llvm-project/clang/test/SemaHIP/amdgpu-builtin-in-lambda.hip
Juan Manuel Martinez Caamaño d995b2ebdc
[Clang][AMDGPU] Accept builtins in lambda declarations (#135027)
`Sema::getCurFunctionDecl(AllowLambda = false)` returns a nullptr when
the lambda declaration is outside a function (for example, when
assigning a lambda to a static constexpr variable).

This triggered an assertion in
`SemaAMDGPU::CheckAMDGCNBuiltinFunctionCall`.

Using `Sema::getCurFunctionDecl(AllowLambda = true)` returns the
declaration of the enclosing lambda.

Stumbled with this issue when refactoring some code in CK.
2025-04-11 19:11:46 +02:00

54 lines
2.3 KiB
Plaintext

// RUN: %clang_cc1 -std=c++20 -triple amdgcn -target-cpu gfx90a -fsyntax-only -fcuda-is-device -verify=gfx90a -o - %s
// RUN: %clang_cc1 -std=c++20 -triple amdgcn -target-cpu gfx950 -fsyntax-only -fcuda-is-device -o - %s
#define __device__ __attribute__((device))
#define __shared__ __attribute__((shared))
struct S {
static constexpr auto make_buffer_rsrc_lambda = [](void *p, short stride, int num, int flags) {
return __builtin_amdgcn_make_buffer_rsrc(p, stride, num, flags);
};
static constexpr auto global_load_lds_lambda = [](void* src, __shared__ void *dst) {
__builtin_amdgcn_global_load_lds(src, dst, 16, 0, 0); // gfx90a-error{{invalid size value}} gfx90a-note{{size must be 1, 2, or 4}}
};
};
__device__ __amdgpu_buffer_rsrc_t test_simple_builtin(void *p, short stride, int num, int flags) {
return S::make_buffer_rsrc_lambda(p, stride, num, flags);
}
__device__ void test_target_dependant_builtin(void *src, __shared__ void *dst) {
S::global_load_lds_lambda(src, dst);
}
constexpr auto make_buffer_rsrc_lambda = [](void *p, short stride, int num, int flags) {
return __builtin_amdgcn_make_buffer_rsrc(p, stride, num, flags);
};
constexpr auto global_load_lds_lambda = [](void* src, __shared__ void *dst) {
__builtin_amdgcn_global_load_lds(src, dst, 16, 0, 0); // gfx90a-error{{invalid size value}} gfx90a-note{{size must be 1, 2, or 4}}
};
__device__ __amdgpu_buffer_rsrc_t global_test_simple_builtin(void *p, short stride, int num, int flags) {
return make_buffer_rsrc_lambda(p, stride, num, flags);
}
__device__ void global_test_target_dependant_builtin(void *src, __shared__ void *dst) {
global_load_lds_lambda(src, dst);
}
__device__ __amdgpu_buffer_rsrc_t local_test_simple_builtin(void *p, short stride, int num, int flags) {
constexpr auto f = [](void *p, short stride, int num, int flags) {
return __builtin_amdgcn_make_buffer_rsrc(p, stride, num, flags);
};
return f(p, stride, num, flags);
}
__device__ void local_test_target_dependant_builtin(void *src, __shared__ void *dst) {
constexpr auto f = [](void* src, __shared__ void *dst) {
__builtin_amdgcn_global_load_lds(src, dst, 16, 0, 0); // gfx90a-error{{invalid size value}} gfx90a-note{{size must be 1, 2, or 4}}
};
f(src, dst);
}