[llvm] b82b305 - [InstCombine] Fix out-of-bounds ashr(shl) optimization
Jeroen Dobbelaere via llvm-commits
llvm-commits at lists.llvm.org
Fri Apr 2 04:49:02 PDT 2021
Author: Jeroen Dobbelaere
Date: 2021-04-02T13:45:11+02:00
New Revision: b82b305cf94a57a7e0e72f576b85aaa136a505c3
URL: https://github.com/llvm/llvm-project/commit/b82b305cf94a57a7e0e72f576b85aaa136a505c3
DIFF: https://github.com/llvm/llvm-project/commit/b82b305cf94a57a7e0e72f576b85aaa136a505c3.diff
LOG: [InstCombine] Fix out-of-bounds ashr(shl) optimization
This fixes a crash found by the oss fuzzer and reported by @fhahn.
The suggestion of @RKSimon seems to be the correct fix here. (See D91343).
The oss fuzz report can be found here: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=32759
Reviewed By: lebedev.ri
Differential Revision: https://reviews.llvm.org/D99792
Added:
llvm/test/Transforms/InstCombine/oss_fuzz_32759.ll
Modified:
llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp
Removed:
################################################################################
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp b/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp
index 54ca2c10dd047..3c097a4952776 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp
@@ -927,7 +927,7 @@ Value *InstCombinerImpl::SimplifyMultipleUseDemandedBits(
unsigned BitWidth = DemandedMask.getBitWidth();
if (match(I,
m_AShr(m_Shl(m_Value(X), m_APInt(ShiftLC)), m_APInt(ShiftRC))) &&
- ShiftLC == ShiftRC &&
+ ShiftLC == ShiftRC && ShiftLC->ult(BitWidth) &&
DemandedMask.isSubsetOf(APInt::getLowBitsSet(
BitWidth, BitWidth - ShiftRC->getZExtValue()))) {
return X;
diff --git a/llvm/test/Transforms/InstCombine/oss_fuzz_32759.ll b/llvm/test/Transforms/InstCombine/oss_fuzz_32759.ll
new file mode 100644
index 0000000000000..4f0443265b110
--- /dev/null
+++ b/llvm/test/Transforms/InstCombine/oss_fuzz_32759.ll
@@ -0,0 +1,29 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -instcombine -S | FileCheck %s
+target datalayout = "n32"
+
+define i1 @oss_fuzz_32759(i1 %y) {
+; CHECK-LABEL: @oss_fuzz_32759(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: br i1 undef, label [[COND_TRUE:%.*]], label [[END:%.*]]
+; CHECK: cond.true:
+; CHECK-NEXT: br label [[END]]
+; CHECK: end:
+; CHECK-NEXT: ret i1 false
+;
+entry:
+ br i1 undef, label %cond.true, label %end
+
+cond.true: ; preds = %entry
+ %zy = zext i1 %y to i32
+ %B6 = shl i32 %zy, 2147483647
+ %B3 = ashr i32 %B6, 2147483647
+ %B7 = srem i32 %B3, 123
+ %cond = xor i32 %B7, %B3
+ br label %end
+
+end: ; preds = %cond.true, %entry
+ %p = phi i32 [ %cond, %cond.true ], [ -1, %entry ]
+ %r = icmp eq i32 %p, 0
+ ret i1 %r
+}
More information about the llvm-commits
mailing list