[SLP] Fixes crash in BoUpSLP::isGatherShuffledEntry()

[vporpo]

Summary

Crash caused by: 708eb1b96d9a36f9c0182b7d53c492059778fa35

[ABataev]

llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp

8410–8411

Better to replace TE->Scalars[I] with VL[I]

[vporpo]

Replace TE->Scalars with VL, as suggested.

[ABataev]

LG, thanks!

llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp

if (UsedTEs.size() == 1) {
     // Try to find the perfect match in another gather node at first.
     auto *It = find_if(FirstEntries, [=](const TreeEntry *EntryPtr) {
       return EntryPtr->isSame(VL) || EntryPtr->isSame(TE->Scalars);
     });
     if (It != FirstEntries.end()) {
       Entries.push_back(*It);
       std::iota(Mask.begin(), Mask.end(), 0);
       // Clear undef scalars.
-      for (int I = 0, Sz = VL.size(); I < Sz; ++I)
+      for (int I = 0, Sz = TE->Scalars.size(); I < Sz; ++I)
         if (isa<PoisonValue>(TE->Scalars[I]))
           Mask[I] = UndefMaskElem;
       return TargetTransformInfo::SK_PermuteSingleSrc;
     }
     // No perfect match, just shuffle, so choose the first tree node from the
     // tree.
     Entries.push_back(FirstEntries.front());
   } else {
     // Try to find nodes with the same vector factor.

llvm/test/Transforms/SLPVectorizer/X86/crash_clear_undefs.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -passes=slp-vectorizer -S -mtriple=x86_64-- -mcpu=corei7 | FileCheck %s
+target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
+
+define i1 @foo() {
+; CHECK-LABEL: @foo(
+; CHECK-NEXT:    [[TMP1:%.*]] = load float, ptr null, align 4
+; CHECK-NEXT:    br i1 false, label [[TMP15:%.*]], label [[TMP2:%.*]]
+; CHECK:       2:
+; CHECK-NEXT:    [[TMP3:%.*]] = extractelement <4 x float> zeroinitializer, i64 0
+; CHECK-NEXT:    [[TMP4:%.*]] = insertelement <2 x float> poison, float [[TMP1]], i32 0
+; CHECK-NEXT:    [[TMP5:%.*]] = insertelement <2 x float> [[TMP4]], float [[TMP3]], i32 1
+; CHECK-NEXT:    [[TMP6:%.*]] = fadd <2 x float> [[TMP5]], zeroinitializer
+; CHECK-NEXT:    [[TMP7:%.*]] = shufflevector <2 x float> [[TMP6]], <2 x float> poison, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
+; CHECK-NEXT:    [[TMP8:%.*]] = insertelement <4 x float> poison, float [[TMP1]], i32 0
+; CHECK-NEXT:    [[TMP9:%.*]] = insertelement <4 x float> [[TMP8]], float [[TMP3]], i32 1
+; CHECK-NEXT:    [[TMP10:%.*]] = shufflevector <4 x float> [[TMP9]], <4 x float> poison, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
+; CHECK-NEXT:    [[TMP11:%.*]] = select <4 x i1> zeroinitializer, <4 x float> [[TMP7]], <4 x float> [[TMP10]]
+; CHECK-NEXT:    [[TMP12:%.*]] = fsub <4 x float> [[TMP11]], zeroinitializer
+; CHECK-NEXT:    [[TMP13:%.*]] = fadd <4 x float> [[TMP11]], zeroinitializer
+; CHECK-NEXT:    [[TMP14:%.*]] = shufflevector <4 x float> [[TMP12]], <4 x float> [[TMP13]], <4 x i32> <i32 0, i32 1, i32 6, i32 7>
+; CHECK-NEXT:    br label [[TMP15]]
+; CHECK:       15:
+; CHECK-NEXT:    [[TMP16:%.*]] = phi <4 x float> [ [[TMP14]], [[TMP2]] ], [ zeroinitializer, [[TMP0:%.*]] ]
+; CHECK-NEXT:    ret i1 false
+;
+  %1 = load float, ptr null, align 4
+  br i1 false, label %14, label %2
+
+2:                                                ; preds = %0
+  %3 = extractelement <4 x float> zeroinitializer, i64 0
+  %4 = fadd float %1, 0.000000e+00
+  %5 = fadd float %3, 0.000000e+00
+  %6 = select i1 false, float %4, float %1
+  %7 = fsub float %6, 0.000000e+00
+  %8 = select i1 false, float %5, float %3
+  %9 = fsub float %8, 0.000000e+00
+  %10 = select i1 false, float %4, float %1
+  %11 = fadd float 0.000000e+00, %10
+  %12 = select i1 false, float %5, float %3
+  %13 = fadd float 0.000000e+00, %12
+  br label %14
+
+14:                                               ; preds = %2, %0
+  %15 = phi float [ %7, %2 ], [ 0.000000e+00, %0 ]
+  %16 = phi float [ %9, %2 ], [ 0.000000e+00, %0 ]
+  %17 = phi float [ %11, %2 ], [ 0.000000e+00, %0 ]
+  %18 = phi float [ %13, %2 ], [ 0.000000e+00, %0 ]
+  ret i1 false
+}