mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-24 03:33:20 +01:00
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
116 lines
5.7 KiB
LLVM
116 lines
5.7 KiB
LLVM
; RUN: llc -O0 -verify-machineinstrs -asm-verbose=0 -march=amdgcn < %s | FileCheck -check-prefix=SI -check-prefix=COMMON %s
|
|
; RUN: llc -O0 -verify-machineinstrs -asm-verbose=0 -march=amdgcn -mcpu=tonga < %s | FileCheck -check-prefix=SI -check-prefix=COMMON %s
|
|
; XUN: llc -O0 -verify-machineinstrs -asm-verbose=0 -march=r600 -mcpu=redwood < %s | FileCheck -check-prefix=EG -check-prefix=COMMON %s
|
|
|
|
; SI hits an assertion at -O0, evergreen hits a not implemented unreachable.
|
|
|
|
; COMMON-LABEL: {{^}}branch_true:
|
|
define amdgpu_kernel void @branch_true(i8 addrspace(1)* nocapture %main, i32 %main_stride) #0 {
|
|
entry:
|
|
br i1 true, label %for.end, label %for.body.lr.ph
|
|
|
|
for.body.lr.ph: ; preds = %entry
|
|
%add.ptr.sum = shl i32 %main_stride, 1
|
|
%add.ptr1.sum = add i32 %add.ptr.sum, %main_stride
|
|
%add.ptr4.sum = shl i32 %main_stride, 2
|
|
br label %for.body
|
|
|
|
for.body: ; preds = %for.body, %for.body.lr.ph
|
|
%main.addr.011 = phi i8 addrspace(1)* [ %main, %for.body.lr.ph ], [ %add.ptr6, %for.body ]
|
|
%0 = bitcast i8 addrspace(1)* %main.addr.011 to i32 addrspace(1)*
|
|
%1 = load i32, i32 addrspace(1)* %0, align 4
|
|
%add.ptr = getelementptr inbounds i8, i8 addrspace(1)* %main.addr.011, i32 %main_stride
|
|
%2 = bitcast i8 addrspace(1)* %add.ptr to i32 addrspace(1)*
|
|
%3 = load i32, i32 addrspace(1)* %2, align 4
|
|
%add.ptr1 = getelementptr inbounds i8, i8 addrspace(1)* %main.addr.011, i32 %add.ptr.sum
|
|
%4 = bitcast i8 addrspace(1)* %add.ptr1 to i32 addrspace(1)*
|
|
%5 = load i32, i32 addrspace(1)* %4, align 4
|
|
%add.ptr2 = getelementptr inbounds i8, i8 addrspace(1)* %main.addr.011, i32 %add.ptr1.sum
|
|
%6 = bitcast i8 addrspace(1)* %add.ptr2 to i32 addrspace(1)*
|
|
%7 = load i32, i32 addrspace(1)* %6, align 4
|
|
%add.ptr3 = getelementptr inbounds i8, i8 addrspace(1)* %main.addr.011, i32 %add.ptr4.sum
|
|
%8 = bitcast i8 addrspace(1)* %add.ptr3 to i32 addrspace(1)*
|
|
%9 = load i32, i32 addrspace(1)* %8, align 4
|
|
%add.ptr6 = getelementptr inbounds i8, i8 addrspace(1)* %main.addr.011, i32 undef
|
|
br i1 undef, label %for.end, label %for.body
|
|
|
|
for.end: ; preds = %for.body, %entry
|
|
ret void
|
|
}
|
|
|
|
; COMMON-LABEL: {{^}}branch_false:
|
|
; SI: s_cbranch_vccnz
|
|
; SI: s_cbranch_scc1
|
|
; SI: s_endpgm
|
|
define amdgpu_kernel void @branch_false(i8 addrspace(1)* nocapture %main, i32 %main_stride) #0 {
|
|
entry:
|
|
br i1 false, label %for.end, label %for.body.lr.ph
|
|
|
|
for.body.lr.ph: ; preds = %entry
|
|
%add.ptr.sum = shl i32 %main_stride, 1
|
|
%add.ptr1.sum = add i32 %add.ptr.sum, %main_stride
|
|
%add.ptr4.sum = shl i32 %main_stride, 2
|
|
br label %for.body
|
|
|
|
for.body: ; preds = %for.body, %for.body.lr.ph
|
|
%main.addr.011 = phi i8 addrspace(1)* [ %main, %for.body.lr.ph ], [ %add.ptr6, %for.body ]
|
|
%0 = bitcast i8 addrspace(1)* %main.addr.011 to i32 addrspace(1)*
|
|
%1 = load i32, i32 addrspace(1)* %0, align 4
|
|
%add.ptr = getelementptr inbounds i8, i8 addrspace(1)* %main.addr.011, i32 %main_stride
|
|
%2 = bitcast i8 addrspace(1)* %add.ptr to i32 addrspace(1)*
|
|
%3 = load i32, i32 addrspace(1)* %2, align 4
|
|
%add.ptr1 = getelementptr inbounds i8, i8 addrspace(1)* %main.addr.011, i32 %add.ptr.sum
|
|
%4 = bitcast i8 addrspace(1)* %add.ptr1 to i32 addrspace(1)*
|
|
%5 = load i32, i32 addrspace(1)* %4, align 4
|
|
%add.ptr2 = getelementptr inbounds i8, i8 addrspace(1)* %main.addr.011, i32 %add.ptr1.sum
|
|
%6 = bitcast i8 addrspace(1)* %add.ptr2 to i32 addrspace(1)*
|
|
%7 = load i32, i32 addrspace(1)* %6, align 4
|
|
%add.ptr3 = getelementptr inbounds i8, i8 addrspace(1)* %main.addr.011, i32 %add.ptr4.sum
|
|
%8 = bitcast i8 addrspace(1)* %add.ptr3 to i32 addrspace(1)*
|
|
%9 = load i32, i32 addrspace(1)* %8, align 4
|
|
%add.ptr6 = getelementptr inbounds i8, i8 addrspace(1)* %main.addr.011, i32 undef
|
|
br i1 undef, label %for.end, label %for.body
|
|
|
|
for.end: ; preds = %for.body, %entry
|
|
ret void
|
|
}
|
|
|
|
; COMMON-LABEL: {{^}}branch_undef:
|
|
; SI: s_cbranch_scc1
|
|
; SI: s_cbranch_scc1
|
|
; SI: s_endpgm
|
|
define amdgpu_kernel void @branch_undef(i8 addrspace(1)* nocapture %main, i32 %main_stride) #0 {
|
|
entry:
|
|
br i1 undef, label %for.end, label %for.body.lr.ph
|
|
|
|
for.body.lr.ph: ; preds = %entry
|
|
%add.ptr.sum = shl i32 %main_stride, 1
|
|
%add.ptr1.sum = add i32 %add.ptr.sum, %main_stride
|
|
%add.ptr4.sum = shl i32 %main_stride, 2
|
|
br label %for.body
|
|
|
|
for.body: ; preds = %for.body, %for.body.lr.ph
|
|
%main.addr.011 = phi i8 addrspace(1)* [ %main, %for.body.lr.ph ], [ %add.ptr6, %for.body ]
|
|
%0 = bitcast i8 addrspace(1)* %main.addr.011 to i32 addrspace(1)*
|
|
%1 = load i32, i32 addrspace(1)* %0, align 4
|
|
%add.ptr = getelementptr inbounds i8, i8 addrspace(1)* %main.addr.011, i32 %main_stride
|
|
%2 = bitcast i8 addrspace(1)* %add.ptr to i32 addrspace(1)*
|
|
%3 = load i32, i32 addrspace(1)* %2, align 4
|
|
%add.ptr1 = getelementptr inbounds i8, i8 addrspace(1)* %main.addr.011, i32 %add.ptr.sum
|
|
%4 = bitcast i8 addrspace(1)* %add.ptr1 to i32 addrspace(1)*
|
|
%5 = load i32, i32 addrspace(1)* %4, align 4
|
|
%add.ptr2 = getelementptr inbounds i8, i8 addrspace(1)* %main.addr.011, i32 %add.ptr1.sum
|
|
%6 = bitcast i8 addrspace(1)* %add.ptr2 to i32 addrspace(1)*
|
|
%7 = load i32, i32 addrspace(1)* %6, align 4
|
|
%add.ptr3 = getelementptr inbounds i8, i8 addrspace(1)* %main.addr.011, i32 %add.ptr4.sum
|
|
%8 = bitcast i8 addrspace(1)* %add.ptr3 to i32 addrspace(1)*
|
|
%9 = load i32, i32 addrspace(1)* %8, align 4
|
|
%add.ptr6 = getelementptr inbounds i8, i8 addrspace(1)* %main.addr.011, i32 undef
|
|
br i1 undef, label %for.end, label %for.body
|
|
|
|
for.end: ; preds = %for.body, %entry
|
|
ret void
|
|
}
|
|
|
|
attributes #0 = { nounwind }
|