[llvm] [LV][EVL] Support in-loop reduction using tail folding with EVL. (PR #90184)

[Florian Hahn via llvm-commits]

================
@@ -1446,23 +1454,39 @@ bool VPlanTransforms::tryAddExplicitVectorLength(VPlan &Plan) {
 
   for (VPValue *HeaderMask : collectAllHeaderMasks(Plan)) {
     for (VPUser *U : collectUsersRecursively(HeaderMask)) {
-      auto *MemR = dyn_cast<VPWidenMemoryRecipe>(U);
-      if (!MemR)
+      VPRecipeBase *NewRecipe = nullptr;
+      auto *CurRecipe = dyn_cast<VPRecipeBase>(U);
+      if (!CurRecipe || CurRecipe->getNumDefinedValues() > 1)
----------------
fhahn wrote:

Right, but why does it matter? Any recipe that isn't handled below will be skipped so there should be no need to check # of defined values unless I am missing something?


https://github.com/llvm/llvm-project/pull/90184