[X86] combineExtractWithShuffle - more aggressively peek through bitcasts

Fixes regression introduced by rL352743

llvm-svn: 352745

Author: RKSimon

llvm/lib/Target/X86/X86ISelLowering.cpp

@@ -33861,15 +33861,19 @@ static SDValue combineExtractWithShuffle(SDNode *N, SelectionDAG &DAG,
   if (SrcSVT == MVT::i1 || !isa<ConstantSDNode>(Idx))
     return SDValue();
 
+  SDValue SrcBC = peekThroughBitcasts(Src);
+
   // Handle extract(broadcast(scalar_value)), it doesn't matter what index is.
-  if (X86ISD::VBROADCAST == Src.getOpcode() &&
-      Src.getOperand(0).getValueType() == VT)
-    return Src.getOperand(0);
+  if (X86ISD::VBROADCAST == SrcBC.getOpcode()) {
+    SDValue SrcOp = SrcBC.getOperand(0);
+    if (SrcOp.getValueSizeInBits() == VT.getSizeInBits())
+      return DAG.getBitcast(VT, SrcOp);
+  }
 
   // Resolve the target shuffle inputs and mask.
   SmallVector<int, 16> Mask;
   SmallVector<SDValue, 2> Ops;
-  if (!resolveTargetShuffleInputs(peekThroughBitcasts(Src), Ops, Mask, DAG))
+  if (!resolveTargetShuffleInputs(SrcBC, Ops, Mask, DAG))
     return SDValue();
 
   // Attempt to narrow/widen the shuffle mask to the correct size.

llvm/test/CodeGen/X86/extractelement-load.ll

@@ -1,7 +1,7 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -mtriple=i686-unknown -mattr=+sse2 | FileCheck %s --check-prefix=X32-SSE2
-; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+ssse3 | FileCheck %s --check-prefix=X64-SSSE3
-; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+avx | FileCheck %s --check-prefix=X64-AVX
+; RUN: llc < %s -mtriple=i686-unknown -mattr=+sse2 | FileCheck %s --check-prefixes=X32-SSE2
+; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+ssse3 | FileCheck %s --check-prefixes=X64,X64-SSSE3
+; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+avx | FileCheck %s --check-prefixes=X64,X64-AVX
 
 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
 
@@ -12,15 +12,10 @@ define i32 @t(<2 x i64>* %val) nounwind  {
 ; X32-SSE2-NEXT:    movl 8(%eax), %eax
 ; X32-SSE2-NEXT:    retl
 ;
-; X64-SSSE3-LABEL: t:
-; X64-SSSE3:       # %bb.0:
-; X64-SSSE3-NEXT:    movl 8(%rdi), %eax
-; X64-SSSE3-NEXT:    retq
-;
-; X64-AVX-LABEL: t:
-; X64-AVX:       # %bb.0:
-; X64-AVX-NEXT:    movl 8(%rdi), %eax
-; X64-AVX-NEXT:    retq
+; X64-LABEL: t:
+; X64:       # %bb.0:
+; X64-NEXT:    movl 8(%rdi), %eax
+; X64-NEXT:    retq
   %tmp2 = load <2 x i64>, <2 x i64>* %val, align 16		; <<2 x i64>> [#uses=1]
   %tmp3 = bitcast <2 x i64> %tmp2 to <4 x i32>		; <<4 x i32>> [#uses=1]
   %tmp4 = extractelement <4 x i32> %tmp3, i32 2		; <i32> [#uses=1]
@@ -34,13 +29,9 @@ define i32 @t2(<8 x i32>* %xp) {
 ; X32-SSE2:       # %bb.0:
 ; X32-SSE2-NEXT:    retl
 ;
-; X64-SSSE3-LABEL: t2:
-; X64-SSSE3:       # %bb.0:
-; X64-SSSE3-NEXT:    retq
-;
-; X64-AVX-LABEL: t2:
-; X64-AVX:       # %bb.0:
-; X64-AVX-NEXT:    retq
+; X64-LABEL: t2:
+; X64:       # %bb.0:
+; X64-NEXT:    retq
   %x = load <8 x i32>, <8 x i32>* %xp
   %Shuff68 = shufflevector <8 x i32> %x, <8 x i32> undef, <8 x i32> <i32 undef, i32 7, i32 9, i32 undef, i32 13, i32 15, i32 1, i32 3>
   %y = extractelement <8 x i32> %Shuff68, i32 0
@@ -91,16 +82,10 @@ define i64 @t4(<2 x double>* %a) {
 ; X32-SSE2-NEXT:    movd %xmm0, %edx
 ; X32-SSE2-NEXT:    retl
 ;
-; X64-SSSE3-LABEL: t4:
-; X64-SSSE3:       # %bb.0:
-; X64-SSSE3-NEXT:    movq (%rdi), %rax
-; X64-SSSE3-NEXT:    retq
-;
-; X64-AVX-LABEL: t4:
-; X64-AVX:       # %bb.0:
-; X64-AVX-NEXT:    vmovddup {{.*#+}} xmm0 = mem[0,0]
-; X64-AVX-NEXT:    vpextrq $1, %xmm0, %rax
-; X64-AVX-NEXT:    retq
+; X64-LABEL: t4:
+; X64:       # %bb.0:
+; X64-NEXT:    movq (%rdi), %rax
+; X64-NEXT:    retq
   %b = load <2 x double>, <2 x double>* %a, align 16
   %c = shufflevector <2 x double> %b, <2 x double> %b, <2 x i32> <i32 1, i32 0>
   %d = bitcast <2 x double> %c to <2 x i64>