diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
--- a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -3914,7 +3914,7 @@
   // a Select choosing between the vectorized LoopExitInst and vectorized Phi,
   // instead of the former. For an inloop reduction the reduction will already
   // be predicated, and does not need to be handled here.
-  if (Cost->foldTailByMasking() && !PhiR->isInLoop()) {
+  if (PhiR->getParent()->getPlan()->hasTailFolded() && !PhiR->isInLoop()) {
     for (unsigned Part = 0; Part < UF; ++Part) {
       Value *VecLoopExitInst = State.get(LoopExitInstDef, Part);
       SelectInst *Sel = nullptr;
@@ -9256,7 +9256,7 @@
   // If tail is folded by masking, introduce selects between the phi
   // and the live-out instruction of each reduction, at the beginning of the
   // dedicated latch block.
-  if (CM.foldTailByMasking()) {
+  if (Plan->hasTailFolded()) {
     Builder.setInsertPoint(LatchVPBB, LatchVPBB->begin());
     for (VPRecipeBase &R :
          Plan->getVectorLoopRegion()->getEntryBasicBlock()->phis()) {
diff --git a/llvm/lib/Transforms/Vectorize/VPlan.h b/llvm/lib/Transforms/Vectorize/VPlan.h
--- a/llvm/lib/Transforms/Vectorize/VPlan.h
+++ b/llvm/lib/Transforms/Vectorize/VPlan.h
@@ -2653,6 +2653,9 @@
   VPBasicBlock *getPreheader() { return Preheader; }
   const VPBasicBlock *getPreheader() const { return Preheader; }
 
+  /// Returns true if the tail loop is folded in the VPlan.
+  bool hasTailFolded();
+
 private:
   /// Add to the given dominator tree the header block and every new basic block
   /// that was created between it and the latch block, inclusive.
diff --git a/llvm/lib/Transforms/Vectorize/VPlan.cpp b/llvm/lib/Transforms/Vectorize/VPlan.cpp
--- a/llvm/lib/Transforms/Vectorize/VPlan.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlan.cpp
@@ -845,6 +845,32 @@
   LiveOuts.insert({PN, new VPLiveOut(PN, V)});
 }
 
+bool VPlan::hasTailFolded() {
+  VPBasicBlock *HeaderVPBB = getVectorLoopRegion()->getEntryBasicBlock();
+  return any_of(*HeaderVPBB, [this](VPRecipeBase &R) {
+    // A widened canonical IV or active-lane-mask PHI imply the tail being
+    // folded.
+    if (isa<VPWidenCanonicalIVRecipe>(&R) || isa<VPActiveLaneMaskPHIRecipe>(&R))
+      return true;
+
+    auto *VPI = dyn_cast<VPInstruction>(&R);
+    if (!VPI)
+      return false;
+    if (VPI->getOpcode() == VPInstruction::ActiveLaneMask)
+      return true;
+    if (VPI->getOpcode() != VPInstruction::ICmpULE)
+      return false;
+    auto IsWidenCanonicalIV = [](const VPRecipeBase &R) {
+      if (auto *Ind = dyn_cast<VPWidenIntOrFpInductionRecipe>(&R))
+        return Ind->isCanonical();
+      return isa<VPWidenCanonicalIVRecipe>(&R);
+    };
+    if (IsWidenCanonicalIV(*VPI->getOperand(0)->getDefiningRecipe()) &&
+        VPI->getOperand(1) == getOrCreateBackedgeTakenCount())
+      return true;
+  });
+}
+
 void VPlan::updateDominatorTree(DominatorTree *DT, BasicBlock *LoopHeaderBB,
                                 BasicBlock *LoopLatchBB,
                                 BasicBlock *LoopExitBB) {