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

[Mel Chen via llvm-commits]

https://github.com/Mel-Chen updated https://github.com/llvm/llvm-project/pull/152479

>From eb09e4e622f97f77d7a78e71af9625988dd9036a Mon Sep 17 00:00:00 2001
From: Mel Chen <mel.chen at sifive.com>
Date: Thu, 7 Aug 2025 03:29:28 -0700
Subject: [PATCH 1/2] nfc, unified evl mask

---
 .../Transforms/Vectorize/VPlanTransforms.cpp  | 73 +++++++++----------
 1 file changed, 36 insertions(+), 37 deletions(-)

diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
index 34b2abf449ece..221dcbb94913d 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
@@ -2242,48 +2242,47 @@ 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);
   for (VPValue *HeaderMask : collectAllHeaderMasks(Plan)) {
-    // TODO: Split optimizeMaskToEVL out and move into
-    // VPlanTransforms::optimize. transformRecipestoEVLRecipes should be run in
-    // tryToBuildVPlanWithVPRecipes beforehand.
-    for (VPUser *U : collectUsersRecursively(HeaderMask)) {
-      auto *CurRecipe = cast<VPRecipeBase>(U);
-      VPRecipeBase *EVLRecipe =
-          optimizeMaskToEVL(HeaderMask, *CurRecipe, TypeInfo, *AllOneMask, EVL);
-      if (!EVLRecipe)
-        continue;
-
-      [[maybe_unused]] unsigned NumDefVal = EVLRecipe->getNumDefinedValues();
-      assert(NumDefVal == CurRecipe->getNumDefinedValues() &&
-             "New recipe must define the same number of values as the "
-             "original.");
-      assert(
-          NumDefVal <= 1 &&
-          "Only supports recipes with a single definition or without users.");
-      EVLRecipe->insertBefore(CurRecipe);
-      if (isa<VPSingleDefRecipe, VPWidenLoadEVLRecipe>(EVLRecipe)) {
-        VPValue *CurVPV = CurRecipe->getVPSingleValue();
-        CurVPV->replaceAllUsesWith(EVLRecipe->getVPSingleValue());
-      }
-      ToErase.push_back(CurRecipe);
-    }
-
-    // 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);
     HeaderMask->replaceAllUsesWith(EVLMask);
     ToErase.push_back(HeaderMask->getDefiningRecipe());
   }
 
+  // Try to optimize header mask recipes away to their EVL variants.
+  // TODO: Split optimizeMaskToEVL out and move into
+  // VPlanTransforms::optimize. transformRecipestoEVLRecipes should be run in
+  // tryToBuildVPlanWithVPRecipes beforehand.
+  for (VPUser *U : collectUsersRecursively(EVLMask)) {
+    auto *CurRecipe = cast<VPRecipeBase>(U);
+    VPRecipeBase *EVLRecipe =
+        optimizeMaskToEVL(EVLMask, *CurRecipe, TypeInfo, *AllOneMask, EVL);
+    if (!EVLRecipe)
+      continue;
+
+    [[maybe_unused]] unsigned NumDefVal = EVLRecipe->getNumDefinedValues();
+    assert(NumDefVal == CurRecipe->getNumDefinedValues() &&
+           "New recipe must define the same number of values as the "
+           "original.");
+    assert(NumDefVal <= 1 &&
+           "Only supports recipes with a single definition or without users.");
+    EVLRecipe->insertBefore(CurRecipe);
+    if (isa<VPSingleDefRecipe, VPWidenLoadEVLRecipe>(EVLRecipe)) {
+      VPValue *CurVPV = CurRecipe->getVPSingleValue();
+      CurVPV->replaceAllUsesWith(EVLRecipe->getVPSingleValue());
+    }
+    ToErase.push_back(CurRecipe);
+  }
+
   for (VPRecipeBase *R : reverse(ToErase)) {
     SmallVector<VPValue *> PossiblyDead(R->operands());
     R->eraseFromParent();

>From 53f0022824a69fb9dd98057effdeea989463c2dd Mon Sep 17 00:00:00 2001
From: Mel Chen <mel.chen at sifive.com>
Date: Fri, 8 Aug 2025 05:00:50 -0700
Subject: [PATCH 2/2] remove dead mask

---
 llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
index 221dcbb94913d..c4a6994276274 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
@@ -2282,6 +2282,9 @@ static void transformRecipestoEVLRecipes(VPlan &Plan, VPValue &EVL) {
     }
     ToErase.push_back(CurRecipe);
   }
+  // Remove dead EVL mask.
+  if (EVLMask->getNumUsers() == 0)
+    EVLMask->getDefiningRecipe()->eraseFromParent();
 
   for (VPRecipeBase *R : reverse(ToErase)) {
     SmallVector<VPValue *> PossiblyDead(R->operands());