fc0653f31c
Recent upstream trends have moved away from explicitly using `-verify-machineinstrs`, as it's already covered by the expensive checks. This PR removes almost all `-verify-machineinstrs` from tests in `llvm/test/CodeGen/AMDGPU/*.ll`, leaving only those tests where its removal currently causes failures.
34 lines
1.2 KiB
LLVM
34 lines
1.2 KiB
LLVM
; RUN: llc -mtriple=amdgcn--amdhsa -mcpu=kaveri < %s | FileCheck -check-prefix=GCN %s
|
|
; RUN: not llc -mtriple=amdgcn-unknown-unknown -mcpu=kaveri < %s 2>&1 | FileCheck -check-prefix=ERROR %s
|
|
|
|
; ERROR: in function test{{.*}}: unsupported hsa intrinsic without hsa target
|
|
|
|
; GCN-LABEL: {{^}}test:
|
|
; GCN: s_load_dword s{{[0-9]+}}, s[6:7], 0x0
|
|
; GCN: .amdhsa_user_sgpr_queue_ptr 1
|
|
define amdgpu_kernel void @test(ptr addrspace(1) %out) {
|
|
%queue_ptr = call noalias ptr addrspace(4) @llvm.amdgcn.queue.ptr() #0
|
|
%value = load i32, ptr addrspace(4) %queue_ptr
|
|
store i32 %value, ptr addrspace(1) %out
|
|
ret void
|
|
}
|
|
|
|
; FIXME: Should really be able to delete the load
|
|
; GCN-LABEL: {{^}}test_ub:
|
|
; GCN: s_load_dword s{{[0-9]+}}, s[0:1], 0x0
|
|
; GCN: .amdhsa_user_sgpr_queue_ptr 0
|
|
define amdgpu_kernel void @test_ub(ptr addrspace(1) %out) #1 {
|
|
%queue_ptr = call noalias ptr addrspace(4) @llvm.amdgcn.queue.ptr() #0
|
|
%value = load i32, ptr addrspace(4) %queue_ptr
|
|
store i32 %value, ptr addrspace(1) %out
|
|
ret void
|
|
}
|
|
|
|
declare noalias ptr addrspace(4) @llvm.amdgcn.queue.ptr() #0
|
|
|
|
attributes #0 = { nounwind readnone }
|
|
attributes #1 = { "amdgpu-no-queue-ptr" }
|
|
|
|
!llvm.module.flags = !{!0}
|
|
!0 = !{i32 1, !"amdhsa_code_object_version", i32 400}
|