exit: // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK10: cond.true: // CHECK10-NEXT: br label [[COND_END:%.*]] // CHECK10: cond.false: // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: br label [[COND_END]] // CHECK10: cond.end: // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10: omp.inner.for.cond: // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10: omp.inner.for.body: // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK10-NEXT: call void @_Z3fn6v() // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK10: omp.body.continue: // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10: omp.inner.for.inc: // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10: omp.inner.for.end: // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10: omp.loop.exit: // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_ // CHECK10-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK10-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK10-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK10-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10: omp_offload.failed: // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63() #[[ATTR2]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10: omp_offload.cont: // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK10-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK10-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK10: omp_offload.failed2: // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69() #[[ATTR2]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK10: omp_offload.cont3: // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[ARG_ADDR]], align 4 // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32* // CHECK10-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8 // CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8 // CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: store i8* null, i8** [[TMP10]], align 8 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK10-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK10-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK10: omp_offload.failed5: // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75(i64 [[TMP5]]) #[[ATTR2]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK10: omp_offload.cont6: // CHECK10-NEXT: ret i32 0 // // // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63 // CHECK10-SAME: () #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*)) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK10: cond.true: // CHECK10-NEXT: br label [[COND_END:%.*]] // CHECK10: cond.false: // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: br label [[COND_END]] // CHECK10: cond.end: // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10: omp.inner.for.cond: // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10: omp.inner.for.body: // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10: omp.inner.for.inc: // CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10: omp.inner.for.end: // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10: omp.loop.exit: // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK10: cond.true: // CHECK10-NEXT: br label [[COND_END:%.*]] // CHECK10: cond.false: // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: br label [[COND_END]] // CHECK10: cond.end: // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10: omp.inner.for.cond: // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10: omp.inner.for.body: // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK10-NEXT: call void @_Z3fn1v() // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK10: omp.body.continue: // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10: omp.inner.for.inc: // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10: omp.inner.for.end: // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10: omp.loop.exit: // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69 // CHECK10-SAME: () #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..12 to void (i32*, i32*, ...)*)) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK10: cond.true: // CHECK10-NEXT: br label [[COND_END:%.*]] // CHECK10: cond.false: // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: br label [[COND_END]] // CHECK10: cond.end: // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10: omp.inner.for.cond: // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10: omp.inner.for.body: // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK10-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK10-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: call void @.omp_outlined..13(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK10-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10: omp.inner.for.inc: // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10: omp.inner.for.end: // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10: omp.loop.exit: // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK10: cond.true: // CHECK10-NEXT: br label [[COND_END:%.*]] // CHECK10: cond.false: // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: br label [[COND_END]] // CHECK10: cond.end: // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10: omp.inner.for.cond: // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10: omp.inner.for.body: // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK10-NEXT: call void @_Z3fn2v() // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK10: omp.body.continue: // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10: omp.inner.for.inc: // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10: omp.inner.for.end: // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10: omp.loop.exit: // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75 // CHECK10-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8 // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32* // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32* [[CONV]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[ARG_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[ARG]], i32** [[ARG_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARG_ADDR]], align 8 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK10: cond.true: // CHECK10-NEXT: br label [[COND_END:%.*]] // CHECK10: cond.false: // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: br label [[COND_END]] // CHECK10: cond.end: // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10: omp.inner.for.cond: // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10: omp.inner.for.body: // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK10-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK10-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK10: omp_if.then: // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]]) // CHECK10-NEXT: br label [[OMP_IF_END:%.*]] // CHECK10: omp_if.else: // CHECK10-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK10-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: call void @.omp_outlined..15(i32* [[TMP13]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]]) #[[ATTR2]] // CHECK10-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK10-NEXT: br label [[OMP_IF_END]] // CHECK10: omp_if.end: // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10: omp.inner.for.inc: // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10: omp.inner.for.end: // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10: omp.loop.exit: // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK10: cond.true: // CHECK10-NEXT: br label [[COND_END:%.*]] // CHECK10: cond.false: // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: br label [[COND_END]] // CHECK10: cond.end: // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10: omp.inner.for.cond: // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10: omp.inner.for.body: // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK10-NEXT: call void @_Z3fn3v() // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK10: omp.body.continue: // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10: omp.inner.for.inc: // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10: omp.inner.for.end: // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10: omp.loop.exit: // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK10-NEXT: ret void // // // CHECK10-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK10-SAME: () #[[ATTR5:[0-9]+]] { // CHECK10-NEXT: entry: // CHECK10-NEXT: call void @__tgt_register_requires(i64 1) // CHECK10-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@_Z9gtid_testv // CHECK13-SAME: () #[[ATTR0:[0-9]+]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK13-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100) // CHECK13-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK13-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK13-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK13: omp_offload.failed: // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47() #[[ATTR2:[0-9]+]] // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK13: omp_offload.cont: // CHECK13-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK13-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK13-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK13-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK13: omp_offload.failed2: // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52() #[[ATTR2]] // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK13: omp_offload.cont3: // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47 // CHECK13-SAME: () #[[ATTR1:[0-9]+]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_outlined. // CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK13-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK13: cond.true: // CHECK13-NEXT: br label [[COND_END:%.*]] // CHECK13: cond.false: // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: br label [[COND_END]] // CHECK13: cond.end: // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13: omp.inner.for.cond: // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13: omp.inner.for.body: // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13: omp.inner.for.inc: // CHECK13-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK13-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK13: omp.inner.for.end: // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK13: omp.loop.exit: // CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..1 // CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK13-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK13-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK13: cond.true: // CHECK13-NEXT: br label [[COND_END:%.*]] // CHECK13: cond.false: // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: br label [[COND_END]] // CHECK13: cond.end: // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13: omp.inner.for.cond: // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK13-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13: omp.inner.for.body: // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK13: omp.body.continue: // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13: omp.inner.for.inc: // CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK13-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK13: omp.inner.for.end: // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK13: omp.loop.exit: // CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52 // CHECK13-SAME: () #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*)) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..2 // CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK13-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK13: cond.true: // CHECK13-NEXT: br label [[COND_END:%.*]] // CHECK13: cond.false: // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: br label [[COND_END]] // CHECK13: cond.end: // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13: omp.inner.for.cond: // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13: omp.inner.for.body: // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK13-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK13-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: call void @.omp_outlined..3(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK13-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13: omp.inner.for.inc: // CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK13-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK13: omp.inner.for.end: // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK13: omp.loop.exit: // CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK13-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK13-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK13: cond.true: // CHECK13-NEXT: br label [[COND_END:%.*]] // CHECK13: cond.false: // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: br label [[COND_END]] // CHECK13: cond.end: // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13: omp.inner.for.cond: // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK13-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13: omp.inner.for.body: // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK13-NEXT: call void @_Z9gtid_testv() // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK13: omp.body.continue: // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13: omp.inner.for.inc: // CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK13-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK13: omp.inner.for.end: // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK13: omp.loop.exit: // CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@main // CHECK13-SAME: () #[[ATTR3:[0-9]+]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK13-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK13-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK13-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK13-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK13-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK13-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK13-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK13: omp_offload.failed: // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85() #[[ATTR2]] // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK13: omp_offload.cont: // CHECK13-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK13-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK13-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK13-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK13: omp_offload.failed2: // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94() #[[ATTR2]] // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK13: omp_offload.cont3: // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* @Arg, align 4 // CHECK13-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32* // CHECK13-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4 // CHECK13-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8 // CHECK13-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK13-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK13-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8 // CHECK13-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK13-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK13-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8 // CHECK13-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK13-NEXT: store i8* null, i8** [[TMP10]], align 8 // CHECK13-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK13-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK13-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK13-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK13-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK13-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK13: omp_offload.failed5: // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103(i64 [[TMP5]]) #[[ATTR2]] // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK13: omp_offload.cont6: // CHECK13-NEXT: [[TMP15:%.*]] = load i32, i32* @Arg, align 4 // CHECK13-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP15]]) // CHECK13-NEXT: ret i32 [[CALL]] // // // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85 // CHECK13-SAME: () #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*)) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..4 // CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK13-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK13: cond.true: // CHECK13-NEXT: br label [[COND_END:%.*]] // CHECK13: cond.false: // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: br label [[COND_END]] // CHECK13: cond.end: // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13: omp.inner.for.cond: // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13: omp.inner.for.body: // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13: omp.inner.for.inc: // CHECK13-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK13-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK13: omp.inner.for.end: // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK13: omp.loop.exit: // CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK13-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK13-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK13: cond.true: // CHECK13-NEXT: br label [[COND_END:%.*]] // CHECK13: cond.false: // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: br label [[COND_END]] // CHECK13: cond.end: // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13: omp.inner.for.cond: // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK13-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13: omp.inner.for.body: // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK13-NEXT: call void @_Z3fn4v() // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK13: omp.body.continue: // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13: omp.inner.for.inc: // CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK13-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK13: omp.inner.for.end: // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK13: omp.loop.exit: // CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94 // CHECK13-SAME: () #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*)) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK13-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK13: cond.true: // CHECK13-NEXT: br label [[COND_END:%.*]] // CHECK13: cond.false: // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: br label [[COND_END]] // CHECK13: cond.end: // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13: omp.inner.for.cond: // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13: omp.inner.for.body: // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK13-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK13-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: call void @.omp_outlined..7(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK13-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13: omp.inner.for.inc: // CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK13-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK13: omp.inner.for.end: // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK13: omp.loop.exit: // CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK13-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK13-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK13: cond.true: // CHECK13-NEXT: br label [[COND_END:%.*]] // CHECK13: cond.false: // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: br label [[COND_END]] // CHECK13: cond.end: // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13: omp.inner.for.cond: // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK13-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13: omp.inner.for.body: // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK13-NEXT: call void @_Z3fn5v() // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK13: omp.body.continue: // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13: omp.inner.for.inc: // CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK13-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK13: omp.inner.for.end: // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK13: omp.loop.exit: // CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103 // CHECK13-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8 // CHECK13-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32* // CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i32* [[CONV]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[ARG_ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[ARG]], i32** [[ARG_ADDR]], align 8 // CHECK13-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARG_ADDR]], align 8 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK13-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK13: cond.true: // CHECK13-NEXT: br label [[COND_END:%.*]] // CHECK13: cond.false: // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: br label [[COND_END]] // CHECK13: cond.end: // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13: omp.inner.for.cond: // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13: omp.inner.for.body: // CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK13-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK13-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK13: omp_if.then: // CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]]) // CHECK13-NEXT: br label [[OMP_IF_END:%.*]] // CHECK13: omp_if.else: // CHECK13-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK13-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: call void @.omp_outlined..9(i32* [[TMP13]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]]) #[[ATTR2]] // CHECK13-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK13-NEXT: br label [[OMP_IF_END]] // CHECK13: omp_if.end: // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13: omp.inner.for.inc: // CHECK13-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK13-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK13: omp.inner.for.end: // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK13: omp.loop.exit: // CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK13-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK13-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK13: cond.true: // CHECK13-NEXT: br label [[COND_END:%.*]] // CHECK13: cond.false: // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: br label [[COND_END]] // CHECK13: cond.end: // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13: omp.inner.for.cond: // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK13-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13: omp.inner.for.body: // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK13-NEXT: call void @_Z3fn6v() // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK13: omp.body.continue: // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13: omp.inner.for.inc: // CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK13-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK13: omp.inner.for.end: // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK13: omp.loop.exit: // CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_ // CHECK13-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK13-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK13-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK13-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4 // CHECK13-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK13-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK13-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK13-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK13: omp_offload.failed: // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63() #[[ATTR2]] // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK13: omp_offload.cont: // CHECK13-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK13-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK13-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK13-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK13: omp_offload.failed2: // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69() #[[ATTR2]] // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK13: omp_offload.cont3: // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[ARG_ADDR]], align 4 // CHECK13-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32* // CHECK13-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4 // CHECK13-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8 // CHECK13-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK13-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK13-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8 // CHECK13-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK13-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK13-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8 // CHECK13-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK13-NEXT: store i8* null, i8** [[TMP10]], align 8 // CHECK13-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK13-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK13-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK13-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK13-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK13-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK13: omp_offload.failed5: // CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75(i64 [[TMP5]]) #[[ATTR2]] // CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK13: omp_offload.cont6: // CHECK13-NEXT: ret i32 0 // // // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63 // CHECK13-SAME: () #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*)) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK13-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK13: cond.true: // CHECK13-NEXT: br label [[COND_END:%.*]] // CHECK13: cond.false: // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: br label [[COND_END]] // CHECK13: cond.end: // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13: omp.inner.for.cond: // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13: omp.inner.for.body: // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13: omp.inner.for.inc: // CHECK13-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK13-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK13: omp.inner.for.end: // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK13: omp.loop.exit: // CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK13-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK13-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK13: cond.true: // CHECK13-NEXT: br label [[COND_END:%.*]] // CHECK13: cond.false: // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: br label [[COND_END]] // CHECK13: cond.end: // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13: omp.inner.for.cond: // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK13-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13: omp.inner.for.body: // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK13-NEXT: call void @_Z3fn1v() // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK13: omp.body.continue: // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13: omp.inner.for.inc: // CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK13-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK13: omp.inner.for.end: // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK13: omp.loop.exit: // CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69 // CHECK13-SAME: () #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..12 to void (i32*, i32*, ...)*)) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK13-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK13: cond.true: // CHECK13-NEXT: br label [[COND_END:%.*]] // CHECK13: cond.false: // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: br label [[COND_END]] // CHECK13: cond.end: // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13: omp.inner.for.cond: // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13: omp.inner.for.body: // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK13-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK13-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: call void @.omp_outlined..13(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK13-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13: omp.inner.for.inc: // CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK13-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK13: omp.inner.for.end: // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK13: omp.loop.exit: // CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK13-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK13-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK13: cond.true: // CHECK13-NEXT: br label [[COND_END:%.*]] // CHECK13: cond.false: // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: br label [[COND_END]] // CHECK13: cond.end: // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13: omp.inner.for.cond: // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK13-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13: omp.inner.for.body: // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK13-NEXT: call void @_Z3fn2v() // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK13: omp.body.continue: // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13: omp.inner.for.inc: // CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK13-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK13: omp.inner.for.end: // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK13: omp.loop.exit: // CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75 // CHECK13-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8 // CHECK13-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32* // CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32* [[CONV]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[ARG_ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[ARG]], i32** [[ARG_ADDR]], align 8 // CHECK13-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARG_ADDR]], align 8 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK13-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK13: cond.true: // CHECK13-NEXT: br label [[COND_END:%.*]] // CHECK13: cond.false: // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: br label [[COND_END]] // CHECK13: cond.end: // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13: omp.inner.for.cond: // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13: omp.inner.for.body: // CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK13-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK13-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK13-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK13-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK13: omp_if.then: // CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]]) // CHECK13-NEXT: br label [[OMP_IF_END:%.*]] // CHECK13: omp_if.else: // CHECK13-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK13-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: call void @.omp_outlined..15(i32* [[TMP13]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]]) #[[ATTR2]] // CHECK13-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK13-NEXT: br label [[OMP_IF_END]] // CHECK13: omp_if.end: // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13: omp.inner.for.inc: // CHECK13-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK13-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK13: omp.inner.for.end: // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK13: omp.loop.exit: // CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK13-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK13-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK13-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK13: cond.true: // CHECK13-NEXT: br label [[COND_END:%.*]] // CHECK13: cond.false: // CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: br label [[COND_END]] // CHECK13: cond.end: // CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13: omp.inner.for.cond: // CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK13-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK13-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13: omp.inner.for.body: // CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK13-NEXT: call void @_Z3fn3v() // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK13: omp.body.continue: // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13: omp.inner.for.inc: // CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK13-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK13: omp.inner.for.end: // CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK13: omp.loop.exit: // CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK13-NEXT: ret void // // // CHECK13-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK13-SAME: () #[[ATTR5:[0-9]+]] { // CHECK13-NEXT: entry: // CHECK13-NEXT: call void @__tgt_register_requires(i64 1) // CHECK13-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@_Z9gtid_testv // CHECK14-SAME: () #[[ATTR0:[0-9]+]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK14-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100) // CHECK14-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK14-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK14-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK14: omp_offload.failed: // CHECK14-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47() #[[ATTR2:[0-9]+]] // CHECK14-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK14: omp_offload.cont: // CHECK14-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK14-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK14-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK14-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK14: omp_offload.failed2: // CHECK14-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52() #[[ATTR2]] // CHECK14-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK14: omp_offload.cont3: // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l47 // CHECK14-SAME: () #[[ATTR1:[0-9]+]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_outlined. // CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK14-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK14: cond.true: // CHECK14-NEXT: br label [[COND_END:%.*]] // CHECK14: cond.false: // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: br label [[COND_END]] // CHECK14: cond.end: // CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK14: omp.inner.for.cond: // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK14-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK14: omp.inner.for.body: // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK14: omp.inner.for.inc: // CHECK14-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK14-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK14: omp.inner.for.end: // CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK14: omp.loop.exit: // CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..1 // CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK14-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK14-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK14-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK14-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK14-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK14-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK14-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK14: cond.true: // CHECK14-NEXT: br label [[COND_END:%.*]] // CHECK14: cond.false: // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: br label [[COND_END]] // CHECK14: cond.end: // CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK14: omp.inner.for.cond: // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK14-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK14: omp.inner.for.body: // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK14: omp.body.continue: // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK14: omp.inner.for.inc: // CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK14-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK14: omp.inner.for.end: // CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK14: omp.loop.exit: // CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9gtid_testv_l52 // CHECK14-SAME: () #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*)) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..2 // CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK14-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK14: cond.true: // CHECK14-NEXT: br label [[COND_END:%.*]] // CHECK14: cond.false: // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: br label [[COND_END]] // CHECK14: cond.end: // CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK14: omp.inner.for.cond: // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK14-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK14: omp.inner.for.body: // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK14-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK14-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: call void @.omp_outlined..3(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK14-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK14: omp.inner.for.inc: // CHECK14-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK14-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK14: omp.inner.for.end: // CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK14: omp.loop.exit: // CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK14-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK14-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK14-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK14-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK14-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK14-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK14-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK14: cond.true: // CHECK14-NEXT: br label [[COND_END:%.*]] // CHECK14: cond.false: // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: br label [[COND_END]] // CHECK14: cond.end: // CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK14: omp.inner.for.cond: // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK14-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK14: omp.inner.for.body: // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK14-NEXT: call void @_Z9gtid_testv() // CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK14: omp.body.continue: // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK14: omp.inner.for.inc: // CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK14-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK14: omp.inner.for.end: // CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK14: omp.loop.exit: // CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@main // CHECK14-SAME: () #[[ATTR3:[0-9]+]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK14-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK14-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK14-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK14-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK14-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK14-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK14-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK14: omp_offload.failed: // CHECK14-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85() #[[ATTR2]] // CHECK14-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK14: omp_offload.cont: // CHECK14-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK14-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK14-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK14-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK14: omp_offload.failed2: // CHECK14-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94() #[[ATTR2]] // CHECK14-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK14: omp_offload.cont3: // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* @Arg, align 4 // CHECK14-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32* // CHECK14-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4 // CHECK14-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8 // CHECK14-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK14-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK14-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8 // CHECK14-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK14-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK14-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8 // CHECK14-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK14-NEXT: store i8* null, i8** [[TMP10]], align 8 // CHECK14-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK14-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK14-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK14-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK14-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK14-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK14: omp_offload.failed5: // CHECK14-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103(i64 [[TMP5]]) #[[ATTR2]] // CHECK14-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK14: omp_offload.cont6: // CHECK14-NEXT: [[TMP15:%.*]] = load i32, i32* @Arg, align 4 // CHECK14-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiEiT_(i32 [[TMP15]]) // CHECK14-NEXT: ret i32 [[CALL]] // // // CHECK14-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l85 // CHECK14-SAME: () #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*)) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..4 // CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK14-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK14: cond.true: // CHECK14-NEXT: br label [[COND_END:%.*]] // CHECK14: cond.false: // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: br label [[COND_END]] // CHECK14: cond.end: // CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK14: omp.inner.for.cond: // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK14-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK14: omp.inner.for.body: // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK14: omp.inner.for.inc: // CHECK14-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK14-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK14: omp.inner.for.end: // CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK14: omp.loop.exit: // CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK14-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK14-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK14-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK14-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK14-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK14-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK14-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK14: cond.true: // CHECK14-NEXT: br label [[COND_END:%.*]] // CHECK14: cond.false: // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: br label [[COND_END]] // CHECK14: cond.end: // CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK14: omp.inner.for.cond: // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK14-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK14: omp.inner.for.body: // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK14-NEXT: call void @_Z3fn4v() // CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK14: omp.body.continue: // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK14: omp.inner.for.inc: // CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK14-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK14: omp.inner.for.end: // CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK14: omp.loop.exit: // CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94 // CHECK14-SAME: () #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*)) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK14-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK14: cond.true: // CHECK14-NEXT: br label [[COND_END:%.*]] // CHECK14: cond.false: // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: br label [[COND_END]] // CHECK14: cond.end: // CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK14: omp.inner.for.cond: // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK14-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK14: omp.inner.for.body: // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK14-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK14-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: call void @.omp_outlined..7(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK14-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK14: omp.inner.for.inc: // CHECK14-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK14-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK14: omp.inner.for.end: // CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK14: omp.loop.exit: // CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK14-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK14-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK14-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK14-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK14-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK14-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK14-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK14: cond.true: // CHECK14-NEXT: br label [[COND_END:%.*]] // CHECK14: cond.false: // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: br label [[COND_END]] // CHECK14: cond.end: // CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK14: omp.inner.for.cond: // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK14-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK14: omp.inner.for.body: // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK14-NEXT: call void @_Z3fn5v() // CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK14: omp.body.continue: // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK14: omp.inner.for.inc: // CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK14-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK14: omp.inner.for.end: // CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK14: omp.loop.exit: // CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l103 // CHECK14-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8 // CHECK14-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32* // CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i32* [[CONV]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[ARG_ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[ARG]], i32** [[ARG_ADDR]], align 8 // CHECK14-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARG_ADDR]], align 8 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK14-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 // CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99 // CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK14: cond.true: // CHECK14-NEXT: br label [[COND_END:%.*]] // CHECK14: cond.false: // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: br label [[COND_END]] // CHECK14: cond.end: // CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK14: omp.inner.for.cond: // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK14-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK14: omp.inner.for.body: // CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK14-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK14-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK14-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK14: omp_if.then: // CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]]) // CHECK14-NEXT: br label [[OMP_IF_END:%.*]] // CHECK14: omp_if.else: // CHECK14-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK14-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: call void @.omp_outlined..9(i32* [[TMP13]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]]) #[[ATTR2]] // CHECK14-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK14-NEXT: br label [[OMP_IF_END]] // CHECK14: omp_if.end: // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK14: omp.inner.for.inc: // CHECK14-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK14-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK14: omp.inner.for.end: // CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK14: omp.loop.exit: // CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK14-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK14-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK14-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK14-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK14-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK14-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK14-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK14: cond.true: // CHECK14-NEXT: br label [[COND_END:%.*]] // CHECK14: cond.false: // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: br label [[COND_END]] // CHECK14: cond.end: // CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK14: omp.inner.for.cond: // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK14-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK14: omp.inner.for.body: // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK14-NEXT: call void @_Z3fn6v() // CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK14: omp.body.continue: // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK14: omp.inner.for.inc: // CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK14-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK14: omp.inner.for.end: // CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK14: omp.loop.exit: // CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@_Z5tmainIiEiT_ // CHECK14-SAME: (i32 [[ARG:%.*]]) #[[ATTR0]] comdat { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[ARG_CASTED:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK14-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK14-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK14-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32 [[ARG]], i32* [[ARG_ADDR]], align 4 // CHECK14-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK14-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK14-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0 // CHECK14-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK14: omp_offload.failed: // CHECK14-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63() #[[ATTR2]] // CHECK14-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK14: omp_offload.cont: // CHECK14-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK14-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0) // CHECK14-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 // CHECK14-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]] // CHECK14: omp_offload.failed2: // CHECK14-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69() #[[ATTR2]] // CHECK14-NEXT: br label [[OMP_OFFLOAD_CONT3]] // CHECK14: omp_offload.cont3: // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[ARG_ADDR]], align 4 // CHECK14-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_CASTED]] to i32* // CHECK14-NEXT: store i32 [[TMP4]], i32* [[CONV]], align 4 // CHECK14-NEXT: [[TMP5:%.*]] = load i64, i64* [[ARG_CASTED]], align 8 // CHECK14-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK14-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK14-NEXT: store i64 [[TMP5]], i64* [[TMP7]], align 8 // CHECK14-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK14-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK14-NEXT: store i64 [[TMP5]], i64* [[TMP9]], align 8 // CHECK14-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK14-NEXT: store i8* null, i8** [[TMP10]], align 8 // CHECK14-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK14-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK14-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100) // CHECK14-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75.region_id, i32 1, i8** [[TMP11]], i8** [[TMP12]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK14-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK14-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK14: omp_offload.failed5: // CHECK14-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75(i64 [[TMP5]]) #[[ATTR2]] // CHECK14-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK14: omp_offload.cont6: // CHECK14-NEXT: ret i32 0 // // // CHECK14-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l63 // CHECK14-SAME: () #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*)) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK14-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK14: cond.true: // CHECK14-NEXT: br label [[COND_END:%.*]] // CHECK14: cond.false: // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: br label [[COND_END]] // CHECK14: cond.end: // CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK14: omp.inner.for.cond: // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK14-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK14: omp.inner.for.body: // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]]) // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK14: omp.inner.for.inc: // CHECK14-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK14-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK14: omp.inner.for.end: // CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK14: omp.loop.exit: // CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK14-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK14-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK14-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK14-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK14-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK14-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK14-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK14: cond.true: // CHECK14-NEXT: br label [[COND_END:%.*]] // CHECK14: cond.false: // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: br label [[COND_END]] // CHECK14: cond.end: // CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK14: omp.inner.for.cond: // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK14-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK14: omp.inner.for.body: // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK14-NEXT: call void @_Z3fn1v() // CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK14: omp.body.continue: // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK14: omp.inner.for.inc: // CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK14-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK14: omp.inner.for.end: // CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK14: omp.loop.exit: // CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l69 // CHECK14-SAME: () #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..12 to void (i32*, i32*, ...)*)) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK14-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK14: cond.true: // CHECK14-NEXT: br label [[COND_END:%.*]] // CHECK14: cond.false: // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: br label [[COND_END]] // CHECK14: cond.end: // CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK14: omp.inner.for.cond: // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK14-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK14: omp.inner.for.body: // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK14-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK14-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: call void @.omp_outlined..13(i32* [[TMP11]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP8]], i64 [[TMP10]]) #[[ATTR2]] // CHECK14-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]]) // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK14: omp.inner.for.inc: // CHECK14-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK14-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK14: omp.inner.for.end: // CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK14: omp.loop.exit: // CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK14-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK14-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK14-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK14-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK14-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK14-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK14-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK14: cond.true: // CHECK14-NEXT: br label [[COND_END:%.*]] // CHECK14: cond.false: // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: br label [[COND_END]] // CHECK14: cond.end: // CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK14: omp.inner.for.cond: // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK14-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK14: omp.inner.for.body: // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK14-NEXT: call void @_Z3fn2v() // CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK14: omp.body.continue: // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK14: omp.inner.for.inc: // CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK14-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK14: omp.inner.for.end: // CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK14: omp.loop.exit: // CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l75 // CHECK14-SAME: (i64 [[ARG:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[ARG_ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: store i64 [[ARG]], i64* [[ARG_ADDR]], align 8 // CHECK14-NEXT: [[CONV:%.*]] = bitcast i64* [[ARG_ADDR]] to i32* // CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32* [[CONV]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[ARG_ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[ARG]], i32** [[ARG_ADDR]], align 8 // CHECK14-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARG_ADDR]], align 8 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK14-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 // CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99 // CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK14: cond.true: // CHECK14-NEXT: br label [[COND_END:%.*]] // CHECK14: cond.false: // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: br label [[COND_END]] // CHECK14: cond.end: // CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK14: omp.inner.for.cond: // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK14-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK14: omp.inner.for.body: // CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK14-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK14-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK14-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP0]], align 4 // CHECK14-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK14-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK14: omp_if.then: // CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]]) // CHECK14-NEXT: br label [[OMP_IF_END:%.*]] // CHECK14: omp_if.else: // CHECK14-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK14-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: call void @.omp_outlined..15(i32* [[TMP13]], i32* [[DOTBOUND_ZERO_ADDR]], i64 [[TMP9]], i64 [[TMP11]]) #[[ATTR2]] // CHECK14-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]]) // CHECK14-NEXT: br label [[OMP_IF_END]] // CHECK14: omp_if.end: // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK14: omp.inner.for.inc: // CHECK14-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK14-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK14: omp.inner.for.end: // CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK14: omp.loop.exit: // CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR1]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK14-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 // CHECK14-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK14-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK14-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK14-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK14-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32 // CHECK14-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 // CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK14-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4 // CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99 // CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK14: cond.true: // CHECK14-NEXT: br label [[COND_END:%.*]] // CHECK14: cond.false: // CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: br label [[COND_END]] // CHECK14: cond.end: // CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK14-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK14: omp.inner.for.cond: // CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK14-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK14-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK14: omp.inner.for.body: // CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK14-NEXT: call void @_Z3fn3v() // CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK14: omp.body.continue: // CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK14: omp.inner.for.inc: // CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK14-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 // CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK14: omp.inner.for.end: // CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK14: omp.loop.exit: // CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK14-NEXT: ret void // // // CHECK14-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK14-SAME: () #[[ATTR5:[0-9]+]] { // CHECK14-NEXT: entry: // CHECK14-NEXT: call void @__tgt_register_requires(i64 1) // CHECK14-NEXT: ret void // //