1fff116645
This patch changes the code we generate to enter a target region on the device. This is in-line with the new definition in the runtime that was added previously. Additionally we implement this in the OpenMPIRBuilder so that this code can be shared with Flang in the future. Reviewed By: ABataev Differential Revision: https://reviews.llvm.org/D128550
120 lines
9.6 KiB
C++
120 lines
9.6 KiB
C++
// expected-no-diagnostics
|
|
#ifndef HEADER
|
|
#define HEADER
|
|
|
|
|
|
|
|
///==========================================================================///
|
|
// RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CK2 --check-prefix CK2-64
|
|
// RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
|
|
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CK2 --check-prefix CK2-64
|
|
// RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CK2 --check-prefix CK2-32
|
|
// RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
|
|
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CK2 --check-prefix CK2-32
|
|
|
|
// RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CK2 --check-prefix CK2-64
|
|
// RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
|
|
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CK2 --check-prefix CK2-64
|
|
// RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CK2 --check-prefix CK2-32
|
|
// RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
|
|
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CK2 --check-prefix CK2-32
|
|
|
|
// RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CK2 --check-prefix CK2-64
|
|
// RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
|
|
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CK2 --check-prefix CK2-64
|
|
// RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp -fopenmp-version=45 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CK2 --check-prefix CK2-32
|
|
// RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -fopenmp-version=45 -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
|
|
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=45 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CK2 --check-prefix CK2-32
|
|
|
|
// RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY1 %s
|
|
// RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
|
|
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY1 %s
|
|
// RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY1 %s
|
|
// RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
|
|
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY1 %s
|
|
// SIMD-ONLY1-NOT: {{__kmpc|__tgt}}
|
|
#ifdef CK2
|
|
|
|
// CK2-LABEL: @.__omp_offloading_{{.*}}implicit_maps_reference{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
|
|
|
|
// CK2: [[SIZES:@.+]] = {{.+}}constant [1 x i64] [i64 4]
|
|
// Map types: OMP_MAP_PRIVATE_VAL | OMP_MAP_TARGET_PARAM | OMP_MAP_IMPLICIT = 800
|
|
// CK2: [[TYPES:@.+]] = {{.+}}constant [1 x i64] [i64 800]
|
|
// CK2-LABEL: @.__omp_offloading_{{.*}}implicit_maps_reference{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
|
|
// CK2: [[SIZES2:@.+]] = {{.+}}constant [1 x i64] zeroinitializer
|
|
// Map types: OMP_MAP_IS_PTR | OMP_MAP_IMPLICIT = 544
|
|
// CK2: [[TYPES2:@.+]] = {{.+}}constant [1 x i64] [i64 544]
|
|
|
|
// CK2-LABEL: implicit_maps_reference{{.*}}(
|
|
void implicit_maps_reference (int a, int *b){
|
|
int &i = a;
|
|
// CK2-DAG: call i32 @__tgt_target_kernel(%struct.ident_t* @{{.+}}, i64 -1, i32 -1, i32 0, i8* @.{{.+}}.region_id, %struct.__tgt_kernel_arguments* [[ARGS:%.+]])
|
|
// CK2-DAG: [[BPARG:%.+]] = getelementptr inbounds {{.+}}[[ARGS]], i32 0, i32 2
|
|
// CK2-DAG: store i8** [[BPGEP:%.+]], i8*** [[BPARG]]
|
|
// CK2-DAG: [[PGEP:%.+]] = getelementptr inbounds {{.+}}[[ARGS]], i32 0, i32 3
|
|
// CK2-DAG: store i8** [[PGEP:%.+]], i8*** [[BPARG]]
|
|
// CK2-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
|
|
// CK2-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
|
|
// CK2-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
|
|
// CK2-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
|
|
// CK2-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i[[sz:64|32]]*
|
|
// CK2-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i[[sz]]*
|
|
// CK2-DAG: store i[[sz]] [[VAL:%[^,]+]], i[[sz]]* [[CBP1]]
|
|
// CK2-DAG: store i[[sz]] [[VAL]], i[[sz]]* [[CP1]]
|
|
// CK2-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
|
|
// CK2-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
|
|
// CK2-64-DAG: store i32 {{.+}}, i32* [[CADDR]],
|
|
|
|
// CK2: call void [[KERNEL:@.+]](i[[sz]] [[VAL]])
|
|
#pragma omp target
|
|
{
|
|
++i;
|
|
}
|
|
|
|
int *&p = b;
|
|
// CK2-DAG: call i32 @__tgt_target_kernel(%struct.ident_t* @{{.+}}, i64 -1, i32 -1, i32 0, i8* @.{{.+}}.region_id, %struct.__tgt_kernel_arguments* [[ARGS:%.+]])
|
|
// CK2-DAG: [[BPARG:%.+]] = getelementptr inbounds {{.+}}[[ARGS]], i32 0, i32 2
|
|
// CK2-DAG: store i8** [[BPGEP:%.+]], i8*** [[BPARG]]
|
|
// CK2-DAG: [[PGEP:%.+]] = getelementptr inbounds {{.+}}[[ARGS]], i32 0, i32 3
|
|
// CK2-DAG: store i8** [[PGEP:%.+]], i8*** [[BPARG]]
|
|
// CK2-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
|
|
// CK2-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
|
|
// CK2-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
|
|
// CK2-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
|
|
// CK2-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i32**
|
|
// CK2-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32**
|
|
// CK2-DAG: store i32* [[VAL:%[^,]+]], i32** [[CBP1]]
|
|
// CK2-DAG: store i32* [[VAL]], i32** [[CP1]]
|
|
// CK2-DAG: [[VAL]] = load i32*, i32** [[ADDR:%.+]],
|
|
// CK2-DAG: [[ADDR]] = load i32**, i32*** [[ADDR2:%.+]],
|
|
|
|
// CK2: call void [[KERNEL2:@.+]](i32* [[VAL]])
|
|
#pragma omp target
|
|
{
|
|
++p;
|
|
}
|
|
}
|
|
|
|
// CK2: define internal void [[KERNEL]](i[[sz]] noundef [[ARG:%.+]])
|
|
// CK2: [[ADDR:%.+]] = alloca i[[sz]],
|
|
// CK2: [[REF:%.+]] = alloca i32*,
|
|
// CK2: store i[[sz]] [[ARG]], i[[sz]]* [[ADDR]],
|
|
// CK2-64: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
|
|
// CK2-64: store i32* [[CADDR]], i32** [[REF]],
|
|
// CK2-64: [[RVAL:%.+]] = load i32*, i32** [[REF]],
|
|
// CK2-64: {{.+}} = load i32, i32* [[RVAL]],
|
|
// CK2-32: store i32* [[ADDR]], i32** [[REF]],
|
|
// CK2-32: [[RVAL:%.+]] = load i32*, i32** [[REF]],
|
|
// CK2-32: {{.+}} = load i32, i32* [[RVAL]],
|
|
|
|
// CK2: define internal void [[KERNEL2]](i32* noundef [[ARG:%.+]])
|
|
// CK2: [[ADDR:%.+]] = alloca i32*,
|
|
// CK2: [[REF:%.+]] = alloca i32**,
|
|
// CK2: store i32* [[ARG]], i32** [[ADDR]],
|
|
// CK2: store i32** [[ADDR]], i32*** [[REF]],
|
|
// CK2: [[T:%.+]] = load i32**, i32*** [[REF]],
|
|
// CK2: [[TT:%.+]] = load i32*, i32** [[T]],
|
|
// CK2: getelementptr inbounds i32, i32* [[TT]], i32 1
|
|
#endif // CK2
|
|
#endif
|