2 files changed, 22 insertions, 17 deletions

diff --git a/llvm/lib/Transforms/InstCombine/InstCombineVectorOps.cpp b/llvm/lib/Transforms/InstCombine/InstCombineVectorOps.cpp
index 6ef3066..18a45c6 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineVectorOps.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineVectorOps.cpp
@@ -319,20 +319,20 @@ Instruction *InstCombinerImpl::foldBitcastExtElt(ExtractElementInst &Ext) {
   return nullptr;
 }
 
-/// Find elements of V demanded by UserInstr.
-static APInt findDemandedEltsBySingleUser(Value *V, Instruction *UserInstr) {
+/// Find elements of V demanded by UserInstr. If returns false, we were not able
+/// to determine all elements.
+static bool findDemandedEltsBySingleUser(Value *V, Instruction *UserInstr,
+                                         APInt &UnionUsedElts) {
   unsigned VWidth = cast<FixedVectorType>(V->getType())->getNumElements();
 
-  // Conservatively assume that all elements are needed.
-  APInt UsedElts(APInt::getAllOnes(VWidth));
-
   switch (UserInstr->getOpcode()) {
   case Instruction::ExtractElement: {
     ExtractElementInst *EEI = cast<ExtractElementInst>(UserInstr);
     assert(EEI->getVectorOperand() == V);
     ConstantInt *EEIIndexC = dyn_cast<ConstantInt>(EEI->getIndexOperand());
     if (EEIIndexC && EEIIndexC->getValue().ult(VWidth)) {
-      UsedElts = APInt::getOneBitSet(VWidth, EEIIndexC->getZExtValue());
+      UnionUsedElts.setBit(EEIIndexC->getZExtValue());
+      return true;
     }
     break;
   }
@@ -341,23 +341,23 @@ static APInt findDemandedEltsBySingleUser(Value *V, Instruction *UserInstr) {
     unsigned MaskNumElts =
         cast<FixedVectorType>(UserInstr->getType())->getNumElements();
 
-    UsedElts = APInt(VWidth, 0);
-    for (unsigned i = 0; i < MaskNumElts; i++) {
-      unsigned MaskVal = Shuffle->getMaskValue(i);
+    for (auto I : llvm::seq(MaskNumElts)) {
+      unsigned MaskVal = Shuffle->getMaskValue(I);
       if (MaskVal == -1u || MaskVal >= 2 * VWidth)
         continue;
       if (Shuffle->getOperand(0) == V && (MaskVal < VWidth))
-        UsedElts.setBit(MaskVal);
+        UnionUsedElts.setBit(MaskVal);
       if (Shuffle->getOperand(1) == V &&
           ((MaskVal >= VWidth) && (MaskVal < 2 * VWidth)))
-        UsedElts.setBit(MaskVal - VWidth);
+        UnionUsedElts.setBit(MaskVal - VWidth);
     }
-    break;
+    return true;
   }
   default:
     break;
   }
-  return UsedElts;
+
+  return false;
 }
 
 /// Find union of elements of V demanded by all its users.
@@ -370,7 +370,8 @@ static APInt findDemandedEltsByAllUsers(Value *V) {
   APInt UnionUsedElts(VWidth, 0);
   for (const Use &U : V->uses()) {
     if (Instruction *I = dyn_cast<Instruction>(U.getUser())) {
-      UnionUsedElts |= findDemandedEltsBySingleUser(V, I);
+      if (!findDemandedEltsBySingleUser(V, I, UnionUsedElts))
+        return APInt::getAllOnes(VWidth);
     } else {
       UnionUsedElts = APInt::getAllOnes(VWidth);
       break;
diff --git a/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp b/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp
index 5d2d79e..917004c 100644
--- a/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp
@@ -132,9 +132,11 @@ STATISTIC(NumReassoc  , "Number of reassociations");
 DEBUG_COUNTER(VisitCounter, "instcombine-visit",
               "Controls which instructions are visited");
 
-static cl::opt<bool>
-EnableCodeSinking("instcombine-code-sinking", cl::desc("Enable code sinking"),
-                                              cl::init(true));
+namespace llvm {
+
+static cl::opt<bool> EnableCodeSinking("instcombine-code-sinking",
+                                       cl::desc("Enable code sinking"),
+                                       cl::init(true));
 
 static cl::opt<unsigned> MaxSinkNumUsers(
     "instcombine-max-sink-users", cl::init(32),
@@ -156,6 +158,8 @@ extern cl::opt<bool> ProfcheckDisableMetadataFixes;
 static cl::opt<unsigned> ShouldLowerDbgDeclare("instcombine-lower-dbg-declare",
                                                cl::Hidden, cl::init(true));
 
+} // end namespace llvm
+
 std::optional<Instruction *>
 InstCombiner::targetInstCombineIntrinsic(IntrinsicInst &II) {
   // Handle target specific intrinsics