[SLP] Fix infinite loop in isUndefVector.

This fixes an infinite loop if isa<T>(II->getOperand(1)) is true.
Update Base at the top of the loop, before the continue.

Reviewed By: ABataev

Differential Revision: https://reviews.llvm.org/D144292

diff --git a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
index 92ca124..0b427e1 100644 (file)
--- a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
+++ b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
@@ -381,9 +381,9 @@ static SmallBitVector isUndefVector(const Value *V,
     if (!UseMask.empty()) {
       const Value *Base = V;
       while (auto *II = dyn_cast<InsertElementInst>(Base)) {
+        Base = II->getOperand(0);
         if (isa<T>(II->getOperand(1)))
           continue;
-        Base = II->getOperand(0);
         std::optional<unsigned> Idx = getInsertIndex(II);
         if (!Idx)
           continue;

diff --git a/llvm/test/Transforms/SLPVectorizer/AArch64/insertelement.ll b/llvm/test/Transforms/SLPVectorizer/AArch64/insertelement.ll
index 284a240..1198bb1 100644 (file)
--- a/llvm/test/Transforms/SLPVectorizer/AArch64/insertelement.ll
+++ b/llvm/test/Transforms/SLPVectorizer/AArch64/insertelement.ll
@@ -35,3 +35,25 @@ define <vscale x 2 x float> @insertelement-scalable-vector() {
 
 ; Function Attrs: nounwind readnone speculatable willreturn
 declare float @llvm.fabs.f32(float)
+
+
+define <4 x float> @insertelement_poison_lanes(ptr %0) {
+; CHECK-LABEL: @insertelement_poison_lanes(
+; CHECK-NEXT:    [[INS_1:%.*]] = insertelement <4 x float> zeroinitializer, float poison, i64 0
+; CHECK-NEXT:    [[INS_2:%.*]] = insertelement <4 x float> [[INS_1]], float 0.000000e+00, i64 0
+; CHECK-NEXT:    [[GEP_1:%.*]] = getelementptr double, ptr [[TMP0:%.*]], i64 1
+; CHECK-NEXT:    store <2 x double> <double 0.000000e+00, double 1.000000e+00>, ptr [[GEP_1]], align 8
+; CHECK-NEXT:    ret <4 x float> [[INS_2]]
+;
+  %trunc.1 = fptrunc double 0.000000e+00 to float
+  %trunc.2 = fptrunc double 1.000000e+00 to float
+  %ins.1 = insertelement <4 x float> zeroinitializer, float poison, i64 0
+  %ins.2 = insertelement <4 x float> %ins.1, float %trunc.1, i64 0
+  %ext.1 = fpext float %trunc.1 to double
+  %gep.1  = getelementptr double, ptr %0, i64 1
+  store double %ext.1, ptr %gep.1, align 8
+  %ext.2 = fpext float %trunc.2 to double
+  %gep.2 = getelementptr double, ptr %0, i64 2
+  store double %ext.2, ptr %gep.2, align 8
+  ret <4 x float> %ins.2
+}