mirror of
https://github.com/llvm/llvm-project.git
synced 2025-04-28 00:36:06 +00:00
16a385ba21
We already created a versioned `__tgt_kernel_arguments` struct but it
was only briefly used and its content was passed in isolation anyway.
This makes it hard to add more information in the future. With this
patch we fully embrace the struct as means to pass information from the
compiler to the plugin as part of a kernel launch.
The patch also extends and renames the struct, bumping the version
number to 2. Version 1 entries are auto-upgraded. This is in preparation
for "bare" kernel launches, per kernel dynamic shared memory, CUDA/HIP
lowering, etc.
The `__tgt_target_kernel_nowait` interface was deprecated as it was
unused. Once we actually implement support for something like that, we
can add an appropriate API.
Note: Only plugins with the `launch_kernel` interface are now supported.
That means that a new clang won't be able to use an old runtime.
An old clang can still use the new runtime since the libomptarget
interface did not change.
Differential Revision: https://reviews.llvm.org/D141232
3351 lines
242 KiB
C++
3351 lines
242 KiB
C++
// 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 _
|
|
// Test host codegen.
|
|
// RUN: %clang_cc1 -DHAS_INT128 -verify -fopenmp -fopenmp-version=50 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1
|
|
// RUN: %clang_cc1 -DHAS_INT128 -fopenmp -fopenmp-version=50 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
|
|
// RUN: %clang_cc1 -DHAS_INT128 -fopenmp -fopenmp-version=50 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1
|
|
// RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=50 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3
|
|
// RUN: %clang_cc1 -fopenmp -fopenmp-version=50 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
|
|
// RUN: %clang_cc1 -fopenmp -fopenmp-version=50 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK3
|
|
|
|
// Test target codegen - host bc file has to be created first.
|
|
// RUN: %clang_cc1 -DHAS_INT128 -verify -fopenmp -fopenmp-version=50 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm-bc %s -o %t-ppc-host.bc
|
|
// RUN: %clang_cc1 -DHAS_INT128 -verify -fopenmp -fopenmp-version=50 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix=CHECK5
|
|
// RUN: %clang_cc1 -DHAS_INT128 -fopenmp -fopenmp-version=50 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o %t %s
|
|
// RUN: %clang_cc1 -DHAS_INT128 -fopenmp -fopenmp-version=50 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK5
|
|
// RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=50 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm-bc %s -o %t-x86-host.bc
|
|
// RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=50 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix=CHECK7
|
|
// RUN: %clang_cc1 -fopenmp -fopenmp-version=50 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o %t %s
|
|
// RUN: %clang_cc1 -fopenmp -fopenmp-version=50 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK7
|
|
|
|
// expected-no-diagnostics
|
|
#ifndef HEADER
|
|
#define HEADER
|
|
|
|
|
|
void mapWithPrivate() {
|
|
int x, y;
|
|
#pragma omp target teams private(x) map(x,y) private(y)
|
|
;
|
|
}
|
|
|
|
void mapWithFirstprivate() {
|
|
int x, y;
|
|
#pragma omp target teams firstprivate(x) map(x,y) firstprivate(y)
|
|
;
|
|
}
|
|
|
|
void mapWithReduction() {
|
|
int x, y;
|
|
#pragma omp target teams reduction(+:x) map(x,y) reduction(+:y)
|
|
;
|
|
}
|
|
|
|
void mapFrom() {
|
|
int x;
|
|
#pragma omp target teams firstprivate(x) map(from:x)
|
|
;
|
|
}
|
|
|
|
void mapTo() {
|
|
int x;
|
|
#pragma omp target teams firstprivate(x) map(to:x)
|
|
;
|
|
}
|
|
|
|
void mapAlloc() {
|
|
int x;
|
|
#pragma omp target teams firstprivate(x) map(alloc:x)
|
|
;
|
|
}
|
|
|
|
void mapArray() {
|
|
int x[77], y[88], z[99];
|
|
#pragma omp target teams private(x) firstprivate(y) reduction(+:z) map(x,y,z)
|
|
;
|
|
#pragma omp target teams private(x) firstprivate(y) reduction(+:z) map(to:x,y,z)
|
|
;
|
|
}
|
|
|
|
# if HAS_INT128
|
|
void mapInt128() {
|
|
__int128 x, y, z;
|
|
#pragma omp target teams private(x) firstprivate(y) reduction(+:z) map(x,y,z)
|
|
;
|
|
#pragma omp target teams private(x) firstprivate(y) reduction(+:z) map(from:x,y,z)
|
|
;
|
|
}
|
|
# endif
|
|
#endif
|
|
// CHECK1-LABEL: define {{[^@]+}}@_Z14mapWithPrivatev
|
|
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4
|
|
// CHECK1-NEXT: [[Y:%.*]] = alloca i32, align 4
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP1]], align 8
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP3]], align 8
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP4]], align 8
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP5]], align 8
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store i32 2, ptr [[TMP8]], align 4
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store i32 2, ptr [[TMP9]], align 4
|
|
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8
|
|
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 8
|
|
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP12]], align 8
|
|
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP13]], align 8
|
|
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8
|
|
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP15]], align 8
|
|
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP16]], align 8
|
|
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP17]], align 8
|
|
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP18]], align 4
|
|
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP19]], align 4
|
|
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// CHECK1-NEXT: store i32 0, ptr [[TMP20]], align 4
|
|
// CHECK1-NEXT: [[TMP21:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27.region_id, ptr [[KERNEL_ARGS]])
|
|
// CHECK1-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
|
|
// CHECK1-NEXT: br i1 [[TMP22]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// CHECK1: omp_offload.failed:
|
|
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27() #[[ATTR2:[0-9]+]]
|
|
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
|
|
// CHECK1: omp_offload.cont:
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27
|
|
// CHECK1-SAME: () #[[ATTR1:[0-9]+]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @.omp_outlined.)
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
|
|
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4
|
|
// CHECK1-NEXT: [[Y:%.*]] = alloca i32, align 4
|
|
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@_Z19mapWithFirstprivatev
|
|
// CHECK1-SAME: () #[[ATTR0]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4
|
|
// CHECK1-NEXT: [[Y:%.*]] = alloca i32, align 4
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP1]], align 8
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP3]], align 8
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP4]], align 8
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP5]], align 8
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store i32 2, ptr [[TMP8]], align 4
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store i32 2, ptr [[TMP9]], align 4
|
|
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8
|
|
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 8
|
|
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// CHECK1-NEXT: store ptr @.offload_sizes.2, ptr [[TMP12]], align 8
|
|
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// CHECK1-NEXT: store ptr @.offload_maptypes.3, ptr [[TMP13]], align 8
|
|
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8
|
|
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP15]], align 8
|
|
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP16]], align 8
|
|
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP17]], align 8
|
|
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP18]], align 4
|
|
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP19]], align 4
|
|
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// CHECK1-NEXT: store i32 0, ptr [[TMP20]], align 4
|
|
// CHECK1-NEXT: [[TMP21:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33.region_id, ptr [[KERNEL_ARGS]])
|
|
// CHECK1-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
|
|
// CHECK1-NEXT: br i1 [[TMP22]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// CHECK1: omp_offload.failed:
|
|
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33(ptr [[X]], ptr [[Y]]) #[[ATTR2]]
|
|
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
|
|
// CHECK1: omp_offload.cont:
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33
|
|
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[Y:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[X_CASTED:%.*]] = alloca i64, align 8
|
|
// CHECK1-NEXT: [[Y_CASTED:%.*]] = alloca i64, align 8
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[X_CASTED]], align 4
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = load i64, ptr [[X_CASTED]], align 8
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4
|
|
// CHECK1-NEXT: store i32 [[TMP4]], ptr [[Y_CASTED]], align 4
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = load i64, ptr [[Y_CASTED]], align 8
|
|
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..1, i64 [[TMP3]], i64 [[TMP5]])
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1
|
|
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[X:%.*]], i64 noundef [[Y:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[X_ADDR:%.*]] = alloca i64, align 8
|
|
// CHECK1-NEXT: [[Y_ADDR:%.*]] = alloca i64, align 8
|
|
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store i64 [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK1-NEXT: store i64 [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@_Z16mapWithReductionv
|
|
// CHECK1-SAME: () #[[ATTR0]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4
|
|
// CHECK1-NEXT: [[Y:%.*]] = alloca i32, align 4
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP1]], align 8
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP3]], align 8
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP4]], align 8
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP5]], align 8
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store i32 2, ptr [[TMP8]], align 4
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store i32 2, ptr [[TMP9]], align 4
|
|
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8
|
|
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 8
|
|
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// CHECK1-NEXT: store ptr @.offload_sizes.5, ptr [[TMP12]], align 8
|
|
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// CHECK1-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP13]], align 8
|
|
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8
|
|
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP15]], align 8
|
|
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP16]], align 8
|
|
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP17]], align 8
|
|
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP18]], align 4
|
|
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP19]], align 4
|
|
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// CHECK1-NEXT: store i32 0, ptr [[TMP20]], align 4
|
|
// CHECK1-NEXT: [[TMP21:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39.region_id, ptr [[KERNEL_ARGS]])
|
|
// CHECK1-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
|
|
// CHECK1-NEXT: br i1 [[TMP22]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// CHECK1: omp_offload.failed:
|
|
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39(ptr [[X]], ptr [[Y]]) #[[ATTR2]]
|
|
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
|
|
// CHECK1: omp_offload.cont:
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39
|
|
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[Y:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..4, ptr [[TMP0]], ptr [[TMP1]])
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..4
|
|
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[Y:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[X1:%.*]] = alloca i32, align 4
|
|
// CHECK1-NEXT: [[Y2:%.*]] = alloca i32, align 4
|
|
// CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [2 x ptr], align 8
|
|
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: store i32 0, ptr [[X1]], align 4
|
|
// CHECK1-NEXT: store i32 0, ptr [[Y2]], align 4
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
|
|
// CHECK1-NEXT: store ptr [[X1]], ptr [[TMP2]], align 8
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
|
|
// CHECK1-NEXT: store ptr [[Y2]], ptr [[TMP3]], align 8
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP5]], i32 2, i64 16, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK1-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
|
|
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
|
|
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
|
|
// CHECK1-NEXT: ]
|
|
// CHECK1: .omp.reduction.case1:
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[X1]], align 4
|
|
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
|
|
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP0]], align 4
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP1]], align 4
|
|
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[Y2]], align 4
|
|
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
|
|
// CHECK1-NEXT: store i32 [[ADD3]], ptr [[TMP1]], align 4
|
|
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK1: .omp.reduction.case2:
|
|
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[X1]], align 4
|
|
// CHECK1-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP11]] monotonic, align 4
|
|
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[Y2]], align 4
|
|
// CHECK1-NEXT: [[TMP14:%.*]] = atomicrmw add ptr [[TMP1]], i32 [[TMP13]] monotonic, align 4
|
|
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK1: .omp.reduction.default:
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
|
|
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i64 0, i64 0
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 0
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i64 0, i64 1
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
|
|
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 1
|
|
// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[TMP10]], align 8
|
|
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP7]], align 4
|
|
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP5]], align 4
|
|
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
|
|
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
|
|
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP11]], align 4
|
|
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP9]], align 4
|
|
// CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
|
|
// CHECK1-NEXT: store i32 [[ADD2]], ptr [[TMP11]], align 4
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@_Z7mapFromv
|
|
// CHECK1-SAME: () #[[ATTR0]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP1]], align 8
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store i32 2, ptr [[TMP5]], align 4
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// CHECK1-NEXT: store ptr @.offload_sizes.8, ptr [[TMP9]], align 8
|
|
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// CHECK1-NEXT: store ptr @.offload_maptypes.9, ptr [[TMP10]], align 8
|
|
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
|
|
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
|
|
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
|
|
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
|
|
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
|
|
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
|
|
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
|
|
// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45.region_id, ptr [[KERNEL_ARGS]])
|
|
// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
|
|
// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// CHECK1: omp_offload.failed:
|
|
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45(ptr [[X]]) #[[ATTR2]]
|
|
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
|
|
// CHECK1: omp_offload.cont:
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45
|
|
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[X_CASTED:%.*]] = alloca i64, align 8
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK1-NEXT: store i32 [[TMP1]], ptr [[X_CASTED]], align 4
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = load i64, ptr [[X_CASTED]], align 8
|
|
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 1, ptr @.omp_outlined..7, i64 [[TMP2]])
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..7
|
|
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[X:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[X_ADDR:%.*]] = alloca i64, align 8
|
|
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store i64 [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@_Z5mapTov
|
|
// CHECK1-SAME: () #[[ATTR0]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP1]], align 8
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store i32 2, ptr [[TMP5]], align 4
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// CHECK1-NEXT: store ptr @.offload_sizes.11, ptr [[TMP9]], align 8
|
|
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// CHECK1-NEXT: store ptr @.offload_maptypes.12, ptr [[TMP10]], align 8
|
|
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
|
|
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
|
|
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
|
|
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
|
|
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
|
|
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
|
|
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
|
|
// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51.region_id, ptr [[KERNEL_ARGS]])
|
|
// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
|
|
// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// CHECK1: omp_offload.failed:
|
|
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51(ptr [[X]]) #[[ATTR2]]
|
|
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
|
|
// CHECK1: omp_offload.cont:
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51
|
|
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[X_CASTED:%.*]] = alloca i64, align 8
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK1-NEXT: store i32 [[TMP1]], ptr [[X_CASTED]], align 4
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = load i64, ptr [[X_CASTED]], align 8
|
|
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 1, ptr @.omp_outlined..10, i64 [[TMP2]])
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..10
|
|
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[X:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[X_ADDR:%.*]] = alloca i64, align 8
|
|
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store i64 [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@_Z8mapAllocv
|
|
// CHECK1-SAME: () #[[ATTR0]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP1]], align 8
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store i32 2, ptr [[TMP5]], align 4
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// CHECK1-NEXT: store ptr @.offload_sizes.14, ptr [[TMP9]], align 8
|
|
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// CHECK1-NEXT: store ptr @.offload_maptypes.15, ptr [[TMP10]], align 8
|
|
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
|
|
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
|
|
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
|
|
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
|
|
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
|
|
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
|
|
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
|
|
// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57.region_id, ptr [[KERNEL_ARGS]])
|
|
// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
|
|
// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// CHECK1: omp_offload.failed:
|
|
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57(ptr [[X]]) #[[ATTR2]]
|
|
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
|
|
// CHECK1: omp_offload.cont:
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57
|
|
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[X_CASTED:%.*]] = alloca i64, align 8
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK1-NEXT: store i32 [[TMP1]], ptr [[X_CASTED]], align 4
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = load i64, ptr [[X_CASTED]], align 8
|
|
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 1, ptr @.omp_outlined..13, i64 [[TMP2]])
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..13
|
|
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[X:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[X_ADDR:%.*]] = alloca i64, align 8
|
|
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store i64 [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@_Z8mapArrayv
|
|
// CHECK1-SAME: () #[[ATTR0]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[X:%.*]] = alloca [77 x i32], align 4
|
|
// CHECK1-NEXT: [[Y:%.*]] = alloca [88 x i32], align 4
|
|
// CHECK1-NEXT: [[Z:%.*]] = alloca [99 x i32], align 4
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [3 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [3 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [3 x ptr], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP0]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP1]], align 8
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store ptr [[Z]], ptr [[TMP3]], align 8
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store ptr [[Z]], ptr [[TMP4]], align 8
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP5]], align 8
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP6]], align 8
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP7]], align 8
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP8]], align 8
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store i32 2, ptr [[TMP11]], align 4
|
|
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store i32 3, ptr [[TMP12]], align 4
|
|
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP13]], align 8
|
|
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// CHECK1-NEXT: store ptr [[TMP10]], ptr [[TMP14]], align 8
|
|
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// CHECK1-NEXT: store ptr @.offload_sizes.18, ptr [[TMP15]], align 8
|
|
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// CHECK1-NEXT: store ptr @.offload_maptypes.19, ptr [[TMP16]], align 8
|
|
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP17]], align 8
|
|
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP18]], align 8
|
|
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP19]], align 8
|
|
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP20]], align 8
|
|
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP21]], align 4
|
|
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP22]], align 4
|
|
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// CHECK1-NEXT: store i32 0, ptr [[TMP23]], align 4
|
|
// CHECK1-NEXT: [[TMP24:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63.region_id, ptr [[KERNEL_ARGS]])
|
|
// CHECK1-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
|
|
// CHECK1-NEXT: br i1 [[TMP25]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// CHECK1: omp_offload.failed:
|
|
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63(ptr [[Y]], ptr [[Z]]) #[[ATTR2]]
|
|
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
|
|
// CHECK1: omp_offload.cont:
|
|
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP26]], align 8
|
|
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP27]], align 8
|
|
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP28]], align 8
|
|
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 1
|
|
// CHECK1-NEXT: store ptr [[Z]], ptr [[TMP29]], align 8
|
|
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 1
|
|
// CHECK1-NEXT: store ptr [[Z]], ptr [[TMP30]], align 8
|
|
// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 1
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP31]], align 8
|
|
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP32]], align 8
|
|
// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP33]], align 8
|
|
// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 2
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP34]], align 8
|
|
// CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[KERNEL_ARGS4:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
|
|
// CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 0
|
|
// CHECK1-NEXT: store i32 2, ptr [[TMP37]], align 4
|
|
// CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 1
|
|
// CHECK1-NEXT: store i32 3, ptr [[TMP38]], align 4
|
|
// CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[TMP35]], ptr [[TMP39]], align 8
|
|
// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 3
|
|
// CHECK1-NEXT: store ptr [[TMP36]], ptr [[TMP40]], align 8
|
|
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 4
|
|
// CHECK1-NEXT: store ptr @.offload_sizes.22, ptr [[TMP41]], align 8
|
|
// CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 5
|
|
// CHECK1-NEXT: store ptr @.offload_maptypes.23, ptr [[TMP42]], align 8
|
|
// CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 6
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP43]], align 8
|
|
// CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 7
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP44]], align 8
|
|
// CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 8
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP45]], align 8
|
|
// CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 9
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP46]], align 8
|
|
// CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 10
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP47]], align 4
|
|
// CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 11
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP48]], align 4
|
|
// CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 12
|
|
// CHECK1-NEXT: store i32 0, ptr [[TMP49]], align 4
|
|
// CHECK1-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65.region_id, ptr [[KERNEL_ARGS4]])
|
|
// CHECK1-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0
|
|
// CHECK1-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
|
|
// CHECK1: omp_offload.failed5:
|
|
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65(ptr [[Y]], ptr [[Z]]) #[[ATTR2]]
|
|
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT6]]
|
|
// CHECK1: omp_offload.cont6:
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63
|
|
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
|
|
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..16, ptr [[TMP0]], ptr [[TMP1]])
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..16
|
|
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Y1:%.*]] = alloca [88 x i32], align 4
|
|
// CHECK1-NEXT: [[X:%.*]] = alloca [77 x i32], align 4
|
|
// CHECK1-NEXT: [[Z2:%.*]] = alloca [99 x i32], align 4
|
|
// CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
|
|
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i64 352, i1 false)
|
|
// CHECK1-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 99
|
|
// CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
|
|
// CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
|
|
// CHECK1: omp.arrayinit.body:
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
|
|
// CHECK1-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
|
|
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
|
|
// CHECK1: omp.arrayinit.done:
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
|
|
// CHECK1-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.17, ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK1-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
|
|
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
|
|
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
|
|
// CHECK1-NEXT: ]
|
|
// CHECK1: .omp.reduction.case1:
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP7]]
|
|
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
|
|
// CHECK1: omp.arraycpy.body:
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
|
|
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
|
|
// CHECK1-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP7]]
|
|
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
|
|
// CHECK1: omp.arraycpy.done6:
|
|
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK1: .omp.reduction.case2:
|
|
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP10]]
|
|
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
|
|
// CHECK1: omp.arraycpy.body8:
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
|
|
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
|
|
// CHECK1-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP11]] monotonic, align 4
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP10]]
|
|
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
|
|
// CHECK1: omp.arraycpy.done14:
|
|
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK1: .omp.reduction.default:
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.17
|
|
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i64 99
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
|
|
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
|
|
// CHECK1: omp.arraycpy.body:
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
|
|
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
|
|
// CHECK1-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
|
|
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
|
|
// CHECK1: omp.arraycpy.done2:
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65
|
|
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
|
|
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..20, ptr [[TMP0]], ptr [[TMP1]])
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..20
|
|
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Y1:%.*]] = alloca [88 x i32], align 4
|
|
// CHECK1-NEXT: [[X:%.*]] = alloca [77 x i32], align 4
|
|
// CHECK1-NEXT: [[Z2:%.*]] = alloca [99 x i32], align 4
|
|
// CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
|
|
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i64 352, i1 false)
|
|
// CHECK1-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 99
|
|
// CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
|
|
// CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
|
|
// CHECK1: omp.arrayinit.body:
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
|
|
// CHECK1-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
|
|
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
|
|
// CHECK1: omp.arrayinit.done:
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
|
|
// CHECK1-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.21, ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK1-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
|
|
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
|
|
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
|
|
// CHECK1-NEXT: ]
|
|
// CHECK1: .omp.reduction.case1:
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP7]]
|
|
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
|
|
// CHECK1: omp.arraycpy.body:
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
|
|
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
|
|
// CHECK1-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP7]]
|
|
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
|
|
// CHECK1: omp.arraycpy.done6:
|
|
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK1: .omp.reduction.case2:
|
|
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP10]]
|
|
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
|
|
// CHECK1: omp.arraycpy.body8:
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
|
|
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
|
|
// CHECK1-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP11]] monotonic, align 4
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP10]]
|
|
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
|
|
// CHECK1: omp.arraycpy.done14:
|
|
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK1: .omp.reduction.default:
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.21
|
|
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i64 99
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
|
|
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
|
|
// CHECK1: omp.arraycpy.body:
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
|
|
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
|
|
// CHECK1-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
|
|
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
|
|
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
|
|
// CHECK1: omp.arraycpy.done2:
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@_Z9mapInt128v
|
|
// CHECK1-SAME: () #[[ATTR0]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[X:%.*]] = alloca i128, align 16
|
|
// CHECK1-NEXT: [[Y:%.*]] = alloca i128, align 16
|
|
// CHECK1-NEXT: [[Z:%.*]] = alloca i128, align 16
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [3 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [3 x ptr], align 8
|
|
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [3 x ptr], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP0]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP1]], align 8
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store ptr [[Z]], ptr [[TMP3]], align 8
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store ptr [[Z]], ptr [[TMP4]], align 8
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP5]], align 8
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP6]], align 8
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP7]], align 8
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP8]], align 8
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// CHECK1-NEXT: store i32 2, ptr [[TMP11]], align 4
|
|
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// CHECK1-NEXT: store i32 3, ptr [[TMP12]], align 4
|
|
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP13]], align 8
|
|
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// CHECK1-NEXT: store ptr [[TMP10]], ptr [[TMP14]], align 8
|
|
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// CHECK1-NEXT: store ptr @.offload_sizes.26, ptr [[TMP15]], align 8
|
|
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// CHECK1-NEXT: store ptr @.offload_maptypes.27, ptr [[TMP16]], align 8
|
|
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP17]], align 8
|
|
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP18]], align 8
|
|
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP19]], align 8
|
|
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP20]], align 8
|
|
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP21]], align 4
|
|
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP22]], align 4
|
|
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// CHECK1-NEXT: store i32 0, ptr [[TMP23]], align 4
|
|
// CHECK1-NEXT: [[TMP24:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l72.region_id, ptr [[KERNEL_ARGS]])
|
|
// CHECK1-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
|
|
// CHECK1-NEXT: br i1 [[TMP25]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// CHECK1: omp_offload.failed:
|
|
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l72(ptr [[Y]], ptr [[Z]]) #[[ATTR2]]
|
|
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
|
|
// CHECK1: omp_offload.cont:
|
|
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP26]], align 8
|
|
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP27]], align 8
|
|
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP28]], align 8
|
|
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 1
|
|
// CHECK1-NEXT: store ptr [[Z]], ptr [[TMP29]], align 8
|
|
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 1
|
|
// CHECK1-NEXT: store ptr [[Z]], ptr [[TMP30]], align 8
|
|
// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 1
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP31]], align 8
|
|
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP32]], align 8
|
|
// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP33]], align 8
|
|
// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 2
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP34]], align 8
|
|
// CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
|
|
// CHECK1-NEXT: [[KERNEL_ARGS4:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
|
|
// CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 0
|
|
// CHECK1-NEXT: store i32 2, ptr [[TMP37]], align 4
|
|
// CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 1
|
|
// CHECK1-NEXT: store i32 3, ptr [[TMP38]], align 4
|
|
// CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 2
|
|
// CHECK1-NEXT: store ptr [[TMP35]], ptr [[TMP39]], align 8
|
|
// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 3
|
|
// CHECK1-NEXT: store ptr [[TMP36]], ptr [[TMP40]], align 8
|
|
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 4
|
|
// CHECK1-NEXT: store ptr @.offload_sizes.30, ptr [[TMP41]], align 8
|
|
// CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 5
|
|
// CHECK1-NEXT: store ptr @.offload_maptypes.31, ptr [[TMP42]], align 8
|
|
// CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 6
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP43]], align 8
|
|
// CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 7
|
|
// CHECK1-NEXT: store ptr null, ptr [[TMP44]], align 8
|
|
// CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 8
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP45]], align 8
|
|
// CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 9
|
|
// CHECK1-NEXT: store i64 0, ptr [[TMP46]], align 8
|
|
// CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 10
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP47]], align 4
|
|
// CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 11
|
|
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP48]], align 4
|
|
// CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 12
|
|
// CHECK1-NEXT: store i32 0, ptr [[TMP49]], align 4
|
|
// CHECK1-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l74.region_id, ptr [[KERNEL_ARGS4]])
|
|
// CHECK1-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0
|
|
// CHECK1-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
|
|
// CHECK1: omp_offload.failed5:
|
|
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l74(ptr [[Y]], ptr [[Z]]) #[[ATTR2]]
|
|
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT6]]
|
|
// CHECK1: omp_offload.cont6:
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l72
|
|
// CHECK1-SAME: (ptr noundef nonnull align 16 dereferenceable(16) [[Y:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Z:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
|
|
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..24, ptr [[TMP0]], ptr [[TMP1]])
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..24
|
|
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Y:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Z:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Y1:%.*]] = alloca i128, align 16
|
|
// CHECK1-NEXT: [[X:%.*]] = alloca i128, align 16
|
|
// CHECK1-NEXT: [[Z2:%.*]] = alloca i128, align 16
|
|
// CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK1-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i128, align 16
|
|
// CHECK1-NEXT: [[ATOMIC_TEMP3:%.*]] = alloca i128, align 16
|
|
// CHECK1-NEXT: [[TMP:%.*]] = alloca i128, align 16
|
|
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = load i128, ptr [[TMP0]], align 16
|
|
// CHECK1-NEXT: store i128 [[TMP2]], ptr [[Y1]], align 16
|
|
// CHECK1-NEXT: store i128 0, ptr [[Z2]], align 16
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
|
|
// CHECK1-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.25, ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK1-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
|
|
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
|
|
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
|
|
// CHECK1-NEXT: ]
|
|
// CHECK1: .omp.reduction.case1:
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = load i128, ptr [[TMP1]], align 16
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = load i128, ptr [[Z2]], align 16
|
|
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP7]], [[TMP8]]
|
|
// CHECK1-NEXT: store i128 [[ADD]], ptr [[TMP1]], align 16
|
|
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK1: .omp.reduction.case2:
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = load i128, ptr [[Z2]], align 16
|
|
// CHECK1-NEXT: call void @__atomic_load(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], i32 noundef signext 0)
|
|
// CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]]
|
|
// CHECK1: atomic_cont:
|
|
// CHECK1-NEXT: [[TMP10:%.*]] = load i128, ptr [[ATOMIC_TEMP]], align 16
|
|
// CHECK1-NEXT: store i128 [[TMP10]], ptr [[TMP]], align 16
|
|
// CHECK1-NEXT: [[TMP11:%.*]] = load i128, ptr [[TMP]], align 16
|
|
// CHECK1-NEXT: [[TMP12:%.*]] = load i128, ptr [[Z2]], align 16
|
|
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i128 [[TMP11]], [[TMP12]]
|
|
// CHECK1-NEXT: store i128 [[ADD4]], ptr [[ATOMIC_TEMP3]], align 16
|
|
// CHECK1-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @__atomic_compare_exchange(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], ptr noundef [[ATOMIC_TEMP3]], i32 noundef signext 0, i32 noundef signext 0)
|
|
// CHECK1-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]]
|
|
// CHECK1: atomic_exit:
|
|
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK1: .omp.reduction.default:
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.25
|
|
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = load i128, ptr [[TMP7]], align 16
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = load i128, ptr [[TMP5]], align 16
|
|
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP8]], [[TMP9]]
|
|
// CHECK1-NEXT: store i128 [[ADD]], ptr [[TMP7]], align 16
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l74
|
|
// CHECK1-SAME: (ptr noundef nonnull align 16 dereferenceable(16) [[Y:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Z:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
|
|
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..28, ptr [[TMP0]], ptr [[TMP1]])
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..28
|
|
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Y:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Z:%.*]]) #[[ATTR1]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[Y1:%.*]] = alloca i128, align 16
|
|
// CHECK1-NEXT: [[X:%.*]] = alloca i128, align 16
|
|
// CHECK1-NEXT: [[Z2:%.*]] = alloca i128, align 16
|
|
// CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK1-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i128, align 16
|
|
// CHECK1-NEXT: [[ATOMIC_TEMP3:%.*]] = alloca i128, align 16
|
|
// CHECK1-NEXT: [[TMP:%.*]] = alloca i128, align 16
|
|
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = load i128, ptr [[TMP0]], align 16
|
|
// CHECK1-NEXT: store i128 [[TMP2]], ptr [[Y1]], align 16
|
|
// CHECK1-NEXT: store i128 0, ptr [[Z2]], align 16
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
|
|
// CHECK1-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.29, ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK1-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
|
|
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
|
|
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
|
|
// CHECK1-NEXT: ]
|
|
// CHECK1: .omp.reduction.case1:
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = load i128, ptr [[TMP1]], align 16
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = load i128, ptr [[Z2]], align 16
|
|
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP7]], [[TMP8]]
|
|
// CHECK1-NEXT: store i128 [[ADD]], ptr [[TMP1]], align 16
|
|
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK1: .omp.reduction.case2:
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = load i128, ptr [[Z2]], align 16
|
|
// CHECK1-NEXT: call void @__atomic_load(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], i32 noundef signext 0)
|
|
// CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]]
|
|
// CHECK1: atomic_cont:
|
|
// CHECK1-NEXT: [[TMP10:%.*]] = load i128, ptr [[ATOMIC_TEMP]], align 16
|
|
// CHECK1-NEXT: store i128 [[TMP10]], ptr [[TMP]], align 16
|
|
// CHECK1-NEXT: [[TMP11:%.*]] = load i128, ptr [[TMP]], align 16
|
|
// CHECK1-NEXT: [[TMP12:%.*]] = load i128, ptr [[Z2]], align 16
|
|
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i128 [[TMP11]], [[TMP12]]
|
|
// CHECK1-NEXT: store i128 [[ADD4]], ptr [[ATOMIC_TEMP3]], align 16
|
|
// CHECK1-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @__atomic_compare_exchange(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], ptr noundef [[ATOMIC_TEMP3]], i32 noundef signext 0, i32 noundef signext 0)
|
|
// CHECK1-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]]
|
|
// CHECK1: atomic_exit:
|
|
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK1: .omp.reduction.default:
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.29
|
|
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
|
|
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
|
|
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
|
|
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
|
|
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
|
|
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
|
|
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
|
|
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
|
|
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
|
|
// CHECK1-NEXT: [[TMP8:%.*]] = load i128, ptr [[TMP7]], align 16
|
|
// CHECK1-NEXT: [[TMP9:%.*]] = load i128, ptr [[TMP5]], align 16
|
|
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP8]], [[TMP9]]
|
|
// CHECK1-NEXT: store i128 [[ADD]], ptr [[TMP7]], align 16
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
|
|
// CHECK1-SAME: () #[[ATTR7:[0-9]+]] {
|
|
// CHECK1-NEXT: entry:
|
|
// CHECK1-NEXT: call void @__tgt_register_requires(i64 1)
|
|
// CHECK1-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@_Z14mapWithPrivatev
|
|
// CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: [[Y:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP1]], align 4
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
|
|
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP3]], align 4
|
|
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP4]], align 4
|
|
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP5]], align 4
|
|
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store i32 2, ptr [[TMP8]], align 4
|
|
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store i32 2, ptr [[TMP9]], align 4
|
|
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4
|
|
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 4
|
|
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP12]], align 4
|
|
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP13]], align 4
|
|
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4
|
|
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP15]], align 4
|
|
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// CHECK3-NEXT: store i64 0, ptr [[TMP16]], align 8
|
|
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// CHECK3-NEXT: store i64 0, ptr [[TMP17]], align 8
|
|
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP18]], align 4
|
|
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP19]], align 4
|
|
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// CHECK3-NEXT: store i32 0, ptr [[TMP20]], align 4
|
|
// CHECK3-NEXT: [[TMP21:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27.region_id, ptr [[KERNEL_ARGS]])
|
|
// CHECK3-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
|
|
// CHECK3-NEXT: br i1 [[TMP22]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// CHECK3: omp_offload.failed:
|
|
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27() #[[ATTR2:[0-9]+]]
|
|
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
|
|
// CHECK3: omp_offload.cont:
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27
|
|
// CHECK3-SAME: () #[[ATTR1:[0-9]+]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @.omp_outlined.)
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined.
|
|
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: [[Y:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@_Z19mapWithFirstprivatev
|
|
// CHECK3-SAME: () #[[ATTR0]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: [[Y:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP1]], align 4
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
|
|
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP3]], align 4
|
|
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP4]], align 4
|
|
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP5]], align 4
|
|
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store i32 2, ptr [[TMP8]], align 4
|
|
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store i32 2, ptr [[TMP9]], align 4
|
|
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4
|
|
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 4
|
|
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// CHECK3-NEXT: store ptr @.offload_sizes.2, ptr [[TMP12]], align 4
|
|
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// CHECK3-NEXT: store ptr @.offload_maptypes.3, ptr [[TMP13]], align 4
|
|
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4
|
|
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP15]], align 4
|
|
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// CHECK3-NEXT: store i64 0, ptr [[TMP16]], align 8
|
|
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// CHECK3-NEXT: store i64 0, ptr [[TMP17]], align 8
|
|
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP18]], align 4
|
|
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP19]], align 4
|
|
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// CHECK3-NEXT: store i32 0, ptr [[TMP20]], align 4
|
|
// CHECK3-NEXT: [[TMP21:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33.region_id, ptr [[KERNEL_ARGS]])
|
|
// CHECK3-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
|
|
// CHECK3-NEXT: br i1 [[TMP22]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// CHECK3: omp_offload.failed:
|
|
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33(ptr [[X]], ptr [[Y]]) #[[ATTR2]]
|
|
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
|
|
// CHECK3: omp_offload.cont:
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33
|
|
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[Y:%.*]]) #[[ATTR1]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[X_CASTED:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: [[Y_CASTED:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK3-NEXT: store i32 [[TMP2]], ptr [[X_CASTED]], align 4
|
|
// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[X_CASTED]], align 4
|
|
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4
|
|
// CHECK3-NEXT: store i32 [[TMP4]], ptr [[Y_CASTED]], align 4
|
|
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[Y_CASTED]], align 4
|
|
// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..1, i32 [[TMP3]], i32 [[TMP5]])
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..1
|
|
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR1]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[X_ADDR:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: [[Y_ADDR:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: store i32 [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK3-NEXT: store i32 [[Y]], ptr [[Y_ADDR]], align 4
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@_Z16mapWithReductionv
|
|
// CHECK3-SAME: () #[[ATTR0]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: [[Y:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP1]], align 4
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
|
|
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP3]], align 4
|
|
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP4]], align 4
|
|
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP5]], align 4
|
|
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store i32 2, ptr [[TMP8]], align 4
|
|
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store i32 2, ptr [[TMP9]], align 4
|
|
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4
|
|
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 4
|
|
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// CHECK3-NEXT: store ptr @.offload_sizes.5, ptr [[TMP12]], align 4
|
|
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// CHECK3-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP13]], align 4
|
|
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4
|
|
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP15]], align 4
|
|
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// CHECK3-NEXT: store i64 0, ptr [[TMP16]], align 8
|
|
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// CHECK3-NEXT: store i64 0, ptr [[TMP17]], align 8
|
|
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP18]], align 4
|
|
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP19]], align 4
|
|
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// CHECK3-NEXT: store i32 0, ptr [[TMP20]], align 4
|
|
// CHECK3-NEXT: [[TMP21:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39.region_id, ptr [[KERNEL_ARGS]])
|
|
// CHECK3-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
|
|
// CHECK3-NEXT: br i1 [[TMP22]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// CHECK3: omp_offload.failed:
|
|
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39(ptr [[X]], ptr [[Y]]) #[[ATTR2]]
|
|
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
|
|
// CHECK3: omp_offload.cont:
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39
|
|
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[Y:%.*]]) #[[ATTR1]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
|
|
// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..4, ptr [[TMP0]], ptr [[TMP1]])
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..4
|
|
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[Y:%.*]]) #[[ATTR1]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[X1:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: [[Y2:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [2 x ptr], align 4
|
|
// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
|
|
// CHECK3-NEXT: store i32 0, ptr [[X1]], align 4
|
|
// CHECK3-NEXT: store i32 0, ptr [[Y2]], align 4
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[X1]], ptr [[TMP2]], align 4
|
|
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 1
|
|
// CHECK3-NEXT: store ptr [[Y2]], ptr [[TMP3]], align 4
|
|
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
|
|
// CHECK3-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP5]], i32 2, i32 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK3-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
|
|
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
|
|
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
|
|
// CHECK3-NEXT: ]
|
|
// CHECK3: .omp.reduction.case1:
|
|
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[X1]], align 4
|
|
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
|
|
// CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP0]], align 4
|
|
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP1]], align 4
|
|
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[Y2]], align 4
|
|
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
|
|
// CHECK3-NEXT: store i32 [[ADD3]], ptr [[TMP1]], align 4
|
|
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK3: .omp.reduction.case2:
|
|
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[X1]], align 4
|
|
// CHECK3-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP11]] monotonic, align 4
|
|
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[Y2]], align 4
|
|
// CHECK3-NEXT: [[TMP14:%.*]] = atomicrmw add ptr [[TMP1]], i32 [[TMP13]] monotonic, align 4
|
|
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK3: .omp.reduction.default:
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
|
|
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
|
|
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
|
|
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
|
|
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i32 0, i32 1
|
|
// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 4
|
|
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i32 0, i32 1
|
|
// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[TMP10]], align 4
|
|
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP7]], align 4
|
|
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP5]], align 4
|
|
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
|
|
// CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
|
|
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP11]], align 4
|
|
// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP9]], align 4
|
|
// CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
|
|
// CHECK3-NEXT: store i32 [[ADD2]], ptr [[TMP11]], align 4
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@_Z7mapFromv
|
|
// CHECK3-SAME: () #[[ATTR0]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP1]], align 4
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
|
|
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store i32 2, ptr [[TMP5]], align 4
|
|
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
|
|
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
|
|
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
|
|
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// CHECK3-NEXT: store ptr @.offload_sizes.8, ptr [[TMP9]], align 4
|
|
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// CHECK3-NEXT: store ptr @.offload_maptypes.9, ptr [[TMP10]], align 4
|
|
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
|
|
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
|
|
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
|
|
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
|
|
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
|
|
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
|
|
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
|
|
// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45.region_id, ptr [[KERNEL_ARGS]])
|
|
// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
|
|
// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// CHECK3: omp_offload.failed:
|
|
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45(ptr [[X]]) #[[ATTR2]]
|
|
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
|
|
// CHECK3: omp_offload.cont:
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45
|
|
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]]) #[[ATTR1]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[X_CASTED:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK3-NEXT: store i32 [[TMP1]], ptr [[X_CASTED]], align 4
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[X_CASTED]], align 4
|
|
// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 1, ptr @.omp_outlined..7, i32 [[TMP2]])
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..7
|
|
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[X:%.*]]) #[[ATTR1]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[X_ADDR:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: store i32 [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@_Z5mapTov
|
|
// CHECK3-SAME: () #[[ATTR0]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP1]], align 4
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
|
|
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store i32 2, ptr [[TMP5]], align 4
|
|
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
|
|
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
|
|
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
|
|
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// CHECK3-NEXT: store ptr @.offload_sizes.11, ptr [[TMP9]], align 4
|
|
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// CHECK3-NEXT: store ptr @.offload_maptypes.12, ptr [[TMP10]], align 4
|
|
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
|
|
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
|
|
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
|
|
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
|
|
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
|
|
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
|
|
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
|
|
// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51.region_id, ptr [[KERNEL_ARGS]])
|
|
// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
|
|
// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// CHECK3: omp_offload.failed:
|
|
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51(ptr [[X]]) #[[ATTR2]]
|
|
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
|
|
// CHECK3: omp_offload.cont:
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51
|
|
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]]) #[[ATTR1]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[X_CASTED:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK3-NEXT: store i32 [[TMP1]], ptr [[X_CASTED]], align 4
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[X_CASTED]], align 4
|
|
// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 1, ptr @.omp_outlined..10, i32 [[TMP2]])
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..10
|
|
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[X:%.*]]) #[[ATTR1]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[X_ADDR:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: store i32 [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@_Z8mapAllocv
|
|
// CHECK3-SAME: () #[[ATTR0]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP1]], align 4
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
|
|
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store i32 2, ptr [[TMP5]], align 4
|
|
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
|
|
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
|
|
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
|
|
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// CHECK3-NEXT: store ptr @.offload_sizes.14, ptr [[TMP9]], align 4
|
|
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// CHECK3-NEXT: store ptr @.offload_maptypes.15, ptr [[TMP10]], align 4
|
|
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
|
|
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
|
|
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
|
|
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
|
|
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
|
|
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
|
|
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
|
|
// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57.region_id, ptr [[KERNEL_ARGS]])
|
|
// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
|
|
// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// CHECK3: omp_offload.failed:
|
|
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57(ptr [[X]]) #[[ATTR2]]
|
|
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
|
|
// CHECK3: omp_offload.cont:
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57
|
|
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]]) #[[ATTR1]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[X_CASTED:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK3-NEXT: store i32 [[TMP1]], ptr [[X_CASTED]], align 4
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[X_CASTED]], align 4
|
|
// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 1, ptr @.omp_outlined..13, i32 [[TMP2]])
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..13
|
|
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[X:%.*]]) #[[ATTR1]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[X_ADDR:%.*]] = alloca i32, align 4
|
|
// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: store i32 [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@_Z8mapArrayv
|
|
// CHECK3-SAME: () #[[ATTR0]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[X:%.*]] = alloca [77 x i32], align 4
|
|
// CHECK3-NEXT: [[Y:%.*]] = alloca [88 x i32], align 4
|
|
// CHECK3-NEXT: [[Z:%.*]] = alloca [99 x i32], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [3 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [3 x ptr], align 4
|
|
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [3 x ptr], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP0]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP1]], align 4
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
|
|
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store ptr [[Z]], ptr [[TMP3]], align 4
|
|
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store ptr [[Z]], ptr [[TMP4]], align 4
|
|
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP5]], align 4
|
|
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP6]], align 4
|
|
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP7]], align 4
|
|
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP8]], align 4
|
|
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// CHECK3-NEXT: store i32 2, ptr [[TMP11]], align 4
|
|
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// CHECK3-NEXT: store i32 3, ptr [[TMP12]], align 4
|
|
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP13]], align 4
|
|
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// CHECK3-NEXT: store ptr [[TMP10]], ptr [[TMP14]], align 4
|
|
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// CHECK3-NEXT: store ptr @.offload_sizes.18, ptr [[TMP15]], align 4
|
|
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// CHECK3-NEXT: store ptr @.offload_maptypes.19, ptr [[TMP16]], align 4
|
|
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP17]], align 4
|
|
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP18]], align 4
|
|
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// CHECK3-NEXT: store i64 0, ptr [[TMP19]], align 8
|
|
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// CHECK3-NEXT: store i64 0, ptr [[TMP20]], align 8
|
|
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP21]], align 4
|
|
// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP22]], align 4
|
|
// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// CHECK3-NEXT: store i32 0, ptr [[TMP23]], align 4
|
|
// CHECK3-NEXT: [[TMP24:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63.region_id, ptr [[KERNEL_ARGS]])
|
|
// CHECK3-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
|
|
// CHECK3-NEXT: br i1 [[TMP25]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// CHECK3: omp_offload.failed:
|
|
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63(ptr [[Y]], ptr [[Z]]) #[[ATTR2]]
|
|
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
|
|
// CHECK3: omp_offload.cont:
|
|
// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP26]], align 4
|
|
// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP27]], align 4
|
|
// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP28]], align 4
|
|
// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 1
|
|
// CHECK3-NEXT: store ptr [[Z]], ptr [[TMP29]], align 4
|
|
// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 1
|
|
// CHECK3-NEXT: store ptr [[Z]], ptr [[TMP30]], align 4
|
|
// CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 1
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP31]], align 4
|
|
// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 2
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP32]], align 4
|
|
// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 2
|
|
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP33]], align 4
|
|
// CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 2
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP34]], align 4
|
|
// CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[KERNEL_ARGS4:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
|
|
// CHECK3-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 0
|
|
// CHECK3-NEXT: store i32 2, ptr [[TMP37]], align 4
|
|
// CHECK3-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 1
|
|
// CHECK3-NEXT: store i32 3, ptr [[TMP38]], align 4
|
|
// CHECK3-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 2
|
|
// CHECK3-NEXT: store ptr [[TMP35]], ptr [[TMP39]], align 4
|
|
// CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 3
|
|
// CHECK3-NEXT: store ptr [[TMP36]], ptr [[TMP40]], align 4
|
|
// CHECK3-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 4
|
|
// CHECK3-NEXT: store ptr @.offload_sizes.22, ptr [[TMP41]], align 4
|
|
// CHECK3-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 5
|
|
// CHECK3-NEXT: store ptr @.offload_maptypes.23, ptr [[TMP42]], align 4
|
|
// CHECK3-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 6
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP43]], align 4
|
|
// CHECK3-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 7
|
|
// CHECK3-NEXT: store ptr null, ptr [[TMP44]], align 4
|
|
// CHECK3-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 8
|
|
// CHECK3-NEXT: store i64 0, ptr [[TMP45]], align 8
|
|
// CHECK3-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 9
|
|
// CHECK3-NEXT: store i64 0, ptr [[TMP46]], align 8
|
|
// CHECK3-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 10
|
|
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP47]], align 4
|
|
// CHECK3-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 11
|
|
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP48]], align 4
|
|
// CHECK3-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 12
|
|
// CHECK3-NEXT: store i32 0, ptr [[TMP49]], align 4
|
|
// CHECK3-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65.region_id, ptr [[KERNEL_ARGS4]])
|
|
// CHECK3-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0
|
|
// CHECK3-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
|
|
// CHECK3: omp_offload.failed5:
|
|
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65(ptr [[Y]], ptr [[Z]]) #[[ATTR2]]
|
|
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT6]]
|
|
// CHECK3: omp_offload.cont6:
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63
|
|
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR1]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 4
|
|
// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..16, ptr [[TMP0]], ptr [[TMP1]])
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..16
|
|
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR1]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[Y1:%.*]] = alloca [88 x i32], align 4
|
|
// CHECK3-NEXT: [[X:%.*]] = alloca [77 x i32], align 4
|
|
// CHECK3-NEXT: [[Z2:%.*]] = alloca [99 x i32], align 4
|
|
// CHECK3-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 4
|
|
// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 4
|
|
// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i32 352, i1 false)
|
|
// CHECK3-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i32 99
|
|
// CHECK3-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
|
|
// CHECK3-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
|
|
// CHECK3: omp.arrayinit.body:
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
|
|
// CHECK3-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
|
|
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
|
|
// CHECK3: omp.arrayinit.done:
|
|
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 4
|
|
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
|
|
// CHECK3-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.17, ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK3-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
|
|
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
|
|
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
|
|
// CHECK3-NEXT: ]
|
|
// CHECK3: .omp.reduction.case1:
|
|
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP7]]
|
|
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
|
|
// CHECK3: omp.arraycpy.body:
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
|
|
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
|
|
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
|
|
// CHECK3-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP7]]
|
|
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
|
|
// CHECK3: omp.arraycpy.done6:
|
|
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK3: .omp.reduction.case2:
|
|
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP10]]
|
|
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
|
|
// CHECK3: omp.arraycpy.body8:
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
|
|
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
|
|
// CHECK3-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP11]] monotonic, align 4
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP10]]
|
|
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
|
|
// CHECK3: omp.arraycpy.done14:
|
|
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK3: .omp.reduction.default:
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.17
|
|
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
|
|
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
|
|
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
|
|
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i32 99
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
|
|
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
|
|
// CHECK3: omp.arraycpy.body:
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
|
|
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
|
|
// CHECK3-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
|
|
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
|
|
// CHECK3: omp.arraycpy.done2:
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65
|
|
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR1]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 4
|
|
// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..20, ptr [[TMP0]], ptr [[TMP1]])
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..20
|
|
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR1]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[Y1:%.*]] = alloca [88 x i32], align 4
|
|
// CHECK3-NEXT: [[X:%.*]] = alloca [77 x i32], align 4
|
|
// CHECK3-NEXT: [[Z2:%.*]] = alloca [99 x i32], align 4
|
|
// CHECK3-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 4
|
|
// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 4
|
|
// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i32 352, i1 false)
|
|
// CHECK3-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i32 99
|
|
// CHECK3-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
|
|
// CHECK3-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
|
|
// CHECK3: omp.arrayinit.body:
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
|
|
// CHECK3-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
|
|
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
|
|
// CHECK3: omp.arrayinit.done:
|
|
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
|
|
// CHECK3-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 4
|
|
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
|
|
// CHECK3-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.21, ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK3-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
|
|
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
|
|
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
|
|
// CHECK3-NEXT: ]
|
|
// CHECK3: .omp.reduction.case1:
|
|
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP7]]
|
|
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
|
|
// CHECK3: omp.arraycpy.body:
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
|
|
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
|
|
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
|
|
// CHECK3-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP7]]
|
|
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
|
|
// CHECK3: omp.arraycpy.done6:
|
|
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK3: .omp.reduction.case2:
|
|
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP10]]
|
|
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
|
|
// CHECK3: omp.arraycpy.body8:
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
|
|
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
|
|
// CHECK3-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP11]] monotonic, align 4
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP10]]
|
|
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
|
|
// CHECK3: omp.arraycpy.done14:
|
|
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK3: .omp.reduction.default:
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.21
|
|
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
|
|
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
|
|
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
|
|
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
|
|
// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
|
|
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
|
|
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
|
|
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
|
|
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i32 99
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
|
|
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
|
|
// CHECK3: omp.arraycpy.body:
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
|
|
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
|
|
// CHECK3-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
|
|
// CHECK3-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
|
|
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
|
|
// CHECK3: omp.arraycpy.done2:
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
|
|
// CHECK3-SAME: () #[[ATTR6:[0-9]+]] {
|
|
// CHECK3-NEXT: entry:
|
|
// CHECK3-NEXT: call void @__tgt_register_requires(i64 1)
|
|
// CHECK3-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27
|
|
// CHECK5-SAME: () #[[ATTR0:[0-9]+]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1:[0-9]+]], i32 0, ptr @.omp_outlined.)
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined.
|
|
// CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[X:%.*]] = alloca i32, align 4
|
|
// CHECK5-NEXT: [[Y:%.*]] = alloca i32, align 4
|
|
// CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33
|
|
// CHECK5-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[Y:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[X_CASTED:%.*]] = alloca i64, align 8
|
|
// CHECK5-NEXT: [[Y_CASTED:%.*]] = alloca i64, align 8
|
|
// CHECK5-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK5-NEXT: store i32 [[TMP2]], ptr [[X_CASTED]], align 4
|
|
// CHECK5-NEXT: [[TMP3:%.*]] = load i64, ptr [[X_CASTED]], align 8
|
|
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4
|
|
// CHECK5-NEXT: store i32 [[TMP4]], ptr [[Y_CASTED]], align 4
|
|
// CHECK5-NEXT: [[TMP5:%.*]] = load i64, ptr [[Y_CASTED]], align 8
|
|
// CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..1, i64 [[TMP3]], i64 [[TMP5]])
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..1
|
|
// CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[X:%.*]], i64 noundef [[Y:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[X_ADDR:%.*]] = alloca i64, align 8
|
|
// CHECK5-NEXT: [[Y_ADDR:%.*]] = alloca i64, align 8
|
|
// CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store i64 [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK5-NEXT: store i64 [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39
|
|
// CHECK5-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[Y:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..2, ptr [[TMP0]], ptr [[TMP1]])
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..2
|
|
// CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[Y:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[X1:%.*]] = alloca i32, align 4
|
|
// CHECK5-NEXT: [[Y2:%.*]] = alloca i32, align 4
|
|
// CHECK5-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [2 x ptr], align 8
|
|
// CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: store i32 0, ptr [[X1]], align 4
|
|
// CHECK5-NEXT: store i32 0, ptr [[Y2]], align 4
|
|
// CHECK5-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
|
|
// CHECK5-NEXT: store ptr [[X1]], ptr [[TMP2]], align 8
|
|
// CHECK5-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
|
|
// CHECK5-NEXT: store ptr [[Y2]], ptr [[TMP3]], align 8
|
|
// CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
|
|
// CHECK5-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP5]], i32 2, i64 16, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK5-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
|
|
// CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
|
|
// CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
|
|
// CHECK5-NEXT: ]
|
|
// CHECK5: .omp.reduction.case1:
|
|
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[X1]], align 4
|
|
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
|
|
// CHECK5-NEXT: store i32 [[ADD]], ptr [[TMP0]], align 4
|
|
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP1]], align 4
|
|
// CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[Y2]], align 4
|
|
// CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
|
|
// CHECK5-NEXT: store i32 [[ADD3]], ptr [[TMP1]], align 4
|
|
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK5: .omp.reduction.case2:
|
|
// CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[X1]], align 4
|
|
// CHECK5-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP11]] monotonic, align 4
|
|
// CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[Y2]], align 4
|
|
// CHECK5-NEXT: [[TMP14:%.*]] = atomicrmw add ptr [[TMP1]], i32 [[TMP13]] monotonic, align 4
|
|
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK5: .omp.reduction.default:
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
|
|
// CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2:[0-9]+]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
|
|
// CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
|
|
// CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i64 0, i64 0
|
|
// CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
|
|
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 0
|
|
// CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
|
|
// CHECK5-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i64 0, i64 1
|
|
// CHECK5-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
|
|
// CHECK5-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 1
|
|
// CHECK5-NEXT: [[TMP11:%.*]] = load ptr, ptr [[TMP10]], align 8
|
|
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP7]], align 4
|
|
// CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP5]], align 4
|
|
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
|
|
// CHECK5-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
|
|
// CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP11]], align 4
|
|
// CHECK5-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP9]], align 4
|
|
// CHECK5-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
|
|
// CHECK5-NEXT: store i32 [[ADD2]], ptr [[TMP11]], align 4
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45
|
|
// CHECK5-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[X_CASTED:%.*]] = alloca i64, align 8
|
|
// CHECK5-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK5-NEXT: store i32 [[TMP1]], ptr [[X_CASTED]], align 4
|
|
// CHECK5-NEXT: [[TMP2:%.*]] = load i64, ptr [[X_CASTED]], align 8
|
|
// CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 1, ptr @.omp_outlined..3, i64 [[TMP2]])
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..3
|
|
// CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[X:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[X_ADDR:%.*]] = alloca i64, align 8
|
|
// CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store i64 [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51
|
|
// CHECK5-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[X_CASTED:%.*]] = alloca i64, align 8
|
|
// CHECK5-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK5-NEXT: store i32 [[TMP1]], ptr [[X_CASTED]], align 4
|
|
// CHECK5-NEXT: [[TMP2:%.*]] = load i64, ptr [[X_CASTED]], align 8
|
|
// CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 1, ptr @.omp_outlined..4, i64 [[TMP2]])
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..4
|
|
// CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[X:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[X_ADDR:%.*]] = alloca i64, align 8
|
|
// CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store i64 [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57
|
|
// CHECK5-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[X_CASTED:%.*]] = alloca i64, align 8
|
|
// CHECK5-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK5-NEXT: store i32 [[TMP1]], ptr [[X_CASTED]], align 4
|
|
// CHECK5-NEXT: [[TMP2:%.*]] = load i64, ptr [[X_CASTED]], align 8
|
|
// CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 1, ptr @.omp_outlined..5, i64 [[TMP2]])
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..5
|
|
// CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[X:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[X_ADDR:%.*]] = alloca i64, align 8
|
|
// CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store i64 [[X]], ptr [[X_ADDR]], align 8
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63
|
|
// CHECK5-SAME: (ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
|
|
// CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..6, ptr [[TMP0]], ptr [[TMP1]])
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..6
|
|
// CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Y1:%.*]] = alloca [88 x i32], align 4
|
|
// CHECK5-NEXT: [[X:%.*]] = alloca [77 x i32], align 4
|
|
// CHECK5-NEXT: [[Z2:%.*]] = alloca [99 x i32], align 4
|
|
// CHECK5-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
|
|
// CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i64 352, i1 false)
|
|
// CHECK5-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0
|
|
// CHECK5-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 99
|
|
// CHECK5-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
|
|
// CHECK5-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
|
|
// CHECK5: omp.arrayinit.body:
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
|
|
// CHECK5-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
|
|
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
|
|
// CHECK5: omp.arrayinit.done:
|
|
// CHECK5-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
|
|
// CHECK5-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8
|
|
// CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
|
|
// CHECK5-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.7, ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK5-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
|
|
// CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
|
|
// CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
|
|
// CHECK5-NEXT: ]
|
|
// CHECK5: .omp.reduction.case1:
|
|
// CHECK5-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP7]]
|
|
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
|
|
// CHECK5: omp.arraycpy.body:
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
|
|
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
|
|
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
|
|
// CHECK5-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP7]]
|
|
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
|
|
// CHECK5: omp.arraycpy.done6:
|
|
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK5: .omp.reduction.case2:
|
|
// CHECK5-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP10]]
|
|
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
|
|
// CHECK5: omp.arraycpy.body8:
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
|
|
// CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
|
|
// CHECK5-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP11]] monotonic, align 4
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP10]]
|
|
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
|
|
// CHECK5: omp.arraycpy.done14:
|
|
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK5: .omp.reduction.default:
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.7
|
|
// CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
|
|
// CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
|
|
// CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
|
|
// CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
|
|
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
|
|
// CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
|
|
// CHECK5-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i64 99
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
|
|
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
|
|
// CHECK5: omp.arraycpy.body:
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
|
|
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
|
|
// CHECK5-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
|
|
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
|
|
// CHECK5: omp.arraycpy.done2:
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65
|
|
// CHECK5-SAME: (ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
|
|
// CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..8, ptr [[TMP0]], ptr [[TMP1]])
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..8
|
|
// CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Y1:%.*]] = alloca [88 x i32], align 4
|
|
// CHECK5-NEXT: [[X:%.*]] = alloca [77 x i32], align 4
|
|
// CHECK5-NEXT: [[Z2:%.*]] = alloca [99 x i32], align 4
|
|
// CHECK5-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
|
|
// CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i64 352, i1 false)
|
|
// CHECK5-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0
|
|
// CHECK5-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 99
|
|
// CHECK5-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
|
|
// CHECK5-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
|
|
// CHECK5: omp.arrayinit.body:
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
|
|
// CHECK5-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
|
|
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
|
|
// CHECK5: omp.arrayinit.done:
|
|
// CHECK5-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
|
|
// CHECK5-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8
|
|
// CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
|
|
// CHECK5-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.9, ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK5-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
|
|
// CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
|
|
// CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
|
|
// CHECK5-NEXT: ]
|
|
// CHECK5: .omp.reduction.case1:
|
|
// CHECK5-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP7]]
|
|
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
|
|
// CHECK5: omp.arraycpy.body:
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
|
|
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
|
|
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
|
|
// CHECK5-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP7]]
|
|
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
|
|
// CHECK5: omp.arraycpy.done6:
|
|
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK5: .omp.reduction.case2:
|
|
// CHECK5-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP10]]
|
|
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
|
|
// CHECK5: omp.arraycpy.body8:
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
|
|
// CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
|
|
// CHECK5-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP11]] monotonic, align 4
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP10]]
|
|
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
|
|
// CHECK5: omp.arraycpy.done14:
|
|
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK5: .omp.reduction.default:
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.9
|
|
// CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
|
|
// CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
|
|
// CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
|
|
// CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
|
|
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
|
|
// CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
|
|
// CHECK5-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i64 99
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
|
|
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
|
|
// CHECK5: omp.arraycpy.body:
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
|
|
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
|
|
// CHECK5-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
|
|
// CHECK5-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
|
|
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
|
|
// CHECK5: omp.arraycpy.done2:
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l72
|
|
// CHECK5-SAME: (ptr noundef nonnull align 16 dereferenceable(16) [[Y:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Z:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
|
|
// CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..10, ptr [[TMP0]], ptr [[TMP1]])
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..10
|
|
// CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Y:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Z:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Y1:%.*]] = alloca i128, align 16
|
|
// CHECK5-NEXT: [[X:%.*]] = alloca i128, align 16
|
|
// CHECK5-NEXT: [[Z2:%.*]] = alloca i128, align 16
|
|
// CHECK5-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK5-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i128, align 16
|
|
// CHECK5-NEXT: [[ATOMIC_TEMP3:%.*]] = alloca i128, align 16
|
|
// CHECK5-NEXT: [[TMP:%.*]] = alloca i128, align 16
|
|
// CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP2:%.*]] = load i128, ptr [[TMP0]], align 16
|
|
// CHECK5-NEXT: store i128 [[TMP2]], ptr [[Y1]], align 16
|
|
// CHECK5-NEXT: store i128 0, ptr [[Z2]], align 16
|
|
// CHECK5-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
|
|
// CHECK5-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8
|
|
// CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
|
|
// CHECK5-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.11, ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK5-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
|
|
// CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
|
|
// CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
|
|
// CHECK5-NEXT: ]
|
|
// CHECK5: .omp.reduction.case1:
|
|
// CHECK5-NEXT: [[TMP7:%.*]] = load i128, ptr [[TMP1]], align 16
|
|
// CHECK5-NEXT: [[TMP8:%.*]] = load i128, ptr [[Z2]], align 16
|
|
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP7]], [[TMP8]]
|
|
// CHECK5-NEXT: store i128 [[ADD]], ptr [[TMP1]], align 16
|
|
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK5: .omp.reduction.case2:
|
|
// CHECK5-NEXT: [[TMP9:%.*]] = load i128, ptr [[Z2]], align 16
|
|
// CHECK5-NEXT: call void @__atomic_load(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], i32 noundef signext 0) #[[ATTR6:[0-9]+]]
|
|
// CHECK5-NEXT: br label [[ATOMIC_CONT:%.*]]
|
|
// CHECK5: atomic_cont:
|
|
// CHECK5-NEXT: [[TMP10:%.*]] = load i128, ptr [[ATOMIC_TEMP]], align 16
|
|
// CHECK5-NEXT: store i128 [[TMP10]], ptr [[TMP]], align 16
|
|
// CHECK5-NEXT: [[TMP11:%.*]] = load i128, ptr [[TMP]], align 16
|
|
// CHECK5-NEXT: [[TMP12:%.*]] = load i128, ptr [[Z2]], align 16
|
|
// CHECK5-NEXT: [[ADD4:%.*]] = add nsw i128 [[TMP11]], [[TMP12]]
|
|
// CHECK5-NEXT: store i128 [[ADD4]], ptr [[ATOMIC_TEMP3]], align 16
|
|
// CHECK5-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @__atomic_compare_exchange(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], ptr noundef [[ATOMIC_TEMP3]], i32 noundef signext 0, i32 noundef signext 0) #[[ATTR6]]
|
|
// CHECK5-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]]
|
|
// CHECK5: atomic_exit:
|
|
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK5: .omp.reduction.default:
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.11
|
|
// CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
|
|
// CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
|
|
// CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
|
|
// CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
|
|
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
|
|
// CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
|
|
// CHECK5-NEXT: [[TMP8:%.*]] = load i128, ptr [[TMP7]], align 16
|
|
// CHECK5-NEXT: [[TMP9:%.*]] = load i128, ptr [[TMP5]], align 16
|
|
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP8]], [[TMP9]]
|
|
// CHECK5-NEXT: store i128 [[ADD]], ptr [[TMP7]], align 16
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l74
|
|
// CHECK5-SAME: (ptr noundef nonnull align 16 dereferenceable(16) [[Y:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Z:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
|
|
// CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..12, ptr [[TMP0]], ptr [[TMP1]])
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..12
|
|
// CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Y:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Z:%.*]]) #[[ATTR0]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[Y1:%.*]] = alloca i128, align 16
|
|
// CHECK5-NEXT: [[X:%.*]] = alloca i128, align 16
|
|
// CHECK5-NEXT: [[Z2:%.*]] = alloca i128, align 16
|
|
// CHECK5-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
|
|
// CHECK5-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i128, align 16
|
|
// CHECK5-NEXT: [[ATOMIC_TEMP3:%.*]] = alloca i128, align 16
|
|
// CHECK5-NEXT: [[TMP:%.*]] = alloca i128, align 16
|
|
// CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP2:%.*]] = load i128, ptr [[TMP0]], align 16
|
|
// CHECK5-NEXT: store i128 [[TMP2]], ptr [[Y1]], align 16
|
|
// CHECK5-NEXT: store i128 0, ptr [[Z2]], align 16
|
|
// CHECK5-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
|
|
// CHECK5-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8
|
|
// CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
|
|
// CHECK5-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.13, ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK5-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
|
|
// CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
|
|
// CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
|
|
// CHECK5-NEXT: ]
|
|
// CHECK5: .omp.reduction.case1:
|
|
// CHECK5-NEXT: [[TMP7:%.*]] = load i128, ptr [[TMP1]], align 16
|
|
// CHECK5-NEXT: [[TMP8:%.*]] = load i128, ptr [[Z2]], align 16
|
|
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP7]], [[TMP8]]
|
|
// CHECK5-NEXT: store i128 [[ADD]], ptr [[TMP1]], align 16
|
|
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK5: .omp.reduction.case2:
|
|
// CHECK5-NEXT: [[TMP9:%.*]] = load i128, ptr [[Z2]], align 16
|
|
// CHECK5-NEXT: call void @__atomic_load(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], i32 noundef signext 0) #[[ATTR6]]
|
|
// CHECK5-NEXT: br label [[ATOMIC_CONT:%.*]]
|
|
// CHECK5: atomic_cont:
|
|
// CHECK5-NEXT: [[TMP10:%.*]] = load i128, ptr [[ATOMIC_TEMP]], align 16
|
|
// CHECK5-NEXT: store i128 [[TMP10]], ptr [[TMP]], align 16
|
|
// CHECK5-NEXT: [[TMP11:%.*]] = load i128, ptr [[TMP]], align 16
|
|
// CHECK5-NEXT: [[TMP12:%.*]] = load i128, ptr [[Z2]], align 16
|
|
// CHECK5-NEXT: [[ADD4:%.*]] = add nsw i128 [[TMP11]], [[TMP12]]
|
|
// CHECK5-NEXT: store i128 [[ADD4]], ptr [[ATOMIC_TEMP3]], align 16
|
|
// CHECK5-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @__atomic_compare_exchange(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], ptr noundef [[ATOMIC_TEMP3]], i32 noundef signext 0, i32 noundef signext 0) #[[ATTR6]]
|
|
// CHECK5-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]]
|
|
// CHECK5: atomic_exit:
|
|
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK5: .omp.reduction.default:
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.13
|
|
// CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] {
|
|
// CHECK5-NEXT: entry:
|
|
// CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
|
|
// CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
|
|
// CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
|
|
// CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
|
|
// CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
|
|
// CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
|
|
// CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
|
|
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
|
|
// CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
|
|
// CHECK5-NEXT: [[TMP8:%.*]] = load i128, ptr [[TMP7]], align 16
|
|
// CHECK5-NEXT: [[TMP9:%.*]] = load i128, ptr [[TMP5]], align 16
|
|
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP8]], [[TMP9]]
|
|
// CHECK5-NEXT: store i128 [[ADD]], ptr [[TMP7]], align 16
|
|
// CHECK5-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27
|
|
// CHECK7-SAME: () #[[ATTR0:[0-9]+]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1:[0-9]+]], i32 0, ptr @.omp_outlined.)
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@.omp_outlined.
|
|
// CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[X:%.*]] = alloca i32, align 4
|
|
// CHECK7-NEXT: [[Y:%.*]] = alloca i32, align 4
|
|
// CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33
|
|
// CHECK7-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[Y:%.*]]) #[[ATTR0]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[X_CASTED:%.*]] = alloca i32, align 4
|
|
// CHECK7-NEXT: [[Y_CASTED:%.*]] = alloca i32, align 4
|
|
// CHECK7-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK7-NEXT: store i32 [[TMP2]], ptr [[X_CASTED]], align 4
|
|
// CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[X_CASTED]], align 4
|
|
// CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4
|
|
// CHECK7-NEXT: store i32 [[TMP4]], ptr [[Y_CASTED]], align 4
|
|
// CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[Y_CASTED]], align 4
|
|
// CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..1, i32 [[TMP3]], i32 [[TMP5]])
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..1
|
|
// CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR0]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[X_ADDR:%.*]] = alloca i32, align 4
|
|
// CHECK7-NEXT: [[Y_ADDR:%.*]] = alloca i32, align 4
|
|
// CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: store i32 [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK7-NEXT: store i32 [[Y]], ptr [[Y_ADDR]], align 4
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39
|
|
// CHECK7-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[Y:%.*]]) #[[ATTR0]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
|
|
// CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..2, ptr [[TMP0]], ptr [[TMP1]])
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..2
|
|
// CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[Y:%.*]]) #[[ATTR0]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[X1:%.*]] = alloca i32, align 4
|
|
// CHECK7-NEXT: [[Y2:%.*]] = alloca i32, align 4
|
|
// CHECK7-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [2 x ptr], align 4
|
|
// CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
|
|
// CHECK7-NEXT: store i32 0, ptr [[X1]], align 4
|
|
// CHECK7-NEXT: store i32 0, ptr [[Y2]], align 4
|
|
// CHECK7-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
|
|
// CHECK7-NEXT: store ptr [[X1]], ptr [[TMP2]], align 4
|
|
// CHECK7-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 1
|
|
// CHECK7-NEXT: store ptr [[Y2]], ptr [[TMP3]], align 4
|
|
// CHECK7-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
|
|
// CHECK7-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP5]], i32 2, i32 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK7-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
|
|
// CHECK7-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
|
|
// CHECK7-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
|
|
// CHECK7-NEXT: ]
|
|
// CHECK7: .omp.reduction.case1:
|
|
// CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[X1]], align 4
|
|
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
|
|
// CHECK7-NEXT: store i32 [[ADD]], ptr [[TMP0]], align 4
|
|
// CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP1]], align 4
|
|
// CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[Y2]], align 4
|
|
// CHECK7-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
|
|
// CHECK7-NEXT: store i32 [[ADD3]], ptr [[TMP1]], align 4
|
|
// CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK7: .omp.reduction.case2:
|
|
// CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[X1]], align 4
|
|
// CHECK7-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP11]] monotonic, align 4
|
|
// CHECK7-NEXT: [[TMP13:%.*]] = load i32, ptr [[Y2]], align 4
|
|
// CHECK7-NEXT: [[TMP14:%.*]] = atomicrmw add ptr [[TMP1]], i32 [[TMP13]] monotonic, align 4
|
|
// CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK7: .omp.reduction.default:
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
|
|
// CHECK7-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2:[0-9]+]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
|
|
// CHECK7-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
|
|
// CHECK7-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i32 0, i32 0
|
|
// CHECK7-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
|
|
// CHECK7-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i32 0, i32 0
|
|
// CHECK7-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
|
|
// CHECK7-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i32 0, i32 1
|
|
// CHECK7-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 4
|
|
// CHECK7-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i32 0, i32 1
|
|
// CHECK7-NEXT: [[TMP11:%.*]] = load ptr, ptr [[TMP10]], align 4
|
|
// CHECK7-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP7]], align 4
|
|
// CHECK7-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP5]], align 4
|
|
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
|
|
// CHECK7-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
|
|
// CHECK7-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP11]], align 4
|
|
// CHECK7-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP9]], align 4
|
|
// CHECK7-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
|
|
// CHECK7-NEXT: store i32 [[ADD2]], ptr [[TMP11]], align 4
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45
|
|
// CHECK7-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]]) #[[ATTR0]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[X_CASTED:%.*]] = alloca i32, align 4
|
|
// CHECK7-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK7-NEXT: store i32 [[TMP1]], ptr [[X_CASTED]], align 4
|
|
// CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[X_CASTED]], align 4
|
|
// CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 1, ptr @.omp_outlined..3, i32 [[TMP2]])
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..3
|
|
// CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[X:%.*]]) #[[ATTR0]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[X_ADDR:%.*]] = alloca i32, align 4
|
|
// CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: store i32 [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51
|
|
// CHECK7-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]]) #[[ATTR0]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[X_CASTED:%.*]] = alloca i32, align 4
|
|
// CHECK7-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK7-NEXT: store i32 [[TMP1]], ptr [[X_CASTED]], align 4
|
|
// CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[X_CASTED]], align 4
|
|
// CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 1, ptr @.omp_outlined..4, i32 [[TMP2]])
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..4
|
|
// CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[X:%.*]]) #[[ATTR0]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[X_ADDR:%.*]] = alloca i32, align 4
|
|
// CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: store i32 [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57
|
|
// CHECK7-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]]) #[[ATTR0]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[X_CASTED:%.*]] = alloca i32, align 4
|
|
// CHECK7-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
|
|
// CHECK7-NEXT: store i32 [[TMP1]], ptr [[X_CASTED]], align 4
|
|
// CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[X_CASTED]], align 4
|
|
// CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 1, ptr @.omp_outlined..5, i32 [[TMP2]])
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..5
|
|
// CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[X:%.*]]) #[[ATTR0]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[X_ADDR:%.*]] = alloca i32, align 4
|
|
// CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: store i32 [[X]], ptr [[X_ADDR]], align 4
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63
|
|
// CHECK7-SAME: (ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR0]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 4
|
|
// CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..6, ptr [[TMP0]], ptr [[TMP1]])
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..6
|
|
// CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR0]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[Y1:%.*]] = alloca [88 x i32], align 4
|
|
// CHECK7-NEXT: [[X:%.*]] = alloca [77 x i32], align 4
|
|
// CHECK7-NEXT: [[Z2:%.*]] = alloca [99 x i32], align 4
|
|
// CHECK7-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 4
|
|
// CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 4
|
|
// CHECK7-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i32 352, i1 false)
|
|
// CHECK7-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0
|
|
// CHECK7-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i32 99
|
|
// CHECK7-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
|
|
// CHECK7-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
|
|
// CHECK7: omp.arrayinit.body:
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
|
|
// CHECK7-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
|
|
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
|
|
// CHECK7: omp.arrayinit.done:
|
|
// CHECK7-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
|
|
// CHECK7-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 4
|
|
// CHECK7-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
|
|
// CHECK7-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.7, ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK7-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
|
|
// CHECK7-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
|
|
// CHECK7-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
|
|
// CHECK7-NEXT: ]
|
|
// CHECK7: .omp.reduction.case1:
|
|
// CHECK7-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP7]]
|
|
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
|
|
// CHECK7: omp.arraycpy.body:
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
|
|
// CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
|
|
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
|
|
// CHECK7-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP7]]
|
|
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
|
|
// CHECK7: omp.arraycpy.done6:
|
|
// CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK7: .omp.reduction.case2:
|
|
// CHECK7-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP10]]
|
|
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
|
|
// CHECK7: omp.arraycpy.body8:
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
|
|
// CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
|
|
// CHECK7-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP11]] monotonic, align 4
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP10]]
|
|
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
|
|
// CHECK7: omp.arraycpy.done14:
|
|
// CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK7: .omp.reduction.default:
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.7
|
|
// CHECK7-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
|
|
// CHECK7-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
|
|
// CHECK7-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
|
|
// CHECK7-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
|
|
// CHECK7-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
|
|
// CHECK7-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
|
|
// CHECK7-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i32 99
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
|
|
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
|
|
// CHECK7: omp.arraycpy.body:
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
|
|
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
|
|
// CHECK7-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
|
|
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
|
|
// CHECK7: omp.arraycpy.done2:
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65
|
|
// CHECK7-SAME: (ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR0]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 4
|
|
// CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 2, ptr @.omp_outlined..8, ptr [[TMP0]], ptr [[TMP1]])
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..8
|
|
// CHECK7-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR0]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[Y1:%.*]] = alloca [88 x i32], align 4
|
|
// CHECK7-NEXT: [[X:%.*]] = alloca [77 x i32], align 4
|
|
// CHECK7-NEXT: [[Z2:%.*]] = alloca [99 x i32], align 4
|
|
// CHECK7-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 4
|
|
// CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 4
|
|
// CHECK7-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i32 352, i1 false)
|
|
// CHECK7-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0
|
|
// CHECK7-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i32 99
|
|
// CHECK7-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
|
|
// CHECK7-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
|
|
// CHECK7: omp.arrayinit.body:
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
|
|
// CHECK7-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
|
|
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
|
|
// CHECK7: omp.arrayinit.done:
|
|
// CHECK7-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
|
|
// CHECK7-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 4
|
|
// CHECK7-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
|
|
// CHECK7-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.9, ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK7-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
|
|
// CHECK7-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
|
|
// CHECK7-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
|
|
// CHECK7-NEXT: ]
|
|
// CHECK7: .omp.reduction.case1:
|
|
// CHECK7-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP7]]
|
|
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
|
|
// CHECK7: omp.arraycpy.body:
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
|
|
// CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
|
|
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
|
|
// CHECK7-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP7]]
|
|
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
|
|
// CHECK7: omp.arraycpy.done6:
|
|
// CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK7: .omp.reduction.case2:
|
|
// CHECK7-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP10]]
|
|
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
|
|
// CHECK7: omp.arraycpy.body8:
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
|
|
// CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
|
|
// CHECK7-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP11]] monotonic, align 4
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP10]]
|
|
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
|
|
// CHECK7: omp.arraycpy.done14:
|
|
// CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
|
|
// CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
|
|
// CHECK7: .omp.reduction.default:
|
|
// CHECK7-NEXT: ret void
|
|
//
|
|
//
|
|
// CHECK7-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.9
|
|
// CHECK7-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] {
|
|
// CHECK7-NEXT: entry:
|
|
// CHECK7-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
|
|
// CHECK7-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
|
|
// CHECK7-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
|
|
// CHECK7-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
|
|
// CHECK7-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
|
|
// CHECK7-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
|
|
// CHECK7-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
|
|
// CHECK7-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
|
|
// CHECK7-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
|
|
// CHECK7-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i32 99
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
|
|
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
|
|
// CHECK7: omp.arraycpy.body:
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
|
|
// CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
|
|
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
|
|
// CHECK7-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
|
|
// CHECK7-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
|
|
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
|
|
// CHECK7: omp.arraycpy.done2:
|
|
// CHECK7-NEXT: ret void
|
|
//
|