[llvm] 79682c4 - [SLP]Check if the buildvector root is not a part of the graph before deletion

[Alexey Bataev via llvm-commits]

Author: Alexey Bataev
Date: 2024-11-19T11:31:40-08:00
New Revision: 79682c4d57620e623fb30271cc8003d0c9e14a01

URL: https://github.com/llvm/llvm-project/commit/79682c4d57620e623fb30271cc8003d0c9e14a01
DIFF: https://github.com/llvm/llvm-project/commit/79682c4d57620e623fb30271cc8003d0c9e14a01.diff

LOG: [SLP]Check if the buildvector root is not a part of the graph before deletion

If the buildvector root has no uses, it might be still needed as a part
of the graph, so need to check that it is not a part of the graph before
deletion.

Fixes #116852

Added: 
    llvm/test/Transforms/SLPVectorizer/X86/bv-root-part-of-graph.ll

Modified: 
    llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
index dc0dffd9fcbf81..e70627b6afc10d 100644
--- a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
+++ b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
@@ -13614,7 +13614,10 @@ Value *BoUpSLP::gather(
     } else {
       Vec = CreateShuffle(Root, Vec, Mask);
       if (auto *OI = dyn_cast<Instruction>(OriginalRoot);
-          OI && OI->hasNUses(0))
+          OI && OI->hasNUses(0) &&
+          none_of(VectorizableTree, [&](const std::unique_ptr<TreeEntry> &TE) {
+            return TE->VectorizedValue == OI;
+          }))
         eraseInstruction(OI);
     }
   }

diff  --git a/llvm/test/Transforms/SLPVectorizer/X86/bv-root-part-of-graph.ll b/llvm/test/Transforms/SLPVectorizer/X86/bv-root-part-of-graph.ll
new file mode 100644
index 00000000000000..0cc4d3db5c5377
--- /dev/null
+++ b/llvm/test/Transforms/SLPVectorizer/X86/bv-root-part-of-graph.ll
@@ -0,0 +1,55 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
+; RUN: opt -S --passes=slp-vectorizer -mtriple=x86_64-unknown-linux-gnu < %s | FileCheck %s
+
+define void @test() {
+; CHECK-LABEL: define void @test() {
+; CHECK-NEXT:  [[BB:.*]]:
+; CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <4 x float> <float 0.000000e+00, float undef, float 0.000000e+00, float 0.000000e+00>, <4 x float> poison, <4 x i32> <i32 0, i32 0, i32 2, i32 3>
+; CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <4 x float> <float 0.000000e+00, float undef, float 0.000000e+00, float 0.000000e+00>, <4 x float> <float poison, float 0.000000e+00, float poison, float poison>, <4 x i32> <i32 0, i32 5, i32 2, i32 3>
+; CHECK-NEXT:    br label %[[BB1:.*]]
+; CHECK:       [[BB1]]:
+; CHECK-NEXT:    [[PHI:%.*]] = phi i32 [ 0, %[[BB]] ], [ [[TMP9:%.*]], %[[BB1]] ]
+; CHECK-NEXT:    [[FMUL:%.*]] = fmul float 0.000000e+00, 0.000000e+00
+; CHECK-NEXT:    [[TMP2:%.*]] = insertelement <4 x float> [[TMP1]], float [[FMUL]], i32 2
+; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <4 x float> [[TMP2]], <4 x float> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 2>
+; CHECK-NEXT:    [[TMP4:%.*]] = fadd <4 x float> [[TMP0]], [[TMP3]]
+; CHECK-NEXT:    [[TMP5:%.*]] = fadd <4 x float> [[TMP4]], zeroinitializer
+; CHECK-NEXT:    [[TMP6:%.*]] = fcmp ogt <4 x float> [[TMP5]], zeroinitializer
+; CHECK-NEXT:    [[TMP7:%.*]] = select <4 x i1> [[TMP6]], <4 x i32> zeroinitializer, <4 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP8:%.*]] = select <4 x i1> zeroinitializer, <4 x i32> [[TMP7]], <4 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP9]] = call i32 @llvm.vector.reduce.or.v4i32(<4 x i32> [[TMP8]])
+; CHECK-NEXT:    br label %[[BB1]]
+;
+bb:
+  br label %bb1
+
+bb1:
+  %phi = phi i32 [ 0, %bb ], [ %or21, %bb1 ]
+  %sitofp = sitofp i32 0 to float
+  %fadd = fadd float %sitofp, %sitofp
+  %fadd2 = fadd float %fadd, 0.000000e+00
+  %fcmp = fcmp ogt float %fadd2, 0.000000e+00
+  %select = select i1 %fcmp, i32 0, i32 0
+  %select3 = select i1 false, i32 %select, i32 0
+  %fadd4 = fadd float %sitofp, 0.000000e+00
+  %fadd5 = fadd float %fadd4, 0.000000e+00
+  %fcmp6 = fcmp ogt float %fadd5, 0.000000e+00
+  %select7 = select i1 %fcmp6, i32 0, i32 0
+  %select8 = select i1 false, i32 %select7, i32 0
+  %or = or i32 %select3, %select8
+  %sitofp9 = sitofp i32 0 to float
+  %fmul = fmul float 0.000000e+00, 0.000000e+00
+  %fadd10 = fadd float %sitofp9, %fmul
+  %fadd11 = fadd float %fadd10, 0.000000e+00
+  %fcmp12 = fcmp ogt float %fadd11, 0.000000e+00
+  %select13 = select i1 %fcmp12, i32 0, i32 0
+  %select14 = select i1 false, i32 %select13, i32 0
+  %or15 = or i32 %select14, %or
+  %fadd16 = fadd float %fmul, 0.000000e+00
+  %fadd17 = fadd float %fadd16, 0.000000e+00
+  %fcmp18 = fcmp ogt float %fadd17, 0.000000e+00
+  %select19 = select i1 %fcmp18, i32 0, i32 0
+  %select20 = select i1 false, i32 %select19, i32 0
+  %or21 = or i32 %or15, %select20
+  br label %bb1
+}