1c9ec74e3f
If an inlined kernel is called in a loop, the launch point alloca would lead to increasing stack usage every time the kernel is invoked. This could make the application run out of stack space and crash. This problem is fixed by using the alloca insertion point while creating the alloca instruction. Fixes https://github.com/llvm/llvm-project/issues/60602 Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D145820
172 lines
12 KiB
C++
172 lines
12 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]+"
|
|
// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -fopenmp-offload-mandatory -emit-llvm %s -o - | FileCheck %s --check-prefix=MANDATORY
|
|
// expected-no-diagnostics
|
|
|
|
void foo() {}
|
|
#pragma omp declare target(foo)
|
|
|
|
void bar() {}
|
|
#pragma omp declare target device_type(nohost) to(bar)
|
|
|
|
void host() {
|
|
#pragma omp target
|
|
{ bar(); }
|
|
}
|
|
|
|
void host_if(bool cond) {
|
|
#pragma omp target if(cond)
|
|
{ bar(); }
|
|
}
|
|
|
|
void host_dev(int device) {
|
|
#pragma omp target device(device)
|
|
{ bar(); }
|
|
}
|
|
// MANDATORY-LABEL: define {{[^@]+}}@_Z3foov
|
|
// MANDATORY-SAME: () #[[ATTR0:[0-9]+]] {
|
|
// MANDATORY-NEXT: entry:
|
|
// MANDATORY-NEXT: ret void
|
|
//
|
|
//
|
|
// MANDATORY-LABEL: define {{[^@]+}}@_Z4hostv
|
|
// MANDATORY-SAME: () #[[ATTR0]] {
|
|
// MANDATORY-NEXT: entry:
|
|
// MANDATORY-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// MANDATORY-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// MANDATORY-NEXT: store i32 2, ptr [[TMP0]], align 4
|
|
// MANDATORY-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// MANDATORY-NEXT: store i32 0, ptr [[TMP1]], align 4
|
|
// MANDATORY-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP2]], align 8
|
|
// MANDATORY-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP3]], align 8
|
|
// MANDATORY-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP4]], align 8
|
|
// MANDATORY-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP5]], align 8
|
|
// MANDATORY-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP6]], align 8
|
|
// MANDATORY-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP7]], align 8
|
|
// MANDATORY-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// MANDATORY-NEXT: store i64 0, ptr [[TMP8]], align 8
|
|
// MANDATORY-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// MANDATORY-NEXT: store i64 0, ptr [[TMP9]], align 8
|
|
// MANDATORY-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// MANDATORY-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP10]], align 4
|
|
// MANDATORY-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// MANDATORY-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP11]], align 4
|
|
// MANDATORY-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// MANDATORY-NEXT: store i32 0, ptr [[TMP12]], align 4
|
|
// MANDATORY-NEXT: [[TMP13:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4hostv_l12.region_id, ptr [[KERNEL_ARGS]])
|
|
// MANDATORY-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
|
|
// MANDATORY-NEXT: br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// MANDATORY: omp_offload.failed:
|
|
// MANDATORY-NEXT: unreachable
|
|
// MANDATORY: omp_offload.cont:
|
|
// MANDATORY-NEXT: ret void
|
|
//
|
|
//
|
|
// MANDATORY-LABEL: define {{[^@]+}}@_Z7host_ifb
|
|
// MANDATORY-SAME: (i1 noundef zeroext [[COND:%.*]]) #[[ATTR0]] {
|
|
// MANDATORY-NEXT: entry:
|
|
// MANDATORY-NEXT: [[COND_ADDR:%.*]] = alloca i8, align 1
|
|
// MANDATORY-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// MANDATORY-NEXT: [[FROMBOOL:%.*]] = zext i1 [[COND]] to i8
|
|
// MANDATORY-NEXT: store i8 [[FROMBOOL]], ptr [[COND_ADDR]], align 1
|
|
// MANDATORY-NEXT: [[TMP0:%.*]] = load i8, ptr [[COND_ADDR]], align 1
|
|
// MANDATORY-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP0]] to i1
|
|
// MANDATORY-NEXT: br i1 [[TOBOOL]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
|
|
// MANDATORY: omp_if.then:
|
|
// MANDATORY-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// MANDATORY-NEXT: store i32 2, ptr [[TMP1]], align 4
|
|
// MANDATORY-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// MANDATORY-NEXT: store i32 0, ptr [[TMP2]], align 4
|
|
// MANDATORY-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP3]], align 8
|
|
// MANDATORY-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP4]], align 8
|
|
// MANDATORY-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP5]], align 8
|
|
// MANDATORY-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP6]], align 8
|
|
// MANDATORY-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP7]], align 8
|
|
// MANDATORY-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP8]], align 8
|
|
// MANDATORY-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// MANDATORY-NEXT: store i64 0, ptr [[TMP9]], align 8
|
|
// MANDATORY-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// MANDATORY-NEXT: store i64 0, ptr [[TMP10]], align 8
|
|
// MANDATORY-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// MANDATORY-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP11]], align 4
|
|
// MANDATORY-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// MANDATORY-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP12]], align 4
|
|
// MANDATORY-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// MANDATORY-NEXT: store i32 0, ptr [[TMP13]], align 4
|
|
// MANDATORY-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7host_ifb_l17.region_id, ptr [[KERNEL_ARGS]])
|
|
// MANDATORY-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
|
|
// MANDATORY-NEXT: br i1 [[TMP15]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// MANDATORY: omp_offload.failed:
|
|
// MANDATORY-NEXT: unreachable
|
|
// MANDATORY: omp_offload.cont:
|
|
// MANDATORY-NEXT: br label [[OMP_IF_END:%.*]]
|
|
// MANDATORY: omp_if.else:
|
|
// MANDATORY-NEXT: unreachable
|
|
// MANDATORY: omp_if.end:
|
|
// MANDATORY-NEXT: ret void
|
|
//
|
|
//
|
|
// MANDATORY-LABEL: define {{[^@]+}}@_Z8host_devi
|
|
// MANDATORY-SAME: (i32 noundef signext [[DEVICE:%.*]]) #[[ATTR0]] {
|
|
// MANDATORY-NEXT: entry:
|
|
// MANDATORY-NEXT: [[DEVICE_ADDR:%.*]] = alloca i32, align 4
|
|
// MANDATORY-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
|
|
// MANDATORY-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
|
|
// MANDATORY-NEXT: store i32 [[DEVICE]], ptr [[DEVICE_ADDR]], align 4
|
|
// MANDATORY-NEXT: [[TMP0:%.*]] = load i32, ptr [[DEVICE_ADDR]], align 4
|
|
// MANDATORY-NEXT: store i32 [[TMP0]], ptr [[DOTCAPTURE_EXPR_]], align 4
|
|
// MANDATORY-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
|
|
// MANDATORY-NEXT: [[TMP2:%.*]] = sext i32 [[TMP1]] to i64
|
|
// MANDATORY-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
|
|
// MANDATORY-NEXT: store i32 2, ptr [[TMP3]], align 4
|
|
// MANDATORY-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
|
|
// MANDATORY-NEXT: store i32 0, ptr [[TMP4]], align 4
|
|
// MANDATORY-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP5]], align 8
|
|
// MANDATORY-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP6]], align 8
|
|
// MANDATORY-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP7]], align 8
|
|
// MANDATORY-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP8]], align 8
|
|
// MANDATORY-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP9]], align 8
|
|
// MANDATORY-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
|
|
// MANDATORY-NEXT: store ptr null, ptr [[TMP10]], align 8
|
|
// MANDATORY-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
|
|
// MANDATORY-NEXT: store i64 0, ptr [[TMP11]], align 8
|
|
// MANDATORY-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
|
|
// MANDATORY-NEXT: store i64 0, ptr [[TMP12]], align 8
|
|
// MANDATORY-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
|
|
// MANDATORY-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP13]], align 4
|
|
// MANDATORY-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
|
|
// MANDATORY-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP14]], align 4
|
|
// MANDATORY-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
|
|
// MANDATORY-NEXT: store i32 0, ptr [[TMP15]], align 4
|
|
// MANDATORY-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 [[TMP2]], i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8host_devi_l22.region_id, ptr [[KERNEL_ARGS]])
|
|
// MANDATORY-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
|
|
// MANDATORY-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
|
|
// MANDATORY: omp_offload.failed:
|
|
// MANDATORY-NEXT: unreachable
|
|
// MANDATORY: omp_offload.cont:
|
|
// MANDATORY-NEXT: ret void
|
|
//
|
|
//
|
|
// MANDATORY-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
|
|
// MANDATORY-SAME: () #[[ATTR3:[0-9]+]] {
|
|
// MANDATORY-NEXT: entry:
|
|
// MANDATORY-NEXT: call void @__tgt_register_requires(i64 1)
|
|
// MANDATORY-NEXT: ret void
|
|
//
|