[llvm] [LV] Explicitly disallow interleaved access requiring gap mask for scalable VFs. nfc (PR #154122)

[via llvm-commits]

llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-llvm-transforms

Author: Mel Chen (Mel-Chen)

<details>
<summary>Changes</summary>

Currently, VPInterleaveRecipe::execute does not support generating LLVM IR for interleaved accesses that require a gap mask with scalable VFs.
It would be better to detect and prevent such groups from being vectorized as interleaved accesses in LoopVectorizationCostModel::interleavedAccessCanBeWidened, rather than relying on the TTI function getInterleavedMemoryOpCost to return an invalid cost.

---
Full diff: https://github.com/llvm/llvm-project/pull/154122.diff


2 Files Affected:

- (modified) llvm/lib/Transforms/Vectorize/LoopVectorize.cpp (+6) 
- (modified) llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp (+2-2) 


``````````diff
diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
index 675a230bd2c94..1c08665652e40 100644
--- a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -3095,6 +3095,12 @@ bool LoopVectorizationCostModel::interleavedAccessCanBeWidened(
   if (Group->isReverse())
     return false;
 
+  // TODO: Support interleaved access that requires a gap mask for scalable VFs.
+  bool NeedsMaskForGaps = LoadAccessWithGapsRequiresEpilogMasking ||
+                          StoreAccessWithGapsRequiresMasking;
+  if (VF.isScalable() && NeedsMaskForGaps)
+    return false;
+
   auto *Ty = getLoadStoreType(I);
   const Align Alignment = getLoadStoreAlignment(I);
   unsigned AS = getLoadStoreAddressSpace(I);
diff --git a/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp b/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
index 0609510ac8212..2fe0e042c538f 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
@@ -3476,6 +3476,8 @@ static Value *interleaveVectors(IRBuilderBase &Builder, ArrayRef<Value *> Vals,
 //   store <12 x i32> %interleaved.vec              ; Write 4 tuples of R,G,B
 void VPInterleaveRecipe::execute(VPTransformState &State) {
   assert(!State.Lane && "Interleave group being replicated.");
+  assert((!NeedsMaskForGaps || !State.VF.isScalable()) &&
+         "Masking gaps for scalable vectors is not yet supported.");
   const InterleaveGroup<Instruction> *Group = IG;
   Instruction *Instr = Group->getInsertPos();
 
@@ -3593,8 +3595,6 @@ void VPInterleaveRecipe::execute(VPTransformState &State) {
       createBitMaskForGaps(State.Builder, State.VF.getKnownMinValue(), *Group);
   assert(((MaskForGaps != nullptr) == NeedsMaskForGaps) &&
          "Mismatch between NeedsMaskForGaps and MaskForGaps");
-  assert((!MaskForGaps || !State.VF.isScalable()) &&
-         "masking gaps for scalable vectors is not yet supported.");
   ArrayRef<VPValue *> StoredValues = getStoredValues();
   // Collect the stored vector from each member.
   SmallVector<Value *, 4> StoredVecs;

``````````

</details>


https://github.com/llvm/llvm-project/pull/154122