[llvm] a105d30 - [X86] Fold (shift undef, X)->0 for vector shifts by immediate.
Craig Topper via llvm-commits
llvm-commits at lists.llvm.org
Thu May 27 09:32:01 PDT 2021
Author: Craig Topper
Date: 2021-05-27T09:31:47-07:00
New Revision: a105d3024efec365961e940c489c4ed5198736d2
URL: https://github.com/llvm/llvm-project/commit/a105d3024efec365961e940c489c4ed5198736d2
DIFF: https://github.com/llvm/llvm-project/commit/a105d3024efec365961e940c489c4ed5198736d2.diff
LOG: [X86] Fold (shift undef, X)->0 for vector shifts by immediate.
We could previously do this by accident through the later
call to getTargetConstantBitsFromNode I think, but that only worked
if N0 had a single use. This patch makes it explicit for undef and
doesn't have a use count check.
I think this is needed to move the (shl X, 1)->(add X, X)
fold to isel for PR50468. We need to be sure X won't be IMPLICIT_DEF
which might prevent the same vreg from being used for both operands.
Differential Revision: https://reviews.llvm.org/D103192
Added:
Modified:
llvm/lib/Target/X86/X86ISelLowering.cpp
llvm/test/CodeGen/X86/vec_shift5.ll
Removed:
################################################################################
diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index 70bf79c1de68..71d1399fa8c5 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -43866,6 +43866,10 @@ static SDValue combineVectorShiftImm(SDNode *N, SelectionDAG &DAG,
assert(N->getOperand(1).getValueType() == MVT::i8 &&
"Unexpected shift amount type");
+ // (shift undef, X) -> 0
+ if (N0.isUndef())
+ return DAG.getConstant(0, SDLoc(N), VT);
+
// Out of range logical bit shifts are guaranteed to be zero.
// Out of range arithmetic bit shifts splat the sign bit.
unsigned ShiftVal = N->getConstantOperandVal(1);
diff --git a/llvm/test/CodeGen/X86/vec_shift5.ll b/llvm/test/CodeGen/X86/vec_shift5.ll
index f6f96710b81c..0514c6e61ade 100644
--- a/llvm/test/CodeGen/X86/vec_shift5.ll
+++ b/llvm/test/CodeGen/X86/vec_shift5.ll
@@ -176,16 +176,16 @@ define <4 x i32> @test17(<4 x i32> %a0, <4 x i32>* %dummy) {
; X86-LABEL: test17:
; X86: # %bb.0:
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
-; X86-NEXT: pslld $6, %xmm0
-; X86-NEXT: movdqa %xmm0, (%eax)
-; X86-NEXT: pslld $7, %xmm0
+; X86-NEXT: xorps %xmm0, %xmm0
+; X86-NEXT: movaps %xmm0, (%eax)
+; X86-NEXT: xorps %xmm0, %xmm0
; X86-NEXT: retl
;
; X64-LABEL: test17:
; X64: # %bb.0:
-; X64-NEXT: pslld $6, %xmm0
-; X64-NEXT: movdqa %xmm0, (%rdi)
-; X64-NEXT: pslld $7, %xmm0
+; X64-NEXT: xorps %xmm0, %xmm0
+; X64-NEXT: movaps %xmm0, (%rdi)
+; X64-NEXT: xorps %xmm0, %xmm0
; X64-NEXT: retq
%a = call <4 x i32> @llvm.x86.sse2.pslli.d(<4 x i32> undef, i32 6)
store <4 x i32> %a, <4 x i32>* %dummy
@@ -197,16 +197,16 @@ define <4 x i32> @test18(<4 x i32> %a0, <4 x i32>* %dummy) {
; X86-LABEL: test18:
; X86: # %bb.0:
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
-; X86-NEXT: pslld $3, %xmm0
-; X86-NEXT: movdqa %xmm0, (%eax)
-; X86-NEXT: pslld $1, %xmm0
+; X86-NEXT: xorps %xmm0, %xmm0
+; X86-NEXT: movaps %xmm0, (%eax)
+; X86-NEXT: xorps %xmm0, %xmm0
; X86-NEXT: retl
;
; X64-LABEL: test18:
; X64: # %bb.0:
-; X64-NEXT: pslld $3, %xmm0
-; X64-NEXT: movdqa %xmm0, (%rdi)
-; X64-NEXT: pslld $1, %xmm0
+; X64-NEXT: xorps %xmm0, %xmm0
+; X64-NEXT: movaps %xmm0, (%rdi)
+; X64-NEXT: xorps %xmm0, %xmm0
; X64-NEXT: retq
%a = call <4 x i32> @llvm.x86.sse2.pslli.d(<4 x i32> undef, i32 3)
store <4 x i32> %a, <4 x i32>* %dummy
More information about the llvm-commits
mailing list