[MergeICmp] Fix a bug in MergeICmp that can lead to a block being processed more than once.
Summary:
Fix a bug in MergeICmp that can lead to a BCECmp block being processed more than once and eventually lead to a broken LLVM module.
The problem is that if the non-constant value is not produced by the last block, the producer will be processed once when the its parent block
is processed and second time when the last block is processed.
We end up having 2 same BCECmpBlock in the merge queue. And eventually lead to a broken LLVM module.
Reviewers: courbet, davide
Reviewed By: courbet
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D43825
llvm-svn: 326318
diff --git a/llvm/lib/Transforms/Scalar/MergeICmps.cpp b/llvm/lib/Transforms/Scalar/MergeICmps.cpp
index 745d29d..3e69512 100644
--- a/llvm/lib/Transforms/Scalar/MergeICmps.cpp
+++ b/llvm/lib/Transforms/Scalar/MergeICmps.cpp
@@ -571,6 +571,19 @@
DEBUG(dbgs() << "skip: several non-constant values\n");
return false;
}
+ if (!isa<ICmpInst>(Phi.getIncomingValue(I)) ||
+ cast<ICmpInst>(Phi.getIncomingValue(I))->getParent() !=
+ Phi.getIncomingBlock(I)) {
+ // Non-constant incoming value is not from a cmp instruction or not
+ // produced by the last block. We could end up processing the value
+ // producing block more than once.
+ //
+ // This is an uncommon case, so we bail.
+ DEBUG(
+ dbgs()
+ << "skip: non-constant value not from cmp or not from last block.\n");
+ return false;
+ }
LastBlock = Phi.getIncomingBlock(I);
}
if (!LastBlock) {