[SLP]Followup fix for the poisonous logical op in reductions

If the VectorizedTree still may generate poisonous value, but it is not the original operand of the reduction op, need to check if Res still the operand, to generate correct code. Fixes #114905
intel · Dec 26, 2024 · 889215a · 889215a
1 parent 1b476ec
commit 889215a
Show file tree

Hide file tree

Showing 3 changed files with 15 additions and 15 deletions.
diff --git a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
@@ -19821,21 +19821,21 @@ class HorizontalReduction {
         Builder.SetCurrentDebugLocation(
             cast<Instruction>(ReductionOps.front().front())->getDebugLoc());
         if (AnyBoolLogicOp) {
-
-          if (auto It = ReducedValsToOps.find(VectorizedTree);
-              It == ReducedValsToOps.end() ||
+          auto It = ReducedValsToOps.find(VectorizedTree);
+          auto It1 = ReducedValsToOps.find(Res);
+          if ((It == ReducedValsToOps.end() && It1 == ReducedValsToOps.end()) ||
               isGuaranteedNotToBePoison(VectorizedTree, AC) ||
-              any_of(It->getSecond(), [&](Instruction *I) {
-                return isBoolLogicOp(I) &&
-                       getRdxOperand(I, 0) == VectorizedTree;
-              })) {
+              (It != ReducedValsToOps.end() &&
+               any_of(It->getSecond(), [&](Instruction *I) {
+                 return isBoolLogicOp(I) &&
+                        getRdxOperand(I, 0) == VectorizedTree;
+               }))) {
             ;
-          } else if (auto It = ReducedValsToOps.find(Res);
-                     It == ReducedValsToOps.end() ||
-                     isGuaranteedNotToBePoison(Res, AC) ||
-                     any_of(It->getSecond(), [&](Instruction *I) {
+          } else if (isGuaranteedNotToBePoison(Res, AC) ||
+                     (It1 != ReducedValsToOps.end() &&
+                     any_of(It1->getSecond(), [&](Instruction *I) {
                        return isBoolLogicOp(I) && getRdxOperand(I, 0) == Res;
-                     })) {
+                     }))) {
             std::swap(VectorizedTree, Res);
           } else {
             VectorizedTree = Builder.CreateFreeze(VectorizedTree);

diff --git a/llvm/test/Transforms/SLPVectorizer/X86/reduction-logical.ll b/llvm/test/Transforms/SLPVectorizer/X86/reduction-logical.ll
@@ -428,7 +428,7 @@ define i1 @logical_and_icmp_extra_op(<4 x i32> %x, <4 x i32> %y, i1 %c) {
 ; CHECK-NEXT:    [[TMP1:%.*]] = icmp slt <4 x i32> [[X:%.*]], [[Y:%.*]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = freeze <4 x i1> [[TMP1]]
 ; CHECK-NEXT:    [[TMP3:%.*]] = call i1 @llvm.vector.reduce.and.v4i1(<4 x i1> [[TMP2]])
-; CHECK-NEXT:    [[OP_RDX:%.*]] = select i1 [[TMP3]], i1 [[C:%.*]], i1 false
+; CHECK-NEXT:    [[OP_RDX:%.*]] = select i1 [[C:%.*]], i1 [[TMP3]], i1 false
 ; CHECK-NEXT:    ret i1 [[OP_RDX]]
 ;
   %x0 = extractelement <4 x i32> %x, i32 0
@@ -456,7 +456,7 @@ define i1 @logical_or_icmp_extra_op(<4 x i32> %x, <4 x i32> %y, i1 %c) {
 ; CHECK-NEXT:    [[TMP1:%.*]] = icmp slt <4 x i32> [[X:%.*]], [[Y:%.*]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = freeze <4 x i1> [[TMP1]]
 ; CHECK-NEXT:    [[TMP3:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP2]])
-; CHECK-NEXT:    [[OP_RDX:%.*]] = select i1 [[TMP3]], i1 true, i1 [[C:%.*]]
+; CHECK-NEXT:    [[OP_RDX:%.*]] = select i1 [[C:%.*]], i1 true, i1 [[TMP3]]
 ; CHECK-NEXT:    ret i1 [[OP_RDX]]
 ;
   %x0 = extractelement <4 x i32> %x, i32 0

diff --git a/llvm/test/Transforms/SLPVectorizer/logical-ops-poisonous-repeated.ll b/llvm/test/Transforms/SLPVectorizer/logical-ops-poisonous-repeated.ll
@@ -14,7 +14,7 @@ define i1 @test(<4 x i32> %x) {
 ; CHECK-NEXT:    [[C3:%.*]] = icmp slt i32 [[X3]], 0
 ; CHECK-NEXT:    [[TMP2:%.*]] = freeze i1 [[C3]]
 ; CHECK-NEXT:    [[OP_RDX:%.*]] = select i1 [[TMP2]], i1 [[C1]], i1 false
-; CHECK-NEXT:    [[OP_RDX1:%.*]] = select i1 [[OP_RDX]], i1 [[TMP1]], i1 false
+; CHECK-NEXT:    [[OP_RDX1:%.*]] = select i1 [[TMP1]], i1 [[OP_RDX]], i1 false
 ; CHECK-NEXT:    ret i1 [[OP_RDX1]]
 ;
   %x0 = extractelement <4 x i32> %x, i32 0