[PATCH] D107372: [InstCombine] Fixed select + masked load fold failure

[Dylan Fleming via Phabricator via llvm-commits]

This revision was automatically updated to reflect the committed changes.
Closed by commit rG3943a74666cb: [InstCombine] Fixed select + masked load fold failure (authored by DylanFleming-arm).

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D107372/new/

https://reviews.llvm.org/D107372

Files:
  llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
  llvm/test/Transforms/InstCombine/select-masked_load.ll


Index: llvm/test/Transforms/InstCombine/select-masked_load.ll
===================================================================
--- llvm/test/Transforms/InstCombine/select-masked_load.ll
+++ llvm/test/Transforms/InstCombine/select-masked_load.ll
@@ -94,5 +94,18 @@
   ret <4 x float> %masked
 }
 
+define <8 x float> @masked_load_and_scalar_select_cond(<8 x float>* %ptr, <8 x i1> %mask, i1 %cond) {
+; CHECK-LABEL: @masked_load_and_scalar_select_cond(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* [[PTR:%.*]], i32 32, <8 x i1> [[MASK:%.*]], <8 x float> undef)
+; CHECK-NEXT:    [[TMP1:%.*]] = select i1 [[COND:%.*]], <8 x float> zeroinitializer, <8 x float> [[TMP0]]
+; CHECK-NEXT:    ret <8 x float> [[TMP1]]
+entry:
+  %0 = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* %ptr, i32 32, <8 x i1> %mask, <8 x float> undef)
+  %1 = select i1 %cond, <8 x float> zeroinitializer, <8 x float> %0
+  ret <8 x float> %1
+}
+
+declare <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>*, i32 immarg, <8 x i1>, <8 x float>)
 declare <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>*, i32 immarg, <4 x i1>, <4 x i32>)
 declare <4 x float> @llvm.masked.load.v4f32.p0v4f32(<4 x float>*, i32 immarg, <4 x i1>, <4 x float>)
Index: llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
===================================================================
--- llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
+++ llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
@@ -3230,7 +3230,8 @@
   Value *Mask;
   if (match(TrueVal, m_Zero()) &&
       match(FalseVal, m_MaskedLoad(m_Value(), m_Value(), m_Value(Mask),
-                                   m_CombineOr(m_Undef(), m_Zero())))) {
+                                   m_CombineOr(m_Undef(), m_Zero()))) &&
+      (CondVal->getType() == Mask->getType())) {
     // We can remove the select by ensuring the load zeros all lanes the
     // select would have.  We determine this by proving there is no overlap
     // between the load and select masks.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D107372.363810.patch
Type: text/x-patch
Size: 2092 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20210803/14173d31/attachment.bin>