mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-10-23 13:02:52 +02:00
dd9ab77318
Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
40 lines
1.6 KiB
LLVM
40 lines
1.6 KiB
LLVM
; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
|
|
; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
|
|
; RUN: llc -march=r600 -mcpu=redwood < %s | FileCheck -check-prefix=R600 -check-prefix=FUNC %s
|
|
|
|
; FUNC-LABEL: {{^}}atomic_sub_local:
|
|
; R600: LDS_SUB *
|
|
; SI: ds_sub_u32
|
|
define amdgpu_kernel void @atomic_sub_local(i32 addrspace(3)* %local) {
|
|
%unused = atomicrmw volatile sub i32 addrspace(3)* %local, i32 5 seq_cst
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}atomic_sub_local_const_offset:
|
|
; R600: LDS_SUB *
|
|
; SI: ds_sub_u32 v{{[0-9]+}}, v{{[0-9]+}} offset:16
|
|
define amdgpu_kernel void @atomic_sub_local_const_offset(i32 addrspace(3)* %local) {
|
|
%gep = getelementptr i32, i32 addrspace(3)* %local, i32 4
|
|
%val = atomicrmw volatile sub i32 addrspace(3)* %gep, i32 5 seq_cst
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}atomic_sub_ret_local:
|
|
; R600: LDS_SUB_RET *
|
|
; SI: ds_sub_rtn_u32
|
|
define amdgpu_kernel void @atomic_sub_ret_local(i32 addrspace(1)* %out, i32 addrspace(3)* %local) {
|
|
%val = atomicrmw volatile sub i32 addrspace(3)* %local, i32 5 seq_cst
|
|
store i32 %val, i32 addrspace(1)* %out
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}atomic_sub_ret_local_const_offset:
|
|
; R600: LDS_SUB_RET *
|
|
; SI: ds_sub_rtn_u32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} offset:20
|
|
define amdgpu_kernel void @atomic_sub_ret_local_const_offset(i32 addrspace(1)* %out, i32 addrspace(3)* %local) {
|
|
%gep = getelementptr i32, i32 addrspace(3)* %local, i32 5
|
|
%val = atomicrmw volatile sub i32 addrspace(3)* %gep, i32 5 seq_cst
|
|
store i32 %val, i32 addrspace(1)* %out
|
|
ret void
|
|
}
|