[X86][ISel] Fix VPTERNLOG matching ensuring the InnerOp is logicOp #166591
+14
−2
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Member
|
@llvm/pr-subscribers-backend-x86 Author: Yi-Chi Lee (yichi170) ChangesThis patch fixes a crash in Previously, This patch applies A new test is added to reproduce the same crash mentioned in #164863 by @RKSimon . Full diff: https://github.com/llvm/llvm-project/pull/166591.diff 2 Files Affected:
diff --git a/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp b/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
index d4418c8563780..6c16fcfb282e8 100644
--- a/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
+++ b/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
@@ -4728,9 +4728,9 @@ bool X86DAGToDAGISel::tryVPTERNLOG(SDNode *N) {
auto tryPeelOuterNotWrappingLogic = [&](SDNode *Op) {
if (Op->getOpcode() == ISD::XOR && Op->hasOneUse() &&
ISD::isBuildVectorAllOnes(Op->getOperand(1).getNode())) {
- SDValue InnerOp = Op->getOperand(0);
+ SDValue InnerOp = getFoldableLogicOp(Op->getOperand(0));
- if (!getFoldableLogicOp(InnerOp))
+ if (!InnerOp)
return SDValue();
N0 = InnerOp.getOperand(0);
diff --git a/llvm/test/CodeGen/X86/issue163738.ll b/llvm/test/CodeGen/X86/vpternlog.ll
similarity index 59%
rename from llvm/test/CodeGen/X86/issue163738.ll
rename to llvm/test/CodeGen/X86/vpternlog.ll
index 61fe043a970dd..bd7478d3a82d5 100644
--- a/llvm/test/CodeGen/X86/issue163738.ll
+++ b/llvm/test/CodeGen/X86/vpternlog.ll
@@ -11,3 +11,15 @@ define <8 x i64> @foo(<8 x i64> %a, <8 x i64> %b, <8 x i64> %c) {
%and3 = xor <8 x i64> %and3.demorgan, splat (i64 -1)
ret <8 x i64> %and3
}
+
+define <8 x i64> @xorbitcast(<64 x i8> %a, <64 x i8> %b, <64 x i8> %c) {
+; CHECK-LABEL: xorbitcast:
+; CHECK: # %bb.0:
+; CHECK-NEXT: vpternlogq {{.*#+}} zmm0 = ~(zmm0 | zmm2 | zmm1)
+; CHECK-NEXT: retq
+ %or1 = or <64 x i8> %a, %b
+ %or2 = or <64 x i8> %or1, %c
+ %cast = bitcast <64 x i8> %or2 to <8 x i64>
+ %xor = xor <8 x i64> %cast, splat (i64 -1)
+ ret <8 x i64> %xor
+}
|
Collaborator
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/10/builds/16774 Here is the relevant piece of the build log for the reference |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This patch fixes a crash in
tryVPTERNLOGwhen trying to peel out the outer not in cases like~(A | B | C).Previously,
InnerOpwas taken directly fromOp->getOperand(0)before verifying that it was a logical operation. As a result, the code could later accessInnerOp->getOperand(0)orInnerOp->getOperand(1)even whenInnerOpwas something like a bitcast, causing an error.This patch applies
getFoldableLogicOptoInnerOp, ensuring thatInnerOpis a valid logic operation before it is dereferenced.A simplified test is added to reproduce the same crash mentioned in #164863 by @RKSimon.