[llvm] [LV][EVL] Simplify EVL recipe transformation by using a single EVL mask. nfc (PR #152479)

[Luke Lau via llvm-commits]

================
@@ -2227,48 +2227,47 @@ static void transformRecipestoEVLRecipes(VPlan &Plan, VPValue &EVL) {
     }
   }
 
-  // Try to optimize header mask recipes away to their EVL variants.
+  // Replace header masks with a mask equivalent to predicating by EVL:
+  //
+  // icmp ule widen-canonical-iv backedge-taken-count
+  // ->
+  // icmp ult step-vector, EVL
+  VPRecipeBase *EVLR = EVL.getDefiningRecipe();
+  VPBuilder Builder(EVLR->getParent(), std::next(EVLR->getIterator()));
+  Type *EVLType = TypeInfo.inferScalarType(&EVL);
+  VPValue *EVLMask = Builder.createICmp(
+      CmpInst::ICMP_ULT,
+      Builder.createNaryOp(VPInstruction::StepVector, {}, EVLType), &EVL);
----------------
lukel97 wrote:

Given that this runs after optimization maybe it's possible it might get optimized away? In that case we would end up with an unused EVLMask.

https://github.com/llvm/llvm-project/pull/152479