[llvm] b606e2d - [Evaluator] Bitcast result of pointer stripping
Arthur Eubanks via llvm-commits
llvm-commits at lists.llvm.org
Wed Apr 21 13:36:58 PDT 2021
Author: Arthur Eubanks
Date: 2021-04-21T13:32:29-07:00
New Revision: b606e2df4dd49c17c9f2733f2339b6dd0dd427de
URL: https://github.com/llvm/llvm-project/commit/b606e2df4dd49c17c9f2733f2339b6dd0dd427de
DIFF: https://github.com/llvm/llvm-project/commit/b606e2df4dd49c17c9f2733f2339b6dd0dd427de.diff
LOG: [Evaluator] Bitcast result of pointer stripping
Trying to evaluate a GEP would assert with
"Ty == cast<PointerType>(C->getType()->getScalarType())->getElementType()"
because the type of the pointer we would evaluate the GEP argument to
would be a different type than the GEP was expecting. We should treat
pointer stripping as a bitcast.
The test adds a redundant GEP that would crash due to type mismatch.
Reviewed By: rnk
Differential Revision: https://reviews.llvm.org/D100970
Added:
Modified:
llvm/lib/Transforms/Utils/Evaluator.cpp
llvm/test/Transforms/GlobalOpt/invariant.group.ll
Removed:
################################################################################
diff --git a/llvm/lib/Transforms/Utils/Evaluator.cpp b/llvm/lib/Transforms/Utils/Evaluator.cpp
index 7404ceb83ab87..02cfb09241721 100644
--- a/llvm/lib/Transforms/Utils/Evaluator.cpp
+++ b/llvm/lib/Transforms/Utils/Evaluator.cpp
@@ -566,6 +566,7 @@ bool Evaluator::EvaluateBlock(BasicBlock::iterator CurInst, BasicBlock *&NextBB,
<< "Stripped pointer casts for alias analysis for "
"intrinsic call.\n");
StrippedPointerCastsForAliasAnalysis = true;
+ InstResult = ConstantExpr::getBitCast(InstResult, II->getType());
} else {
LLVM_DEBUG(dbgs() << "Unknown intrinsic. Cannot evaluate.\n");
return false;
diff --git a/llvm/test/Transforms/GlobalOpt/invariant.group.ll b/llvm/test/Transforms/GlobalOpt/invariant.group.ll
index 586432b60b650..6d192939a5e04 100644
--- a/llvm/test/Transforms/GlobalOpt/invariant.group.ll
+++ b/llvm/test/Transforms/GlobalOpt/invariant.group.ll
@@ -32,9 +32,10 @@ enter:
%0 = bitcast i32* %valptr to i8*
%barr = call i8* @llvm.launder.invariant.group(i8* %0)
- %1 = bitcast i8* %barr to i32*
+ %1 = getelementptr i8, i8* %barr, i32 0
+ %2 = bitcast i8* %1 to i32*
- %val2 = load i32, i32* %1
+ %val2 = load i32, i32* %2
store i32 %val2, i32* @tmp2
ret void
}
More information about the llvm-commits
mailing list