[DAGCombine] Allow vector constant folding of any value type before type legalization

The patch comes in 2 parts:

1 - it makes use of the SelectionDAG::NewNodesMustHaveLegalTypes flag to tell when it can safely constant fold illegal types.

2 - it correctly resets SelectionDAG::NewNodesMustHaveLegalTypes at the start of each call to SelectionDAGISel::CodeGenAndEmitDAG so all the pre-legalization stages can make use of it - not just the first basic block that gets handled.

Fix for PR30760

Differential Revision: https://reviews.llvm.org/D29568

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294749 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/CodeGen/SelectionDAG/SelectionDAG.cpp b/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
index 9e802b99ed4a838d855b7c6943a7932d93a5f39f..16425a082004ce0712b795cca1de93ff8181e70d 100644 (file)
--- a/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
+++ b/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
@@ -3762,7 +3762,7 @@ SDValue SelectionDAG::FoldConstantVectorArithmetic(unsigned Opcode,
   // Find legal integer scalar type for constant promotion and
   // ensure that its scalar size is at least as large as source.
   EVT LegalSVT = VT.getScalarType();
-  if (LegalSVT.isInteger()) {
+  if (NewNodesMustHaveLegalTypes && LegalSVT.isInteger()) {
     LegalSVT = TLI->getTypeToTransformTo(*getContext(), LegalSVT);
     if (LegalSVT.bitsLT(VT.getScalarType()))
       return SDValue();

diff --git a/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp b/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
index 25646358d96ca72ccfa2db4863937e93340f73b8..989c33e0fcba72c22d53b8e23636886230b0d391 100644 (file)
--- a/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
+++ b/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
@@ -763,6 +763,10 @@ void SelectionDAGISel::CodeGenAndEmitDAG() {
   int BlockNumber = -1;
   (void)BlockNumber;
   bool MatchFilterBB = false; (void)MatchFilterBB;
+
+  // Pre-type legalization allow creation of any node types.
+  CurDAG->NewNodesMustHaveLegalTypes = false;
+
 #ifndef NDEBUG
   MatchFilterBB = (FilterDAGBasicBlockName.empty() ||
                    FilterDAGBasicBlockName ==
@@ -809,6 +813,7 @@ void SelectionDAGISel::CodeGenAndEmitDAG() {
   DEBUG(dbgs() << "Type-legalized selection DAG: BB#" << BlockNumber
         << " '" << BlockName << "'\n"; CurDAG->dump());
 
+  // Only allow creation of legal node types.
   CurDAG->NewNodesMustHaveLegalTypes = true;
 
   if (Changed) {

diff --git a/test/CodeGen/X86/vector-lzcnt-128.ll b/test/CodeGen/X86/vector-lzcnt-128.ll
index 6445a363787cce5dfd847463d8560628b857d94b..9e11edcc29dc5f3ebef0d3d0c85d0b4b068ed31a 100644 (file)
--- a/test/CodeGen/X86/vector-lzcnt-128.ll
+++ b/test/CodeGen/X86/vector-lzcnt-128.ll
@@ -1596,35 +1596,8 @@ define <2 x i64> @foldv2i64() nounwind {
 ;
 ; X32-SSE-LABEL: foldv2i64:
 ; X32-SSE:       # BB#0:
-; X32-SSE-NEXT:    movdqa {{.*#+}} xmm2 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
-; X32-SSE-NEXT:    movdqa {{.*#+}} xmm1 = [256,0,4294967295,4294967295]
-; X32-SSE-NEXT:    movdqa %xmm1, %xmm0
-; X32-SSE-NEXT:    pand %xmm2, %xmm0
-; X32-SSE-NEXT:    movdqa {{.*#+}} xmm3 = [4,3,2,2,1,1,1,1,0,0,0,0,0,0,0,0]
-; X32-SSE-NEXT:    movdqa %xmm3, %xmm4
-; X32-SSE-NEXT:    pshufb %xmm0, %xmm4
-; X32-SSE-NEXT:    movdqa %xmm1, %xmm0
-; X32-SSE-NEXT:    psrlw $4, %xmm0
-; X32-SSE-NEXT:    pand %xmm2, %xmm0
-; X32-SSE-NEXT:    pxor %xmm2, %xmm2
-; X32-SSE-NEXT:    pshufb %xmm0, %xmm3
-; X32-SSE-NEXT:    pcmpeqb %xmm2, %xmm0
-; X32-SSE-NEXT:    pand %xmm4, %xmm0
-; X32-SSE-NEXT:    paddb %xmm3, %xmm0
-; X32-SSE-NEXT:    movdqa %xmm1, %xmm3
-; X32-SSE-NEXT:    pcmpeqb %xmm2, %xmm3
-; X32-SSE-NEXT:    psrlw $8, %xmm3
-; X32-SSE-NEXT:    pand %xmm0, %xmm3
-; X32-SSE-NEXT:    psrlw $8, %xmm0
-; X32-SSE-NEXT:    paddw %xmm3, %xmm0
-; X32-SSE-NEXT:    pcmpeqw %xmm2, %xmm1
-; X32-SSE-NEXT:    psrld $16, %xmm1
-; X32-SSE-NEXT:    pand %xmm0, %xmm1
-; X32-SSE-NEXT:    psrld $16, %xmm0
-; X32-SSE-NEXT:    paddd %xmm1, %xmm0
-; X32-SSE-NEXT:    pblendw {{.*#+}} xmm2 = xmm0[0,1],xmm2[2,3,4,5,6,7]
-; X32-SSE-NEXT:    psrlq $32, %xmm0
-; X32-SSE-NEXT:    paddq %xmm2, %xmm0
+; X32-SSE-NEXT:    movl $55, %eax
+; X32-SSE-NEXT:    movd %eax, %xmm0
 ; X32-SSE-NEXT:    retl
   %out = call <2 x i64> @llvm.ctlz.v2i64(<2 x i64> <i64 256, i64 -1>, i1 0)
   ret <2 x i64> %out
@@ -1651,35 +1624,8 @@ define <2 x i64> @foldv2i64u() nounwind {
 ;
 ; X32-SSE-LABEL: foldv2i64u:
 ; X32-SSE:       # BB#0:
-; X32-SSE-NEXT:    movdqa {{.*#+}} xmm2 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
-; X32-SSE-NEXT:    movdqa {{.*#+}} xmm1 = [256,0,4294967295,4294967295]
-; X32-SSE-NEXT:    movdqa %xmm1, %xmm0
-; X32-SSE-NEXT:    pand %xmm2, %xmm0
-; X32-SSE-NEXT:    movdqa {{.*#+}} xmm3 = [4,3,2,2,1,1,1,1,0,0,0,0,0,0,0,0]
-; X32-SSE-NEXT:    movdqa %xmm3, %xmm4
-; X32-SSE-NEXT:    pshufb %xmm0, %xmm4
-; X32-SSE-NEXT:    movdqa %xmm1, %xmm0
-; X32-SSE-NEXT:    psrlw $4, %xmm0
-; X32-SSE-NEXT:    pand %xmm2, %xmm0
-; X32-SSE-NEXT:    pxor %xmm2, %xmm2
-; X32-SSE-NEXT:    pshufb %xmm0, %xmm3
-; X32-SSE-NEXT:    pcmpeqb %xmm2, %xmm0
-; X32-SSE-NEXT:    pand %xmm4, %xmm0
-; X32-SSE-NEXT:    paddb %xmm3, %xmm0
-; X32-SSE-NEXT:    movdqa %xmm1, %xmm3
-; X32-SSE-NEXT:    pcmpeqb %xmm2, %xmm3
-; X32-SSE-NEXT:    psrlw $8, %xmm3
-; X32-SSE-NEXT:    pand %xmm0, %xmm3
-; X32-SSE-NEXT:    psrlw $8, %xmm0
-; X32-SSE-NEXT:    paddw %xmm3, %xmm0
-; X32-SSE-NEXT:    pcmpeqw %xmm2, %xmm1
-; X32-SSE-NEXT:    psrld $16, %xmm1
-; X32-SSE-NEXT:    pand %xmm0, %xmm1
-; X32-SSE-NEXT:    psrld $16, %xmm0
-; X32-SSE-NEXT:    paddd %xmm1, %xmm0
-; X32-SSE-NEXT:    pblendw {{.*#+}} xmm2 = xmm0[0,1],xmm2[2,3,4,5,6,7]
-; X32-SSE-NEXT:    psrlq $32, %xmm0
-; X32-SSE-NEXT:    paddq %xmm2, %xmm0
+; X32-SSE-NEXT:    movl $55, %eax
+; X32-SSE-NEXT:    movd %eax, %xmm0
 ; X32-SSE-NEXT:    retl
   %out = call <2 x i64> @llvm.ctlz.v2i64(<2 x i64> <i64 256, i64 -1>, i1 -1)
   ret <2 x i64> %out

diff --git a/test/CodeGen/X86/vector-lzcnt-256.ll b/test/CodeGen/X86/vector-lzcnt-256.ll
index c683954930232e213656351df619cbcecd8b4271..04f74c9475918af3c5cbb00f5c28b32eb85bfb79 100644 (file)
--- a/test/CodeGen/X86/vector-lzcnt-256.ll
+++ b/test/CodeGen/X86/vector-lzcnt-256.ll
@@ -830,31 +830,7 @@ define <4 x i64> @foldv4i64() nounwind {
 ;
 ; X32-AVX-LABEL: foldv4i64:
 ; X32-AVX:       # BB#0:
-; X32-AVX-NEXT:    vmovdqa {{.*#+}} ymm0 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
-; X32-AVX-NEXT:    vmovdqa {{.*#+}} ymm1 = [256,0,4294967295,4294967295,0,0,255,0]
-; X32-AVX-NEXT:    vpand %ymm0, %ymm1, %ymm2
-; X32-AVX-NEXT:    vmovdqa {{.*#+}} ymm3 = [4,3,2,2,1,1,1,1,0,0,0,0,0,0,0,0,4,3,2,2,1,1,1,1,0,0,0,0,0,0,0,0]
-; X32-AVX-NEXT:    vpshufb %ymm2, %ymm3, %ymm2
-; X32-AVX-NEXT:    vpsrlw $4, %ymm1, %ymm4
-; X32-AVX-NEXT:    vpand %ymm0, %ymm4, %ymm0
-; X32-AVX-NEXT:    vpxor %ymm4, %ymm4, %ymm4
-; X32-AVX-NEXT:    vpcmpeqb %ymm4, %ymm0, %ymm5
-; X32-AVX-NEXT:    vpand %ymm5, %ymm2, %ymm2
-; X32-AVX-NEXT:    vpshufb %ymm0, %ymm3, %ymm0
-; X32-AVX-NEXT:    vpaddb %ymm0, %ymm2, %ymm0
-; X32-AVX-NEXT:    vpcmpeqb %ymm4, %ymm1, %ymm2
-; X32-AVX-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; X32-AVX-NEXT:    vpand %ymm2, %ymm0, %ymm2
-; X32-AVX-NEXT:    vpsrlw $8, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpaddw %ymm2, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpcmpeqw %ymm4, %ymm1, %ymm1
-; X32-AVX-NEXT:    vpsrld $16, %ymm1, %ymm1
-; X32-AVX-NEXT:    vpand %ymm1, %ymm0, %ymm1
-; X32-AVX-NEXT:    vpsrld $16, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpaddd %ymm1, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpblendd {{.*#+}} ymm1 = ymm0[0],ymm4[1,2,3],ymm0[4],ymm4[5],ymm0[6],ymm4[7]
-; X32-AVX-NEXT:    vpsrlq $32, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpaddq %ymm1, %ymm0, %ymm0
+; X32-AVX-NEXT:    vmovaps {{.*#+}} ymm0 = [55,0,0,0,64,0,56,0]
 ; X32-AVX-NEXT:    retl
   %out = call <4 x i64> @llvm.ctlz.v4i64(<4 x i64> <i64 256, i64 -1, i64 0, i64 255>, i1 0)
   ret <4 x i64> %out
@@ -873,31 +849,7 @@ define <4 x i64> @foldv4i64u() nounwind {
 ;
 ; X32-AVX-LABEL: foldv4i64u:
 ; X32-AVX:       # BB#0:
-; X32-AVX-NEXT:    vmovdqa {{.*#+}} ymm0 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
-; X32-AVX-NEXT:    vmovdqa {{.*#+}} ymm1 = [256,0,4294967295,4294967295,0,0,255,0]
-; X32-AVX-NEXT:    vpand %ymm0, %ymm1, %ymm2
-; X32-AVX-NEXT:    vmovdqa {{.*#+}} ymm3 = [4,3,2,2,1,1,1,1,0,0,0,0,0,0,0,0,4,3,2,2,1,1,1,1,0,0,0,0,0,0,0,0]
-; X32-AVX-NEXT:    vpshufb %ymm2, %ymm3, %ymm2
-; X32-AVX-NEXT:    vpsrlw $4, %ymm1, %ymm4
-; X32-AVX-NEXT:    vpand %ymm0, %ymm4, %ymm0
-; X32-AVX-NEXT:    vpxor %ymm4, %ymm4, %ymm4
-; X32-AVX-NEXT:    vpcmpeqb %ymm4, %ymm0, %ymm5
-; X32-AVX-NEXT:    vpand %ymm5, %ymm2, %ymm2
-; X32-AVX-NEXT:    vpshufb %ymm0, %ymm3, %ymm0
-; X32-AVX-NEXT:    vpaddb %ymm0, %ymm2, %ymm0
-; X32-AVX-NEXT:    vpcmpeqb %ymm4, %ymm1, %ymm2
-; X32-AVX-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; X32-AVX-NEXT:    vpand %ymm2, %ymm0, %ymm2
-; X32-AVX-NEXT:    vpsrlw $8, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpaddw %ymm2, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpcmpeqw %ymm4, %ymm1, %ymm1
-; X32-AVX-NEXT:    vpsrld $16, %ymm1, %ymm1
-; X32-AVX-NEXT:    vpand %ymm1, %ymm0, %ymm1
-; X32-AVX-NEXT:    vpsrld $16, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpaddd %ymm1, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpblendd {{.*#+}} ymm1 = ymm0[0],ymm4[1,2,3],ymm0[4],ymm4[5],ymm0[6],ymm4[7]
-; X32-AVX-NEXT:    vpsrlq $32, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpaddq %ymm1, %ymm0, %ymm0
+; X32-AVX-NEXT:    vmovaps {{.*#+}} ymm0 = [55,0,0,0,64,0,56,0]
 ; X32-AVX-NEXT:    retl
   %out = call <4 x i64> @llvm.ctlz.v4i64(<4 x i64> <i64 256, i64 -1, i64 0, i64 255>, i1 -1)
   ret <4 x i64> %out

diff --git a/test/CodeGen/X86/vector-tzcnt-128.ll b/test/CodeGen/X86/vector-tzcnt-128.ll
index bf32e672138ce0e9387b4cf6af13aa725a096212..3967b1d1c1b2a57648a3d2652345ed095929640d 100644 (file)
--- a/test/CodeGen/X86/vector-tzcnt-128.ll
+++ b/test/CodeGen/X86/vector-tzcnt-128.ll
@@ -1258,23 +1258,8 @@ define <2 x i64> @foldv2i64() nounwind {
 ;
 ; X32-SSE-LABEL: foldv2i64:
 ; X32-SSE:       # BB#0:
-; X32-SSE-NEXT:    movdqa {{.*#+}} xmm0 = [256,0,4294967295,4294967295]
-; X32-SSE-NEXT:    pxor %xmm1, %xmm1
-; X32-SSE-NEXT:    pxor %xmm2, %xmm2
-; X32-SSE-NEXT:    psubq %xmm0, %xmm2
-; X32-SSE-NEXT:    pand %xmm0, %xmm2
-; X32-SSE-NEXT:    psubq {{\.LCPI.*}}, %xmm2
-; X32-SSE-NEXT:    movdqa {{.*#+}} xmm3 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
-; X32-SSE-NEXT:    movdqa %xmm2, %xmm4
-; X32-SSE-NEXT:    pand %xmm3, %xmm4
-; X32-SSE-NEXT:    movdqa {{.*#+}} xmm0 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
-; X32-SSE-NEXT:    movdqa %xmm0, %xmm5
-; X32-SSE-NEXT:    pshufb %xmm4, %xmm5
-; X32-SSE-NEXT:    psrlw $4, %xmm2
-; X32-SSE-NEXT:    pand %xmm3, %xmm2
-; X32-SSE-NEXT:    pshufb %xmm2, %xmm0
-; X32-SSE-NEXT:    paddb %xmm5, %xmm0
-; X32-SSE-NEXT:    psadbw %xmm1, %xmm0
+; X32-SSE-NEXT:    movl $8, %eax
+; X32-SSE-NEXT:    movd %eax, %xmm0
 ; X32-SSE-NEXT:    retl
   %out = call <2 x i64> @llvm.cttz.v2i64(<2 x i64> <i64 256, i64 -1>, i1 0)
   ret <2 x i64> %out
@@ -1295,23 +1280,8 @@ define <2 x i64> @foldv2i64u() nounwind {
 ;
 ; X32-SSE-LABEL: foldv2i64u:
 ; X32-SSE:       # BB#0:
-; X32-SSE-NEXT:    movdqa {{.*#+}} xmm0 = [256,0,4294967295,4294967295]
-; X32-SSE-NEXT:    pxor %xmm1, %xmm1
-; X32-SSE-NEXT:    pxor %xmm2, %xmm2
-; X32-SSE-NEXT:    psubq %xmm0, %xmm2
-; X32-SSE-NEXT:    pand %xmm0, %xmm2
-; X32-SSE-NEXT:    psubq {{\.LCPI.*}}, %xmm2
-; X32-SSE-NEXT:    movdqa {{.*#+}} xmm3 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
-; X32-SSE-NEXT:    movdqa %xmm2, %xmm4
-; X32-SSE-NEXT:    pand %xmm3, %xmm4
-; X32-SSE-NEXT:    movdqa {{.*#+}} xmm0 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
-; X32-SSE-NEXT:    movdqa %xmm0, %xmm5
-; X32-SSE-NEXT:    pshufb %xmm4, %xmm5
-; X32-SSE-NEXT:    psrlw $4, %xmm2
-; X32-SSE-NEXT:    pand %xmm3, %xmm2
-; X32-SSE-NEXT:    pshufb %xmm2, %xmm0
-; X32-SSE-NEXT:    paddb %xmm5, %xmm0
-; X32-SSE-NEXT:    psadbw %xmm1, %xmm0
+; X32-SSE-NEXT:    movl $8, %eax
+; X32-SSE-NEXT:    movd %eax, %xmm0
 ; X32-SSE-NEXT:    retl
   %out = call <2 x i64> @llvm.cttz.v2i64(<2 x i64> <i64 256, i64 -1>, i1 -1)
   ret <2 x i64> %out

diff --git a/test/CodeGen/X86/vector-tzcnt-256.ll b/test/CodeGen/X86/vector-tzcnt-256.ll
index 0ced0c5b263f757f882745f56bd5b432e62b8725..a0b277ddd7327929cadac6aecea4aff3dd74d9a4 100644 (file)
--- a/test/CodeGen/X86/vector-tzcnt-256.ll
+++ b/test/CodeGen/X86/vector-tzcnt-256.ll
@@ -871,20 +871,7 @@ define <4 x i64> @foldv4i64() nounwind {
 ;
 ; X32-AVX-LABEL: foldv4i64:
 ; X32-AVX:       # BB#0:
-; X32-AVX-NEXT:    vmovdqa {{.*#+}} ymm0 = [256,0,4294967295,4294967295,0,0,255,0]
-; X32-AVX-NEXT:    vpxor %ymm1, %ymm1, %ymm1
-; X32-AVX-NEXT:    vpsubq %ymm0, %ymm1, %ymm2
-; X32-AVX-NEXT:    vpand %ymm2, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpsubq {{\.LCPI.*}}, %ymm0, %ymm0
-; X32-AVX-NEXT:    vmovdqa {{.*#+}} ymm2 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
-; X32-AVX-NEXT:    vpand %ymm2, %ymm0, %ymm3
-; X32-AVX-NEXT:    vmovdqa {{.*#+}} ymm4 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4,0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
-; X32-AVX-NEXT:    vpshufb %ymm3, %ymm4, %ymm3
-; X32-AVX-NEXT:    vpsrlw $4, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpand %ymm2, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpshufb %ymm0, %ymm4, %ymm0
-; X32-AVX-NEXT:    vpaddb %ymm3, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpsadbw %ymm1, %ymm0, %ymm0
+; X32-AVX-NEXT:    vmovaps {{.*#+}} ymm0 = [8,0,0,0,64,0,0,0]
 ; X32-AVX-NEXT:    retl
   %out = call <4 x i64> @llvm.cttz.v4i64(<4 x i64> <i64 256, i64 -1, i64 0, i64 255>, i1 0)
   ret <4 x i64> %out
@@ -898,20 +885,7 @@ define <4 x i64> @foldv4i64u() nounwind {
 ;
 ; X32-AVX-LABEL: foldv4i64u:
 ; X32-AVX:       # BB#0:
-; X32-AVX-NEXT:    vmovdqa {{.*#+}} ymm0 = [256,0,4294967295,4294967295,0,0,255,0]
-; X32-AVX-NEXT:    vpxor %ymm1, %ymm1, %ymm1
-; X32-AVX-NEXT:    vpsubq %ymm0, %ymm1, %ymm2
-; X32-AVX-NEXT:    vpand %ymm2, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpsubq {{\.LCPI.*}}, %ymm0, %ymm0
-; X32-AVX-NEXT:    vmovdqa {{.*#+}} ymm2 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
-; X32-AVX-NEXT:    vpand %ymm2, %ymm0, %ymm3
-; X32-AVX-NEXT:    vmovdqa {{.*#+}} ymm4 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4,0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
-; X32-AVX-NEXT:    vpshufb %ymm3, %ymm4, %ymm3
-; X32-AVX-NEXT:    vpsrlw $4, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpand %ymm2, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpshufb %ymm0, %ymm4, %ymm0
-; X32-AVX-NEXT:    vpaddb %ymm3, %ymm0, %ymm0
-; X32-AVX-NEXT:    vpsadbw %ymm1, %ymm0, %ymm0
+; X32-AVX-NEXT:    vmovaps {{.*#+}} ymm0 = [8,0,0,0,64,0,0,0]
 ; X32-AVX-NEXT:    retl
   %out = call <4 x i64> @llvm.cttz.v4i64(<4 x i64> <i64 256, i64 -1, i64 0, i64 255>, i1 -1)
   ret <4 x i64> %out