[llvm] [LV][EVL] Introduce MergeUntilPivot VPInstruction to enable out-loop reduction in EVL vectorization. (PR #101641)

[Mel Chen via llvm-commits]

================
@@ -1481,7 +1486,8 @@ bool VPlanTransforms::tryAddExplicitVectorLength(VPlan &Plan) {
   NextEVLIV->insertBefore(CanonicalIVIncrement);
   EVLPhi->addOperand(NextEVLIV);
 
-  transformRecipestoEVLRecipes(Plan, *VPEVL);
+  LLVMContext &Ctx = CanonicalIVPHI->getScalarType()->getContext();
+  transformRecipestoEVLRecipes(Plan, *VPEVL, Ctx);
----------------
Mel-Chen wrote:

e124f9f1c9157b8210fb617f30c5ca2280306c49

https://github.com/llvm/llvm-project/pull/101641