fa5d22a045
Replace the `ident_t` handling in Clang with the methods offered by the OMPIRBuilder. This cuts down on the clang code as well as the differences between the two, making further transitions easier. Tests have changed but there should not be a real functional change. The most interesting difference is probably that we stop generating local ident_t allocations for now and just use globals. Given that this happens only with debug info, the location part of the `ident_t` is probably bigger than the test anyway. As the location part is already a global, we can avoid the allocation, memcpy, and store in favor of a constant global that is slightly bigger. This can be revisited if there are complications. Reviewed By: ABataev Differential Revision: https://reviews.llvm.org/D80735
104 lines
5.3 KiB
C++
104 lines
5.3 KiB
C++
// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -x c++ -emit-llvm %s -o - | FileCheck %s
|
|
// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-apple-darwin10 -emit-pch -o %t %s
|
|
// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-apple-darwin10 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
|
|
|
|
// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp-simd -x c++ -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s
|
|
// RUN: %clang_cc1 -fopenmp-simd -x c++ -triple x86_64-apple-darwin10 -emit-pch -o %t %s
|
|
// RUN: %clang_cc1 -fopenmp-simd -x c++ -triple x86_64-apple-darwin10 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
|
|
// SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
|
|
// expected-no-diagnostics
|
|
#ifndef HEADER
|
|
#define HEADER
|
|
|
|
typedef void **omp_allocator_handle_t;
|
|
extern const omp_allocator_handle_t omp_null_allocator;
|
|
extern const omp_allocator_handle_t omp_default_mem_alloc;
|
|
extern const omp_allocator_handle_t omp_large_cap_mem_alloc;
|
|
extern const omp_allocator_handle_t omp_const_mem_alloc;
|
|
extern const omp_allocator_handle_t omp_high_bw_mem_alloc;
|
|
extern const omp_allocator_handle_t omp_low_lat_mem_alloc;
|
|
extern const omp_allocator_handle_t omp_cgroup_mem_alloc;
|
|
extern const omp_allocator_handle_t omp_pteam_mem_alloc;
|
|
extern const omp_allocator_handle_t omp_thread_mem_alloc;
|
|
|
|
// CHECK: [[PRIVATES:%.+]] = type { i8*, i8* }
|
|
|
|
struct S {
|
|
int a;
|
|
S() : a(0) {}
|
|
S(const S&) {}
|
|
S& operator=(const S&) {return *this;}
|
|
~S() {}
|
|
friend S operator+(const S&a, const S&b) {return a;}
|
|
};
|
|
|
|
|
|
int main(int argc, char **argv) {
|
|
int a;
|
|
float b;
|
|
S c[5];
|
|
short d[argc];
|
|
#pragma omp taskgroup task_reduction(+: a, b, argc)
|
|
{
|
|
#pragma omp taskgroup task_reduction(-:c, d)
|
|
#pragma omp parallel
|
|
#pragma omp task in_reduction(+:a) in_reduction(-:d) allocate(omp_high_bw_mem_alloc: d)
|
|
a += d[a];
|
|
}
|
|
#pragma omp task in_reduction(+:a)
|
|
++a;
|
|
return 0;
|
|
}
|
|
|
|
// CHECK-LABEL: @main
|
|
// CHECK: void @__kmpc_taskgroup(%struct.ident_t* @1, i32 [[GTID:%.+]])
|
|
// CHECK: [[TD1:%.+]] = call i8* @__kmpc_taskred_init(i32 [[GTID]], i32 3, i8* %
|
|
// CHECK-NEXT: store i8* [[TD1]], i8** [[TD1_ADDR:%[^,]+]],
|
|
// CHECK-NEXT: call void @__kmpc_taskgroup(%struct.ident_t* @1, i32 [[GTID]])
|
|
// CHECK: [[TD2:%.+]] = call i8* @__kmpc_taskred_init(i32 [[GTID]], i32 2, i8* %
|
|
// CHECK-NEXT: store i8* [[TD2]], i8** [[TD2_ADDR:%[^,]+]],
|
|
// CHECK-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @1, i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i16*, i8**, i8**)* [[OMP_PARALLEL:@.+]] to void (i32*, i32*, ...)*), i32* %{{.+}}, i64 %{{.+}}, i16* %{{.+}}, i8** [[TD1_ADDR]], i8** [[TD2_ADDR]])
|
|
// CHECK-NEXT: call void @__kmpc_end_taskgroup(%struct.ident_t* @1, i32 [[GTID]])
|
|
// CHECK-NEXT: call void @__kmpc_end_taskgroup(%struct.ident_t* @1, i32 [[GTID]])
|
|
|
|
// CHECK: define internal void [[OMP_PARALLEL]](
|
|
// CHECK: [[TASK_T:%.+]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @1, i32 [[GTID:%.+]], i32 1, i64 56, i64 40, i32 (i32, i8*)* bitcast (i32 (i32, [[T:%.+]]*)* [[OMP_TASK:@.+]] to i32 (i32, i8*)*))
|
|
// CHECK-NEXT: [[TASK_T_WITH_PRIVS:%.+]] = bitcast i8* [[TASK_T]] to [[T]]*
|
|
// CHECK: [[PRIVS:%.+]] = getelementptr inbounds [[T]], [[T]]* [[TASK_T_WITH_PRIVS]], i32 0, i32 1
|
|
// CHECK: [[TD1_REF:%.+]] = getelementptr inbounds [[PRIVATES]], [[PRIVATES]]* [[PRIVS]], i32 0, i32 0
|
|
// CHECK-NEXT: [[TD1:%.+]] = load i8*, i8** %{{.+}},
|
|
// CHECK-NEXT: store i8* [[TD1]], i8** [[TD1_REF]],
|
|
// CHECK-NEXT: [[TD2_REF:%.+]] = getelementptr inbounds [[PRIVATES]], [[PRIVATES]]* [[PRIVS]], i32 0, i32 1
|
|
// CHECK-NEXT: [[TD2:%.+]] = load i8*, i8** %{{.+}},
|
|
// CHECK-NEXT: store i8* [[TD2]], i8** [[TD2_REF]],
|
|
// CHECK-NEXT: call i32 @__kmpc_omp_task(%struct.ident_t* @1, i32 [[GTID]], i8* [[TASK_T]])
|
|
// CHECK-NEXT: ret void
|
|
// CHECK-NEXT: }
|
|
|
|
// CHECK: define internal {{.*}} [[OMP_TASK]](
|
|
// CHECK: call void (i8*, ...) %{{[^(]+}}(i8* %{{.+}}, i8*** [[TD1_REF:%[^,]+]], i8*** [[TD2_REF:%[^,]+]])
|
|
// CHECK-NEXT: [[TD1_ADDR:%.+]] = load i8**, i8*** [[TD1_REF]],
|
|
// CHECK-NEXT: [[TD2_ADDR:%.+]] = load i8**, i8*** [[TD2_REF]],
|
|
// CHECK-NEXT: [[A_REF:%.+]] = getelementptr inbounds %
|
|
// CHECK-NEXT: [[A_ADDR:%.+]] = load i32*, i32** [[A_REF]],
|
|
// CHECK-NEXT: [[TD1:%.+]] = load i8*, i8** [[TD1_ADDR]],
|
|
// CHECK-NEXT: [[GTID:%.+]] = load i32, i32* %
|
|
// CHECK-NEXT: [[A_PTR:%.+]] = bitcast i32* [[A_ADDR]] to i8*
|
|
// CHECK-NEXT: call i8* @__kmpc_task_reduction_get_th_data(i32 [[GTID]], i8* [[TD1]], i8* [[A_PTR]])
|
|
// CHECK: [[D_REF:%.+]] = getelementptr inbounds %
|
|
// CHECK-NEXT: [[D_ADDR:%.+]] = load i16*, i16** [[D_REF]],
|
|
// CHECK: call i8* @__kmpc_threadprivate_cached(
|
|
// CHECK: [[TD2:%.+]] = load i8*, i8** [[TD2_ADDR]],
|
|
// CHECK-NEXT: [[D_PTR:%.+]] = bitcast i16* [[D_ADDR]] to i8*
|
|
// CHECK-NEXT: call i8* @__kmpc_task_reduction_get_th_data(i32 [[GTID]], i8* [[TD2]], i8* [[D_PTR]])
|
|
// CHECK: add nsw i32
|
|
// CHECK: store i32 %
|
|
// CHECK-NOT: call i8* @__kmpc_threadprivate_cached(
|
|
|
|
// CHECK: [[A_PTR:%.+]] = call i8* @__kmpc_task_reduction_get_th_data(i32 %{{.+}}, i8* null, i8* %{{.+}})
|
|
// CHECK-NEXT: [[A_ADDR:%.+]] = bitcast i8* [[A_PTR]] to i32*
|
|
// CHECK-NEXT: [[A:%.+]] = load i32, i32* [[A_ADDR]],
|
|
// CHECK-NEXT: [[NEW:%.+]] = add nsw i32 [[A]], 1
|
|
// CHECK-NEXT: store i32 [[NEW]], i32* [[A_ADDR]],
|
|
#endif
|