maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

Revert rGe6a7e3b5e3e7 "[X86][SSE] matchShuffleWithSHUFPD - add support for unary shuffles." 

This reverts commit e6a7e3b5e3.

Avoids register pressure regression reported at PR45263
This commit is contained in:
Simon Pilgrim 2020-03-21 12:14:02 +00:00
parent 49ccb32fd4
commit f424d51c3e
2 changed files with 3 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
llvm/lib/Target/X86/X86ISelLowering.cpp
View File

@ -15948,7 +15948,6 @@ static bool matchShuffleWithSHUFPD(MVT VT, SDValue &V1, SDValue &V2,
// Mask for V8F64: 0/1, 8/9, 2/3, 10/11, 4/5, ..
// Mask for V4F64; 0/1, 4/5, 2/3, 6/7..
ShuffleImm = 0;
bool UnaryMask = isUndefOrZeroOrInRange(Mask, 0, NumElts);
bool ShufpdMask = true;
bool CommutableMask = true;
for (int i = 0; i < NumElts; ++i) {
@ -15956,7 +15955,7 @@ static bool matchShuffleWithSHUFPD(MVT VT, SDValue &V1, SDValue &V2,
continue;
if (Mask[i] < 0)
return false;
int Val = (i & 6) + (UnaryMask ? 0 : (NumElts * (i & 1)));
int Val = (i & 6) + NumElts * (i & 1);
int CommutVal = (i & 0xe) + NumElts * ((i & 1) ^ 1);
if (Mask[i] < Val || Mask[i] > Val + 1)
ShufpdMask = false;
@ -15968,9 +15967,7 @@ static bool matchShuffleWithSHUFPD(MVT VT, SDValue &V1, SDValue &V2,
if (!ShufpdMask && !CommutableMask)
return false;
if (UnaryMask)
V2 = V1;
else if (!ShufpdMask && CommutableMask)
if (!ShufpdMask && CommutableMask)
std::swap(V1, V2);
ForceV1Zero = ZeroLane[0];

2
llvm/test/CodeGen/X86/sse-intrinsics-fast-isel.ll
View File

@ -2791,7 +2791,7 @@ define void @test_mm_storeh_pi(x86_mmx *%a0, <4 x float> %a1) nounwind {
;
; X64-SSE2-LABEL: test_mm_storeh_pi:
; X64-SSE2: # %bb.0:
; X64-SSE2-NEXT: shufps $78, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x4e]
; X64-SSE2-NEXT: pshufd $78, %xmm0, %xmm0 # encoding: [0x66,0x0f,0x70,0xc0,0x4e]
; X64-SSE2-NEXT: # xmm0 = xmm0[2,3,0,1]
; X64-SSE2-NEXT: movq %xmm0, %rax # encoding: [0x66,0x48,0x0f,0x7e,0xc0]
; X64-SSE2-NEXT: movq %rax, (%rdi) # encoding: [0x48,0x89,0x07]