6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2025-01-08 03:30:22 +00:00
Code Issues Projects Releases Wiki Activity

R600: Better fix for bug 20982

Browse Source 
Just do the left shift as unsigned to avoid the UB.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218092 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Matt Arsenault 2014-09-19 00:42:06 +00:00
parent 9b676fd6f2
commit bd2b96a12d
2 changed files with 6 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
lib/Target/R600/AMDGPUISelLowering.cpp
View File

@ -1896,7 +1896,8 @@ template <typename IntTy>
static SDValue constantFoldBFE(SelectionDAG &DAG, IntTy Src0,
uint32_t Offset, uint32_t Width) {
if (Width + Offset < 32) {
IntTy Result = (Src0 << (32 - Offset - Width)) >> (32 - Width);
uint32_t Shl = static_cast<uint32_t>(Src0) << (32 - Offset - Width);
IntTy Result = static_cast<IntTy>(Shl) >> (32 - Width);
return DAG.getConstant(Result, MVT::i32);
}
@ -2053,12 +2054,8 @@ SDValue AMDGPUTargetLowering::PerformDAGCombine(SDNode *N,
// Avoid undefined left shift of a negative in the constant fold.
// TODO: I'm not sure what the behavior of the hardware is, this should
// probably follow that instead.
int32_t Val = CVal->getSExtValue();
if (Val < 0)
return SDValue();
return constantFoldBFE<int32_t>(DAG,
Val,
CVal->getSExtValue(),
OffsetVal,
WidthVal);
}

7
test/CodeGen/R600/llvm.AMDGPU.bfe.i32.ll
View File

@ -370,13 +370,12 @@ define void @bfe_i32_constant_fold_test_15(i32 addrspace(1)* %out) nounwind {
ret void
}
; FIXME: This should fold to something
; FUNC-LABEL: @bfe_i32_constant_fold_test_16
; SI: S_BFE_I32 [[SREG:s[0-9]+]], -1, 0x70001
; SI: V_MOV_B32_e32 [[VREG:v[0-9]+]], [[SREG]]
; SI-NOT: BFE
; SI: V_MOV_B32_e32 [[VREG:v[0-9]+]], -1
; SI: BUFFER_STORE_DWORD [[VREG]],
; SI: S_ENDPGM
; EG-NOT: BFE
define void @bfe_i32_constant_fold_test_16(i32 addrspace(1)* %out) nounwind {
%bfe_i32 = call i32 @llvm.AMDGPU.bfe.i32(i32 4294967295, i32 1, i32 7) nounwind readnone
store i32 %bfe_i32, i32 addrspace(1)* %out, align 4