[CodeGen] Replace the reused stores' chain for extractelt expansion.

This fixes a subtle issue that was introduced in r205153.

When reusing a store for the extractelement expansion (to load directly
from it, inserting of going through the stack), later stores to the
same location might have overwritten the data we were expecting to
extract from.

To fix that, we need to explicitly replace the chain going out of the
reused store, so that later stores also have an explicit dependency on
the generated element-extracting loads, and can't clobber them.

rdar://20066785
Differential Revision: http://reviews.llvm.org/D8180

llvm-svn: 231721

lib/CodeGen/SelectionDAG/LegalizeDAG.cpp

@@ -1442,13 +1442,27 @@ SDValue SelectionDAGLegalize::ExpandExtractFromVectorThroughStack(SDValue Op) {
   Idx = DAG.getZExtOrTrunc(Idx, dl, TLI.getPointerTy());
   StackPtr = DAG.getNode(ISD::ADD, dl, Idx.getValueType(), Idx, StackPtr);
 
+  SDValue NewLoad;
+
   if (Op.getValueType().isVector())
-    return DAG.getLoad(Op.getValueType(), dl, Ch, StackPtr,MachinePointerInfo(),
+    NewLoad = DAG.getLoad(Op.getValueType(), dl, Ch, StackPtr,
-                       false, false, false, 0);
+                          MachinePointerInfo(), false, false, false, 0);
-  return DAG.getExtLoad(ISD::EXTLOAD, dl, Op.getValueType(), Ch, StackPtr,
+  else
-                        MachinePointerInfo(),
+    NewLoad = DAG.getExtLoad(
-                        Vec.getValueType().getVectorElementType(),
+        ISD::EXTLOAD, dl, Op.getValueType(), Ch, StackPtr, MachinePointerInfo(),
-                        false, false, false, 0);
+        Vec.getValueType().getVectorElementType(), false, false, false, 0);
+
+  // Replace the chain going out of the store, by the one out of the load.
+  DAG.ReplaceAllUsesOfValueWith(Ch, SDValue(NewLoad.getNode(), 1));
+
+  // We introduced a cycle though, so update the loads operands, making sure
+  // to use the original store's chain as an incoming chain.
+  SmallVector<SDValue, 6> NewLoadOperands(NewLoad->op_begin(),
+                                          NewLoad->op_end());
+  NewLoadOperands[0] = Ch;
+  NewLoad =
+      SDValue(DAG.UpdateNodeOperands(NewLoad.getNode(), NewLoadOperands), 0);
+  return NewLoad;
 }
 
 SDValue SelectionDAGLegalize::ExpandInsertToVectorThroughStack(SDValue Op) {

test/CodeGen/X86/extractelement-legalization-store-ordering.ll

@@ -0,0 +1,57 @@
+; RUN: llc < %s -mtriple i386-apple-darwin -mcpu=yonah | FileCheck %s
+
+target datalayout = "e-m:o-p:32:32-f64:32:64-f80:128-n8:16:32-S128"
+
+; Make sure we don't break load/store ordering when turning an extractelement
+; into loads, off the stack or a previous store.
+; Be very explicit about the ordering/stack offsets.
+
+; CHECK-LABEL: test_extractelement_legalization_storereuse:
+; CHECK:      # BB#0
+; CHECK-NEXT: pushl %ebx
+; CHECK-NEXT: pushl %edi
+; CHECK-NEXT: pushl %esi
+; CHECK-NEXT: movl 16(%esp), %eax
+; CHECK-NEXT: movl 24(%esp), %ecx
+; CHECK-NEXT: movl 20(%esp), %edx
+; CHECK-NEXT: paddd (%edx), %xmm0
+; CHECK-NEXT: movdqa %xmm0, (%edx)
+; CHECK-NEXT: shll $4, %ecx
+; CHECK-NEXT: movl (%ecx,%edx), %esi
+; CHECK-NEXT: movl 12(%ecx,%edx), %edi
+; CHECK-NEXT: movl 8(%ecx,%edx), %ebx
+; CHECK-NEXT: movl 4(%ecx,%edx), %edx
+; CHECK-NEXT: movl %esi, 12(%eax,%ecx)
+; CHECK-NEXT: movl %edx, (%eax,%ecx)
+; CHECK-NEXT: movl %ebx, 8(%eax,%ecx)
+; CHECK-NEXT: movl %edi, 4(%eax,%ecx)
+; CHECK-NEXT: popl %esi
+; CHECK-NEXT: popl %edi
+; CHECK-NEXT: popl %ebx
+; CHECK-NEXT: retl
+define void @test_extractelement_legalization_storereuse(<4 x i32> %a, i32* nocapture %x, i32* nocapture readonly %y, i32 %i) #0 {
+entry:
+  %0 = bitcast i32* %y to <4 x i32>*
+  %1 = load <4 x i32>, <4 x i32>* %0, align 16
+  %am = add <4 x i32> %a, %1
+  store <4 x i32> %am, <4 x i32>* %0, align 16
+  %ip0 = shl nsw i32 %i, 2
+  %ip1 = or i32 %ip0, 1
+  %ip2 = or i32 %ip0, 2
+  %ip3 = or i32 %ip0, 3
+  %vecext = extractelement <4 x i32> %am, i32 %ip0
+  %arrayidx = getelementptr inbounds i32, i32* %x, i32 %ip3
+  store i32 %vecext, i32* %arrayidx, align 4
+  %vecext5 = extractelement <4 x i32> %am, i32 %ip1
+  %arrayidx8 = getelementptr inbounds i32, i32* %x, i32 %ip0
+  store i32 %vecext5, i32* %arrayidx8, align 4
+  %vecext11 = extractelement <4 x i32> %am, i32 %ip2
+  %arrayidx14 = getelementptr inbounds i32, i32* %x, i32 %ip2
+  store i32 %vecext11, i32* %arrayidx14, align 4
+  %vecext17 = extractelement <4 x i32> %am, i32 %ip3
+  %arrayidx20 = getelementptr inbounds i32, i32* %x, i32 %ip1
+  store i32 %vecext17, i32* %arrayidx20, align 4
+  ret void
+}
+
+attributes #0 = { nounwind }

test/CodeGen/X86/vector-idiv.ll

@@ -460,6 +460,9 @@ define <16 x i8> @test7(<16 x i8> %a) #0 {
 ;
 ; SSE-LABEL: test7:
 ; SSE:       # BB#0:
+; SSE-NEXT:    pushq %rbp
+; SSE-NEXT:    pushq %r14
+; SSE-NEXT:    pushq %rbx
 ; SSE-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
 ; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %eax
 ; SSE-NEXT:    imull $-109, %eax, %ecx
@@ -471,156 +474,156 @@ define <16 x i8> @test7(<16 x i8> %a) #0 {
 ; SSE-NEXT:    addb %al, %cl
 ; SSE-NEXT:    movzbl %cl, %eax
 ; SSE-NEXT:    movd %eax, %xmm0
+; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %r14d
+; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %edx
+; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %r9d
 ; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %eax
-; SSE-NEXT:    imull $-109, %eax, %ecx
+; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %r11d
-; SSE-NEXT:    shrl $8, %ecx
+; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %ecx
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %r8d
-; SSE-NEXT:    movb %cl, %al
+; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %esi
-; SSE-NEXT:    shrb $7, %al
+; SSE-NEXT:    imull $-109, %esi, %edi
-; SSE-NEXT:    sarb $2, %cl
+; SSE-NEXT:    shrl $8, %edi
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    addb %sil, %dil
-; SSE-NEXT:    movzbl %cl, %eax
+; SSE-NEXT:    movb %dil, %bl
-; SSE-NEXT:    movd %eax, %xmm1
+; SSE-NEXT:    shrb $7, %bl
+; SSE-NEXT:    sarb $2, %dil
+; SSE-NEXT:    addb %bl, %dil
+; SSE-NEXT:    movzbl %dil, %esi
+; SSE-NEXT:    movd %esi, %xmm1
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    imull $-109, %eax, %esi
-; SSE-NEXT:    imull $-109, %eax, %ecx
+; SSE-NEXT:    shrl $8, %esi
-; SSE-NEXT:    shrl $8, %ecx
+; SSE-NEXT:    addb %al, %sil
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movb %sil, %al
-; SSE-NEXT:    movb %cl, %al
 ; SSE-NEXT:    shrb $7, %al
-; SSE-NEXT:    sarb $2, %cl
+; SSE-NEXT:    sarb $2, %sil
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    addb %al, %sil
-; SSE-NEXT:    movzbl %cl, %eax
+; SSE-NEXT:    movzbl %sil, %eax
 ; SSE-NEXT:    movd %eax, %xmm2
-; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %ebp
-; SSE-NEXT:    imull $-109, %eax, %ecx
+; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %esi
-; SSE-NEXT:    shrl $8, %ecx
+; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %r10d
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %edi
-; SSE-NEXT:    movb %cl, %al
+; SSE-NEXT:    imull $-109, %edi, %ebx
+; SSE-NEXT:    shrl $8, %ebx
+; SSE-NEXT:    addb %dil, %bl
+; SSE-NEXT:    movb %bl, %al
 ; SSE-NEXT:    shrb $7, %al
-; SSE-NEXT:    sarb $2, %cl
+; SSE-NEXT:    sarb $2, %bl
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    addb %al, %bl
-; SSE-NEXT:    movzbl %cl, %eax
+; SSE-NEXT:    movzbl %bl, %eax
 ; SSE-NEXT:    movd %eax, %xmm0
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
-; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    imull $-109, %edx, %eax
-; SSE-NEXT:    imull $-109, %eax, %ecx
+; SSE-NEXT:    shrl $8, %eax
-; SSE-NEXT:    shrl $8, %ecx
+; SSE-NEXT:    addb %dl, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movb %al, %dl
-; SSE-NEXT:    movb %cl, %al
+; SSE-NEXT:    shrb $7, %dl
-; SSE-NEXT:    shrb $7, %al
+; SSE-NEXT:    sarb $2, %al
-; SSE-NEXT:    sarb $2, %cl
+; SSE-NEXT:    addb %dl, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movzbl %al, %eax
-; SSE-NEXT:    movzbl %cl, %eax
 ; SSE-NEXT:    movd %eax, %xmm1
-; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    imull $-109, %esi, %eax
-; SSE-NEXT:    imull $-109, %eax, %ecx
+; SSE-NEXT:    shrl $8, %eax
-; SSE-NEXT:    shrl $8, %ecx
+; SSE-NEXT:    addb %sil, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movb %al, %dl
-; SSE-NEXT:    movb %cl, %al
+; SSE-NEXT:    shrb $7, %dl
-; SSE-NEXT:    shrb $7, %al
+; SSE-NEXT:    sarb $2, %al
-; SSE-NEXT:    sarb $2, %cl
+; SSE-NEXT:    addb %dl, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movzbl %al, %eax
-; SSE-NEXT:    movzbl %cl, %eax
 ; SSE-NEXT:    movd %eax, %xmm2
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
-; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    imull $-109, %ecx, %eax
-; SSE-NEXT:    imull $-109, %eax, %ecx
+; SSE-NEXT:    shrl $8, %eax
-; SSE-NEXT:    shrl $8, %ecx
+; SSE-NEXT:    addb %cl, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movb %al, %cl
-; SSE-NEXT:    movb %cl, %al
+; SSE-NEXT:    shrb $7, %cl
-; SSE-NEXT:    shrb $7, %al
+; SSE-NEXT:    sarb $2, %al
-; SSE-NEXT:    sarb $2, %cl
+; SSE-NEXT:    addb %cl, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movzbl %al, %eax
-; SSE-NEXT:    movzbl %cl, %eax
 ; SSE-NEXT:    movd %eax, %xmm3
+; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %ecx
 ; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %eax
-; SSE-NEXT:    imull $-109, %eax, %ecx
+; SSE-NEXT:    imull $-109, %eax, %edx
-; SSE-NEXT:    shrl $8, %ecx
+; SSE-NEXT:    shrl $8, %edx
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    addb %al, %dl
-; SSE-NEXT:    movb %cl, %al
+; SSE-NEXT:    movb %dl, %al
 ; SSE-NEXT:    shrb $7, %al
-; SSE-NEXT:    sarb $2, %cl
+; SSE-NEXT:    sarb $2, %dl
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    addb %al, %dl
-; SSE-NEXT:    movzbl %cl, %eax
+; SSE-NEXT:    movzbl %dl, %eax
 ; SSE-NEXT:    movd %eax, %xmm1
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3],xmm1[4],xmm3[4],xmm1[5],xmm3[5],xmm1[6],xmm3[6],xmm1[7],xmm3[7]
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    imull $-109, %r14d, %eax
-; SSE-NEXT:    imull $-109, %eax, %ecx
+; SSE-NEXT:    shrl $8, %eax
-; SSE-NEXT:    shrl $8, %ecx
+; SSE-NEXT:    addb %r14b, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movb %al, %dl
-; SSE-NEXT:    movb %cl, %al
+; SSE-NEXT:    shrb $7, %dl
-; SSE-NEXT:    shrb $7, %al
+; SSE-NEXT:    sarb $2, %al
-; SSE-NEXT:    sarb $2, %cl
+; SSE-NEXT:    addb %dl, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movzbl %al, %eax
-; SSE-NEXT:    movzbl %cl, %eax
 ; SSE-NEXT:    movd %eax, %xmm2
-; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    imull $-109, %ebp, %eax
-; SSE-NEXT:    imull $-109, %eax, %ecx
+; SSE-NEXT:    shrl $8, %eax
-; SSE-NEXT:    shrl $8, %ecx
+; SSE-NEXT:    addb %bpl, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movb %al, %dl
-; SSE-NEXT:    movb %cl, %al
+; SSE-NEXT:    shrb $7, %dl
-; SSE-NEXT:    shrb $7, %al
+; SSE-NEXT:    sarb $2, %al
-; SSE-NEXT:    sarb $2, %cl
+; SSE-NEXT:    addb %dl, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movzbl %al, %eax
-; SSE-NEXT:    movzbl %cl, %eax
 ; SSE-NEXT:    movd %eax, %xmm0
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
-; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    imull $-109, %r11d, %eax
-; SSE-NEXT:    imull $-109, %eax, %ecx
+; SSE-NEXT:    shrl $8, %eax
-; SSE-NEXT:    shrl $8, %ecx
+; SSE-NEXT:    addb %r11b, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movb %al, %dl
-; SSE-NEXT:    movb %cl, %al
+; SSE-NEXT:    shrb $7, %dl
-; SSE-NEXT:    shrb $7, %al
+; SSE-NEXT:    sarb $2, %al
-; SSE-NEXT:    sarb $2, %cl
+; SSE-NEXT:    addb %dl, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movzbl %al, %eax
-; SSE-NEXT:    movzbl %cl, %eax
 ; SSE-NEXT:    movd %eax, %xmm3
-; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    imull $-109, %ecx, %eax
-; SSE-NEXT:    imull $-109, %eax, %ecx
+; SSE-NEXT:    shrl $8, %eax
-; SSE-NEXT:    shrl $8, %ecx
+; SSE-NEXT:    addb %cl, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movb %al, %cl
-; SSE-NEXT:    movb %cl, %al
+; SSE-NEXT:    shrb $7, %cl
-; SSE-NEXT:    shrb $7, %al
+; SSE-NEXT:    sarb $2, %al
-; SSE-NEXT:    sarb $2, %cl
+; SSE-NEXT:    addb %cl, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movzbl %al, %eax
-; SSE-NEXT:    movzbl %cl, %eax
 ; SSE-NEXT:    movd %eax, %xmm2
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3],xmm2[4],xmm3[4],xmm2[5],xmm3[5],xmm2[6],xmm3[6],xmm2[7],xmm3[7]
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    imull $-109, %r9d, %eax
-; SSE-NEXT:    imull $-109, %eax, %ecx
+; SSE-NEXT:    shrl $8, %eax
-; SSE-NEXT:    shrl $8, %ecx
+; SSE-NEXT:    addb %r9b, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movb %al, %cl
-; SSE-NEXT:    movb %cl, %al
+; SSE-NEXT:    shrb $7, %cl
-; SSE-NEXT:    shrb $7, %al
+; SSE-NEXT:    sarb $2, %al
-; SSE-NEXT:    sarb $2, %cl
+; SSE-NEXT:    addb %cl, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movzbl %al, %eax
-; SSE-NEXT:    movzbl %cl, %eax
 ; SSE-NEXT:    movd %eax, %xmm0
-; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    imull $-109, %r10d, %eax
-; SSE-NEXT:    imull $-109, %eax, %ecx
+; SSE-NEXT:    shrl $8, %eax
-; SSE-NEXT:    shrl $8, %ecx
+; SSE-NEXT:    addb %r10b, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movb %al, %cl
-; SSE-NEXT:    movb %cl, %al
+; SSE-NEXT:    shrb $7, %cl
-; SSE-NEXT:    shrb $7, %al
+; SSE-NEXT:    sarb $2, %al
-; SSE-NEXT:    sarb $2, %cl
+; SSE-NEXT:    addb %cl, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movzbl %al, %eax
-; SSE-NEXT:    movzbl %cl, %eax
 ; SSE-NEXT:    movd %eax, %xmm3
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    imull $-109, %r8d, %eax
-; SSE-NEXT:    imull $-109, %eax, %ecx
+; SSE-NEXT:    shrl $8, %eax
-; SSE-NEXT:    shrl $8, %ecx
+; SSE-NEXT:    addb %r8b, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movb %al, %cl
-; SSE-NEXT:    movb %cl, %al
+; SSE-NEXT:    shrb $7, %cl
-; SSE-NEXT:    shrb $7, %al
+; SSE-NEXT:    sarb $2, %al
-; SSE-NEXT:    sarb $2, %cl
+; SSE-NEXT:    addb %cl, %al
-; SSE-NEXT:    addb %al, %cl
+; SSE-NEXT:    movzbl %al, %eax
-; SSE-NEXT:    movzbl %cl, %eax
 ; SSE-NEXT:    movd %eax, %xmm4
 ; SSE-NEXT:    movsbl -{{[0-9]+}}(%rsp), %eax
 ; SSE-NEXT:    imull $-109, %eax, %ecx
@@ -636,6 +639,9 @@ define <16 x i8> @test7(<16 x i8> %a) #0 {
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; SSE-NEXT:    popq %rbx
+; SSE-NEXT:    popq %r14
+; SSE-NEXT:    popq %rbp
 ; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: test7: