Index: lib/Target/X86/X86ISelDAGToDAG.cpp
===================================================================
--- lib/Target/X86/X86ISelDAGToDAG.cpp
+++ lib/Target/X86/X86ISelDAGToDAG.cpp
@@ -443,6 +443,14 @@
 
     bool matchBEXTRFromAnd(SDNode *Node);
 
+    /// If the high bits of an 'and' operation are known zero, check if setting
+    /// the high bits of an 'and' constant operand can produce a smaller
+    /// encoding by creating a small negative immediate rather than a large
+    /// positive one. This reverses a transform in SimplifyDemandedBits that
+    /// shrinks mask constants by clearing bits. Return 'true' if the node is
+    /// replaced.
+    bool shrinkAndImmediate(SDNode *N);
+
     bool isMaskZeroExtended(SDNode *N) const;
   };
 }
@@ -2429,6 +2437,45 @@
   return true;
 }
 
+bool X86DAGToDAGISel::shrinkAndImmediate(SDNode *And) {
+  // i8 is unshrinkable, i16 should be promoted to i32, and vector ops don't
+  // have immediate operands.
+  MVT VT = And->getSimpleValueType(0);
+  if (VT != MVT::i32 && VT != MVT::i64)
+    return false;
+
+  auto *Mask = dyn_cast<ConstantSDNode>(And->getOperand(1));
+  if (!Mask)
+    return false;
+
+  // Bail out if the constant is already negative.
+  APInt MaskVal = Mask->getAPIntValue();
+  unsigned MaskLZ = MaskVal.countLeadingZeros();
+  if (!MaskLZ)
+    return false;
+
+  // We need to set a string of mask bits at least as long as the high zero bits
+  // in the current constant to have any chance of narrowing.
+  SDValue And0 = And->getOperand(0);
+  KnownBits Known;
+  CurDAG->computeKnownBits(And0, Known);
+  if (Known.countMinLeadingZeros() < MaskLZ)
+    return false;
+
+  // Don't bother changing the constant if it won't allow smaller encoding.
+  APInt NegVal = MaskVal | APInt::getHighBitsSet(VT.getSizeInBits(), MaskLZ);
+  unsigned MinWidth = NegVal.getMinSignedBits();
+  if (MinWidth > 32 || (MinWidth > 8 && MaskVal.getMinSignedBits() <= 32))
+    return false;
+
+  SDLoc DL(And);
+  SDValue NewMask = CurDAG->getConstant(NegVal, DL, VT);
+  SDValue NewAnd = CurDAG->getNode(ISD::AND, DL, VT, And0, NewMask);
+  ReplaceNode(And, NewAnd.getNode());
+  SelectCode(NewAnd.getNode());
+  return true;
+}
+
 void X86DAGToDAGISel::Select(SDNode *Node) {
   MVT NVT = Node->getSimpleValueType(0);
   unsigned Opc, MOpc;
@@ -2483,9 +2530,10 @@
   }
 
   case ISD::AND:
-    // Try to match BEXTR/BEXTRI instruction.
     if (matchBEXTRFromAnd(Node))
       return;
+    if (shrinkAndImmediate(Node))
+      return;
 
     LLVM_FALLTHROUGH;
   case ISD::OR:
Index: test/CodeGen/X86/and-encoding.ll
===================================================================
--- test/CodeGen/X86/and-encoding.ll
+++ test/CodeGen/X86/and-encoding.ll
@@ -47,8 +47,7 @@
 ; CHECK-LABEL: lopped32_32to8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    shrl $4, %edi # encoding: [0xc1,0xef,0x04]
-; CHECK-NEXT:    andl $268435440, %edi # encoding: [0x81,0xe7,0xf0,0xff,0xff,0x0f]
-; CHECK-NEXT:    # imm = 0xFFFFFF0
+; CHECK-NEXT:    andl $-16, %edi # encoding: [0x83,0xe7,0xf0]
 ; CHECK-NEXT:    movl %edi, %eax # encoding: [0x89,0xf8]
 ; CHECK-NEXT:    retq # encoding: [0xc3]
   %shr = lshr i32 %x, 4
@@ -62,8 +61,7 @@
 ; CHECK-LABEL: lopped64_32to8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    shrq $36, %rdi # encoding: [0x48,0xc1,0xef,0x24]
-; CHECK-NEXT:    andl $268435440, %edi # encoding: [0x81,0xe7,0xf0,0xff,0xff,0x0f]
-; CHECK-NEXT:    # imm = 0xFFFFFF0
+; CHECK-NEXT:    andq $-16, %rdi # encoding: [0x48,0x83,0xe7,0xf0]
 ; CHECK-NEXT:    movq %rdi, %rax # encoding: [0x48,0x89,0xf8]
 ; CHECK-NEXT:    retq # encoding: [0xc3]
   %shr = lshr i64 %x, 36
@@ -77,9 +75,8 @@
 ; CHECK-LABEL: lopped64_64to8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    shrq $4, %rdi # encoding: [0x48,0xc1,0xef,0x04]
-; CHECK-NEXT:    movabsq $1152921504606846960, %rax # encoding: [0x48,0xb8,0xf0,0xff,0xff,0xff,0xff,0xff,0xff,0x0f]
-; CHECK-NEXT:    # imm = 0xFFFFFFFFFFFFFF0
-; CHECK-NEXT:    andq %rdi, %rax # encoding: [0x48,0x21,0xf8]
+; CHECK-NEXT:    andq $-16, %rdi # encoding: [0x48,0x83,0xe7,0xf0]
+; CHECK-NEXT:    movq %rdi, %rax # encoding: [0x48,0x89,0xf8]
 ; CHECK-NEXT:    retq # encoding: [0xc3]
   %shr = lshr i64 %x, 4
   %and = and i64 %shr, 1152921504606846960
@@ -92,9 +89,9 @@
 ; CHECK-LABEL: lopped64_64to32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    shrq $4, %rdi # encoding: [0x48,0xc1,0xef,0x04]
-; CHECK-NEXT:    movabsq $1152921504605863920, %rax # encoding: [0x48,0xb8,0xf0,0xff,0xf0,0xff,0xff,0xff,0xff,0x0f]
-; CHECK-NEXT:    # imm = 0xFFFFFFFFFF0FFF0
-; CHECK-NEXT:    andq %rdi, %rax # encoding: [0x48,0x21,0xf8]
+; CHECK-NEXT:    andq $-983056, %rdi # encoding: [0x48,0x81,0xe7,0xf0,0xff,0xf0,0xff]
+; CHECK-NEXT:    # imm = 0xFFF0FFF0
+; CHECK-NEXT:    movq %rdi, %rax # encoding: [0x48,0x89,0xf8]
 ; CHECK-NEXT:    retq # encoding: [0xc3]
   %shr = lshr i64 %x, 4
   %and = and i64 %shr, 1152921504605863920
@@ -112,8 +109,7 @@
 ; CHECK-NEXT:    # imm = 0xF0F0F0F1
 ; CHECK-NEXT:    imulq %rcx, %rax # encoding: [0x48,0x0f,0xaf,0xc1]
 ; CHECK-NEXT:    shrq $36, %rax # encoding: [0x48,0xc1,0xe8,0x24]
-; CHECK-NEXT:    andl $268435328, %eax # encoding: [0x25,0x80,0xff,0xff,0x0f]
-; CHECK-NEXT:    # imm = 0xFFFFF80
+; CHECK-NEXT:    andl $-128, %eax # encoding: [0x83,0xe0,0x80]
 ; CHECK-NEXT:    # kill: def %eax killed %eax killed %rax
 ; CHECK-NEXT:    retq # encoding: [0xc3]
   %div = udiv i32 %x, 17
Index: test/CodeGen/X86/divide-by-constant.ll
===================================================================
--- test/CodeGen/X86/divide-by-constant.ll
+++ test/CodeGen/X86/divide-by-constant.ll
@@ -49,7 +49,6 @@
 ; X32-NEXT:    movzbl {{[0-9]+}}(%esp), %eax
 ; X32-NEXT:    imull $171, %eax, %eax
 ; X32-NEXT:    shrl $9, %eax
-; X32-NEXT:    movzwl %ax, %eax
 ; X32-NEXT:    # kill: def %al killed %al killed %eax
 ; X32-NEXT:    retl
 ;
@@ -57,7 +56,6 @@
 ; X64:       # %bb.0: # %entry
 ; X64-NEXT:    imull $171, %esi, %eax
 ; X64-NEXT:    shrl $9, %eax
-; X64-NEXT:    movzwl %ax, %eax
 ; X64-NEXT:    # kill: def %al killed %al killed %eax
 ; X64-NEXT:    retq
 entry:
@@ -168,7 +166,6 @@
 ; X32-NEXT:    movzbl %al, %eax
 ; X32-NEXT:    imull $211, %eax, %eax
 ; X32-NEXT:    shrl $13, %eax
-; X32-NEXT:    movzwl %ax, %eax
 ; X32-NEXT:    # kill: def %al killed %al killed %eax
 ; X32-NEXT:    retl
 ;
@@ -178,7 +175,6 @@
 ; X64-NEXT:    movzbl %dil, %eax
 ; X64-NEXT:    imull $211, %eax, %eax
 ; X64-NEXT:    shrl $13, %eax
-; X64-NEXT:    movzwl %ax, %eax
 ; X64-NEXT:    # kill: def %al killed %al killed %eax
 ; X64-NEXT:    retq
   %div = udiv i8 %x, 78
@@ -193,7 +189,6 @@
 ; X32-NEXT:    movzbl %al, %eax
 ; X32-NEXT:    imull $71, %eax, %eax
 ; X32-NEXT:    shrl $11, %eax
-; X32-NEXT:    movzwl %ax, %eax
 ; X32-NEXT:    # kill: def %al killed %al killed %eax
 ; X32-NEXT:    retl
 ;
@@ -203,7 +198,6 @@
 ; X64-NEXT:    movzbl %dil, %eax
 ; X64-NEXT:    imull $71, %eax, %eax
 ; X64-NEXT:    shrl $11, %eax
-; X64-NEXT:    movzwl %ax, %eax
 ; X64-NEXT:    # kill: def %al killed %al killed %eax
 ; X64-NEXT:    retq
   %div = udiv i8 %x, 116
Index: test/CodeGen/X86/known-bits.ll
===================================================================
--- test/CodeGen/X86/known-bits.ll
+++ test/CodeGen/X86/known-bits.ll
@@ -13,7 +13,6 @@
 ; X32-NEXT:    movzbl (%eax), %eax
 ; X32-NEXT:    imull $101, %eax, %eax
 ; X32-NEXT:    shrl $14, %eax
-; X32-NEXT:    movzwl %ax, %eax
 ; X32-NEXT:    movzbl %al, %eax
 ; X32-NEXT:    vmovd %eax, %xmm0
 ; X32-NEXT:    vpshufb {{.*#+}} xmm0 = zero,zero,zero,zero,xmm0[0],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
@@ -51,7 +50,6 @@
 ; X64-NEXT:    movzbl (%rdi), %eax
 ; X64-NEXT:    imull $101, %eax, %eax
 ; X64-NEXT:    shrl $14, %eax
-; X64-NEXT:    movzwl %ax, %eax
 ; X64-NEXT:    movzbl %al, %eax
 ; X64-NEXT:    vmovd %eax, %xmm0
 ; X64-NEXT:    vpshufb {{.*#+}} xmm0 = zero,zero,zero,zero,xmm0[0],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
Index: test/CodeGen/X86/popcnt.ll
===================================================================
--- test/CodeGen/X86/popcnt.ll
+++ test/CodeGen/X86/popcnt.ll
@@ -71,7 +71,7 @@
 ; X32-NEXT:    andl $13107, %eax # imm = 0x3333
 ; X32-NEXT:    addl %ecx, %eax
 ; X32-NEXT:    movl %eax, %ecx
-; X32-NEXT:    andl $32752, %ecx # imm = 0x7FF0
+; X32-NEXT:    andl $-16, %ecx
 ; X32-NEXT:    shrl $4, %ecx
 ; X32-NEXT:    addl %eax, %ecx
 ; X32-NEXT:    andl $3855, %ecx # imm = 0xF0F
@@ -94,7 +94,7 @@
 ; X64-NEXT:    andl $13107, %edi # imm = 0x3333
 ; X64-NEXT:    addl %eax, %edi
 ; X64-NEXT:    movl %edi, %eax
-; X64-NEXT:    andl $32752, %eax # imm = 0x7FF0
+; X64-NEXT:    andl $-16, %eax
 ; X64-NEXT:    shrl $4, %eax
 ; X64-NEXT:    addl %edi, %eax
 ; X64-NEXT:    andl $3855, %eax # imm = 0xF0F
Index: test/CodeGen/X86/pr33844.ll
===================================================================
--- test/CodeGen/X86/pr33844.ll
+++ test/CodeGen/X86/pr33844.ll
@@ -15,7 +15,7 @@
 ; CHECK-NEXT:    shrl $31, %ecx
 ; CHECK-NEXT:    addl $2147483647, %ecx # imm = 0x7FFFFFFF
 ; CHECK-NEXT:    shrl $31, %ecx
-; CHECK-NEXT:    andl $62, %ecx
+; CHECK-NEXT:    andl $-2, %ecx
 ; CHECK-NEXT:    andl $-536870912, %eax # imm = 0xE0000000
 ; CHECK-NEXT:    orl %ecx, %eax
 ; CHECK-NEXT:    movl %eax, {{.*}}(%rip)
Index: test/CodeGen/X86/shift-pair.ll
===================================================================
--- test/CodeGen/X86/shift-pair.ll
+++ test/CodeGen/X86/shift-pair.ll
@@ -5,7 +5,7 @@
 ; CHECK-LABEL: test:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    shrq $54, %rdi
-; CHECK-NEXT:    andl $1020, %edi # imm = 0x3FC
+; CHECK-NEXT:    andq $-4, %rdi
 ; CHECK-NEXT:    movq %rdi, %rax
 ; CHECK-NEXT:    retq
     %B = lshr i64 %A, 56
Index: test/CodeGen/X86/urem-i8-constant.ll
===================================================================
--- test/CodeGen/X86/urem-i8-constant.ll
+++ test/CodeGen/X86/urem-i8-constant.ll
@@ -9,7 +9,6 @@
 ; CHECK-NEXT:    movzbl {{[0-9]+}}(%esp), %ecx
 ; CHECK-NEXT:    imull $111, %ecx, %eax
 ; CHECK-NEXT:    shrl $12, %eax
-; CHECK-NEXT:    movzwl %ax, %eax
 ; CHECK-NEXT:    movb $37, %dl
 ; CHECK-NEXT:    # kill: def %al killed %al killed %eax
 ; CHECK-NEXT:    mulb %dl
Index: test/CodeGen/X86/vector-sext.ll
===================================================================
--- test/CodeGen/X86/vector-sext.ll
+++ test/CodeGen/X86/vector-sext.ll
@@ -2190,23 +2190,13 @@
 ; SSE2:       # %bb.0: # %entry
 ; SSE2-NEXT:    movzbl (%rdi), %eax
 ; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    shrl $3, %ecx
-; SSE2-NEXT:    andl $1, %ecx
+; SSE2-NEXT:    shrl $7, %ecx
 ; SSE2-NEXT:    movd %ecx, %xmm0
 ; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    shrl $2, %ecx
-; SSE2-NEXT:    andl $1, %ecx
-; SSE2-NEXT:    movd %ecx, %xmm2
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSE2-NEXT:    movl %eax, %ecx
+; SSE2-NEXT:    shrl $6, %ecx
 ; SSE2-NEXT:    andl $1, %ecx
 ; SSE2-NEXT:    movd %ecx, %xmm1
-; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    shrl %ecx
-; SSE2-NEXT:    andl $1, %ecx
-; SSE2-NEXT:    movd %ecx, %xmm0
 ; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
 ; SSE2-NEXT:    movl %eax, %ecx
 ; SSE2-NEXT:    shrl $5, %ecx
 ; SSE2-NEXT:    andl $1, %ecx
@@ -2216,15 +2206,24 @@
 ; SSE2-NEXT:    andl $1, %ecx
 ; SSE2-NEXT:    movd %ecx, %xmm2
 ; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
 ; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    shrl $6, %ecx
+; SSE2-NEXT:    shrl $3, %ecx
 ; SSE2-NEXT:    andl $1, %ecx
 ; SSE2-NEXT:    movd %ecx, %xmm0
-; SSE2-NEXT:    shrl $7, %eax
-; SSE2-NEXT:    movzwl %ax, %eax
-; SSE2-NEXT:    movd %eax, %xmm3
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1]
+; SSE2-NEXT:    movl %eax, %ecx
+; SSE2-NEXT:    shrl $2, %ecx
+; SSE2-NEXT:    andl $1, %ecx
+; SSE2-NEXT:    movd %ecx, %xmm3
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
+; SSE2-NEXT:    movl %eax, %ecx
+; SSE2-NEXT:    andl $1, %ecx
+; SSE2-NEXT:    movd %ecx, %xmm1
+; SSE2-NEXT:    shrl %eax
+; SSE2-NEXT:    andl $1, %eax
+; SSE2-NEXT:    movd %eax, %xmm0
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
 ; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
 ; SSE2-NEXT:    movdqa %xmm1, %xmm0
 ; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
@@ -2239,23 +2238,13 @@
 ; SSSE3:       # %bb.0: # %entry
 ; SSSE3-NEXT:    movzbl (%rdi), %eax
 ; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    shrl $3, %ecx
-; SSSE3-NEXT:    andl $1, %ecx
+; SSSE3-NEXT:    shrl $7, %ecx
 ; SSSE3-NEXT:    movd %ecx, %xmm0
 ; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    shrl $2, %ecx
-; SSSE3-NEXT:    andl $1, %ecx
-; SSSE3-NEXT:    movd %ecx, %xmm2
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSSE3-NEXT:    movl %eax, %ecx
+; SSSE3-NEXT:    shrl $6, %ecx
 ; SSSE3-NEXT:    andl $1, %ecx
 ; SSSE3-NEXT:    movd %ecx, %xmm1
-; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    shrl %ecx
-; SSSE3-NEXT:    andl $1, %ecx
-; SSSE3-NEXT:    movd %ecx, %xmm0
 ; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
 ; SSSE3-NEXT:    movl %eax, %ecx
 ; SSSE3-NEXT:    shrl $5, %ecx
 ; SSSE3-NEXT:    andl $1, %ecx
@@ -2265,15 +2254,24 @@
 ; SSSE3-NEXT:    andl $1, %ecx
 ; SSSE3-NEXT:    movd %ecx, %xmm2
 ; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
+; SSSE3-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
 ; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    shrl $6, %ecx
+; SSSE3-NEXT:    shrl $3, %ecx
 ; SSSE3-NEXT:    andl $1, %ecx
 ; SSSE3-NEXT:    movd %ecx, %xmm0
-; SSSE3-NEXT:    shrl $7, %eax
-; SSSE3-NEXT:    movzwl %ax, %eax
-; SSSE3-NEXT:    movd %eax, %xmm3
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1]
+; SSSE3-NEXT:    movl %eax, %ecx
+; SSSE3-NEXT:    shrl $2, %ecx
+; SSSE3-NEXT:    andl $1, %ecx
+; SSSE3-NEXT:    movd %ecx, %xmm3
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
+; SSSE3-NEXT:    movl %eax, %ecx
+; SSSE3-NEXT:    andl $1, %ecx
+; SSSE3-NEXT:    movd %ecx, %xmm1
+; SSSE3-NEXT:    shrl %eax
+; SSSE3-NEXT:    andl $1, %eax
+; SSSE3-NEXT:    movd %eax, %xmm0
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
 ; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
 ; SSSE3-NEXT:    movdqa %xmm1, %xmm0
 ; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
@@ -2315,7 +2313,6 @@
 ; SSE41-NEXT:    andl $1, %ecx
 ; SSE41-NEXT:    pinsrw $6, %ecx, %xmm1
 ; SSE41-NEXT:    shrl $7, %eax
-; SSE41-NEXT:    movzwl %ax, %eax
 ; SSE41-NEXT:    pinsrw $7, %eax, %xmm1
 ; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm0 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero
 ; SSE41-NEXT:    pslld $31, %xmm0
@@ -3001,52 +2998,32 @@
 ; SSE2:       # %bb.0: # %entry
 ; SSE2-NEXT:    movzwl (%rdi), %eax
 ; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    shrl $7, %ecx
-; SSE2-NEXT:    andl $1, %ecx
+; SSE2-NEXT:    shrl $15, %ecx
 ; SSE2-NEXT:    movd %ecx, %xmm0
 ; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    shrl $6, %ecx
+; SSE2-NEXT:    shrl $14, %ecx
 ; SSE2-NEXT:    andl $1, %ecx
 ; SSE2-NEXT:    movd %ecx, %xmm1
 ; SSE2-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]
 ; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    shrl $5, %ecx
+; SSE2-NEXT:    shrl $13, %ecx
 ; SSE2-NEXT:    andl $1, %ecx
 ; SSE2-NEXT:    movd %ecx, %xmm0
 ; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    shrl $4, %ecx
+; SSE2-NEXT:    shrl $12, %ecx
 ; SSE2-NEXT:    andl $1, %ecx
 ; SSE2-NEXT:    movd %ecx, %xmm2
 ; SSE2-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]
 ; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
 ; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    shrl $3, %ecx
-; SSE2-NEXT:    andl $1, %ecx
-; SSE2-NEXT:    movd %ecx, %xmm0
-; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    shrl $2, %ecx
-; SSE2-NEXT:    andl $1, %ecx
-; SSE2-NEXT:    movd %ecx, %xmm3
-; SSE2-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]
-; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    andl $1, %ecx
-; SSE2-NEXT:    movd %ecx, %xmm1
-; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    shrl %ecx
-; SSE2-NEXT:    andl $1, %ecx
-; SSE2-NEXT:    movd %ecx, %xmm0
-; SSE2-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]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; SSE2-NEXT:    movl %eax, %ecx
 ; SSE2-NEXT:    shrl $11, %ecx
 ; SSE2-NEXT:    andl $1, %ecx
 ; SSE2-NEXT:    movd %ecx, %xmm0
 ; SSE2-NEXT:    movl %eax, %ecx
 ; SSE2-NEXT:    shrl $10, %ecx
 ; SSE2-NEXT:    andl $1, %ecx
-; SSE2-NEXT:    movd %ecx, %xmm2
-; SSE2-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]
+; SSE2-NEXT:    movd %ecx, %xmm1
+; SSE2-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]
 ; SSE2-NEXT:    movl %eax, %ecx
 ; SSE2-NEXT:    shrl $9, %ecx
 ; SSE2-NEXT:    andl $1, %ecx
@@ -3056,26 +3033,45 @@
 ; SSE2-NEXT:    andl $1, %ecx
 ; SSE2-NEXT:    movd %ecx, %xmm0
 ; SSE2-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]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
 ; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    shrl $13, %ecx
+; SSE2-NEXT:    shrl $7, %ecx
+; SSE2-NEXT:    andl $1, %ecx
+; SSE2-NEXT:    movd %ecx, %xmm1
+; SSE2-NEXT:    movl %eax, %ecx
+; SSE2-NEXT:    shrl $6, %ecx
 ; SSE2-NEXT:    andl $1, %ecx
 ; SSE2-NEXT:    movd %ecx, %xmm2
+; SSE2-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]
 ; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    shrl $12, %ecx
+; SSE2-NEXT:    shrl $5, %ecx
+; SSE2-NEXT:    andl $1, %ecx
+; SSE2-NEXT:    movd %ecx, %xmm1
+; SSE2-NEXT:    movl %eax, %ecx
+; SSE2-NEXT:    shrl $4, %ecx
 ; SSE2-NEXT:    andl $1, %ecx
 ; SSE2-NEXT:    movd %ecx, %xmm3
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3],xmm3[4],xmm2[4],xmm3[5],xmm2[5],xmm3[6],xmm2[6],xmm3[7],xmm2[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3],xmm3[4],xmm1[4],xmm3[5],xmm1[5],xmm3[6],xmm1[6],xmm3[7],xmm1[7]
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
 ; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    shrl $14, %ecx
+; SSE2-NEXT:    shrl $3, %ecx
+; SSE2-NEXT:    andl $1, %ecx
+; SSE2-NEXT:    movd %ecx, %xmm1
+; SSE2-NEXT:    movl %eax, %ecx
+; SSE2-NEXT:    shrl $2, %ecx
 ; SSE2-NEXT:    andl $1, %ecx
 ; SSE2-NEXT:    movd %ecx, %xmm2
-; SSE2-NEXT:    shrl $15, %eax
-; SSE2-NEXT:    movzwl %ax, %eax
+; SSE2-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]
+; SSE2-NEXT:    movl %eax, %ecx
+; SSE2-NEXT:    andl $1, %ecx
+; SSE2-NEXT:    movd %ecx, %xmm1
+; SSE2-NEXT:    shrl %eax
+; SSE2-NEXT:    andl $1, %eax
 ; SSE2-NEXT:    movd %eax, %xmm4
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1],xmm2[2],xmm4[2],xmm2[3],xmm4[3],xmm2[4],xmm4[4],xmm2[5],xmm4[5],xmm2[6],xmm4[6],xmm2[7],xmm4[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1],xmm1[2],xmm4[2],xmm1[3],xmm4[3],xmm1[4],xmm4[4],xmm1[5],xmm4[5],xmm1[6],xmm4[6],xmm1[7],xmm4[7]
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
 ; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm0[0]
 ; SSE2-NEXT:    movdqa %xmm1, %xmm0
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
@@ -3090,52 +3086,32 @@
 ; SSSE3:       # %bb.0: # %entry
 ; SSSE3-NEXT:    movzwl (%rdi), %eax
 ; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    shrl $7, %ecx
-; SSSE3-NEXT:    andl $1, %ecx
+; SSSE3-NEXT:    shrl $15, %ecx
 ; SSSE3-NEXT:    movd %ecx, %xmm0
 ; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    shrl $6, %ecx
+; SSSE3-NEXT:    shrl $14, %ecx
 ; SSSE3-NEXT:    andl $1, %ecx
 ; SSSE3-NEXT:    movd %ecx, %xmm1
 ; SSSE3-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]
 ; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    shrl $5, %ecx
+; SSSE3-NEXT:    shrl $13, %ecx
 ; SSSE3-NEXT:    andl $1, %ecx
 ; SSSE3-NEXT:    movd %ecx, %xmm0
 ; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    shrl $4, %ecx
+; SSSE3-NEXT:    shrl $12, %ecx
 ; SSSE3-NEXT:    andl $1, %ecx
 ; SSSE3-NEXT:    movd %ecx, %xmm2
 ; SSSE3-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]
 ; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
 ; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    shrl $3, %ecx
-; SSSE3-NEXT:    andl $1, %ecx
-; SSSE3-NEXT:    movd %ecx, %xmm0
-; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    shrl $2, %ecx
-; SSSE3-NEXT:    andl $1, %ecx
-; SSSE3-NEXT:    movd %ecx, %xmm3
-; SSSE3-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]
-; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    andl $1, %ecx
-; SSSE3-NEXT:    movd %ecx, %xmm1
-; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    shrl %ecx
-; SSSE3-NEXT:    andl $1, %ecx
-; SSSE3-NEXT:    movd %ecx, %xmm0
-; SSSE3-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]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; SSSE3-NEXT:    movl %eax, %ecx
 ; SSSE3-NEXT:    shrl $11, %ecx
 ; SSSE3-NEXT:    andl $1, %ecx
 ; SSSE3-NEXT:    movd %ecx, %xmm0
 ; SSSE3-NEXT:    movl %eax, %ecx
 ; SSSE3-NEXT:    shrl $10, %ecx
 ; SSSE3-NEXT:    andl $1, %ecx
-; SSSE3-NEXT:    movd %ecx, %xmm2
-; SSSE3-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]
+; SSSE3-NEXT:    movd %ecx, %xmm1
+; SSSE3-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]
 ; SSSE3-NEXT:    movl %eax, %ecx
 ; SSSE3-NEXT:    shrl $9, %ecx
 ; SSSE3-NEXT:    andl $1, %ecx
@@ -3145,26 +3121,45 @@
 ; SSSE3-NEXT:    andl $1, %ecx
 ; SSSE3-NEXT:    movd %ecx, %xmm0
 ; SSSE3-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]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
 ; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    shrl $13, %ecx
+; SSSE3-NEXT:    shrl $7, %ecx
+; SSSE3-NEXT:    andl $1, %ecx
+; SSSE3-NEXT:    movd %ecx, %xmm1
+; SSSE3-NEXT:    movl %eax, %ecx
+; SSSE3-NEXT:    shrl $6, %ecx
 ; SSSE3-NEXT:    andl $1, %ecx
 ; SSSE3-NEXT:    movd %ecx, %xmm2
+; SSSE3-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]
 ; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    shrl $12, %ecx
+; SSSE3-NEXT:    shrl $5, %ecx
+; SSSE3-NEXT:    andl $1, %ecx
+; SSSE3-NEXT:    movd %ecx, %xmm1
+; SSSE3-NEXT:    movl %eax, %ecx
+; SSSE3-NEXT:    shrl $4, %ecx
 ; SSSE3-NEXT:    andl $1, %ecx
 ; SSSE3-NEXT:    movd %ecx, %xmm3
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3],xmm3[4],xmm2[4],xmm3[5],xmm2[5],xmm3[6],xmm2[6],xmm3[7],xmm2[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3],xmm3[4],xmm1[4],xmm3[5],xmm1[5],xmm3[6],xmm1[6],xmm3[7],xmm1[7]
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
 ; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    shrl $14, %ecx
+; SSSE3-NEXT:    shrl $3, %ecx
+; SSSE3-NEXT:    andl $1, %ecx
+; SSSE3-NEXT:    movd %ecx, %xmm1
+; SSSE3-NEXT:    movl %eax, %ecx
+; SSSE3-NEXT:    shrl $2, %ecx
 ; SSSE3-NEXT:    andl $1, %ecx
 ; SSSE3-NEXT:    movd %ecx, %xmm2
-; SSSE3-NEXT:    shrl $15, %eax
-; SSSE3-NEXT:    movzwl %ax, %eax
+; SSSE3-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]
+; SSSE3-NEXT:    movl %eax, %ecx
+; SSSE3-NEXT:    andl $1, %ecx
+; SSSE3-NEXT:    movd %ecx, %xmm1
+; SSSE3-NEXT:    shrl %eax
+; SSSE3-NEXT:    andl $1, %eax
 ; SSSE3-NEXT:    movd %eax, %xmm4
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1],xmm2[2],xmm4[2],xmm2[3],xmm4[3],xmm2[4],xmm4[4],xmm2[5],xmm4[5],xmm2[6],xmm4[6],xmm2[7],xmm4[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1],xmm1[2],xmm4[2],xmm1[3],xmm4[3],xmm1[4],xmm4[4],xmm1[5],xmm4[5],xmm1[6],xmm4[6],xmm1[7],xmm4[7]
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
+; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
 ; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm0[0]
 ; SSSE3-NEXT:    movdqa %xmm1, %xmm0
 ; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
@@ -3238,7 +3233,6 @@
 ; SSE41-NEXT:    andl $1, %ecx
 ; SSE41-NEXT:    pinsrb $14, %ecx, %xmm1
 ; SSE41-NEXT:    shrl $15, %eax
-; SSE41-NEXT:    movzwl %ax, %eax
 ; SSE41-NEXT:    pinsrb $15, %eax, %xmm1
 ; SSE41-NEXT:    pmovzxbw {{.*#+}} xmm0 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
 ; SSE41-NEXT:    psllw $15, %xmm0
Index: test/CodeGen/X86/win64_frame.ll
===================================================================
--- test/CodeGen/X86/win64_frame.ll
+++ test/CodeGen/X86/win64_frame.ll
@@ -177,9 +177,8 @@
 ; ALL-NEXT:    movq %rsp, %rbx
 ; ALL-NEXT:    movl 288(%rbp), %esi
 ; ALL-NEXT:    movl %ecx, %eax
-; ALL-NEXT:    leaq 15(,%rax,4), %rcx
-; ALL-NEXT:    movabsq $34359738352, %rax # imm = 0x7FFFFFFF0
-; ALL-NEXT:    andq %rcx, %rax
+; ALL-NEXT:    leaq 15(,%rax,4), %rax
+; ALL-NEXT:    andq $-16, %rax
 ; ALL-NEXT:    callq __chkstk
 ; ALL-NEXT:    subq %rax, %rsp
 ; ALL-NEXT:    subq $32, %rsp
Index: test/CodeGen/X86/x86-64-baseptr.ll
===================================================================
--- test/CodeGen/X86/x86-64-baseptr.ll
+++ test/CodeGen/X86/x86-64-baseptr.ll
@@ -21,16 +21,15 @@
 ; CHECK-NEXT:    subq $32, %rsp
 ; CHECK-NEXT:    movq %rsp, %rbx
 ; CHECK-NEXT:    callq helper
+; CHECK-NEXT:    movq %rsp, %rcx
 ; CHECK-NEXT:    movl %eax, %eax
 ; CHECK-NEXT:    leaq 31(,%rax,4), %rax
-; CHECK-NEXT:    movabsq $34359738336, %rcx # imm = 0x7FFFFFFE0
-; CHECK-NEXT:    andq %rax, %rcx
-; CHECK-NEXT:    movq %rsp, %rax
-; CHECK-NEXT:    movq %rax, %rdx
-; CHECK-NEXT:    subq %rcx, %rdx
-; CHECK-NEXT:    negq %rcx
+; CHECK-NEXT:    andq $-32, %rax
+; CHECK-NEXT:    movq %rcx, %rdx
+; CHECK-NEXT:    subq %rax, %rdx
 ; CHECK-NEXT:    movq %rdx, %rsp
-; CHECK-NEXT:    movl $0, (%rax,%rcx)
+; CHECK-NEXT:    negq %rax
+; CHECK-NEXT:    movl $0, (%rcx,%rax)
 ; CHECK-NEXT:    leaq -8(%rbp), %rsp
 ; CHECK-NEXT:    popq %rbx
 ; CHECK-NEXT:    popq %rbp
Index: test/CodeGen/X86/zext-fold.ll
===================================================================
--- test/CodeGen/X86/zext-fold.ll
+++ test/CodeGen/X86/zext-fold.ll
@@ -6,7 +6,7 @@
 ; CHECK-LABEL: test1:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    movzbl {{[0-9]+}}(%esp), %eax
-; CHECK-NEXT:    andl $224, %eax
+; CHECK-NEXT:    andl $-32, %eax
 ; CHECK-NEXT:    retl
   %A = and i8 %x, -32
   %B = zext i8 %A to i32
@@ -19,7 +19,7 @@
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    movzbl {{[0-9]+}}(%esp), %eax
 ; CHECK-NEXT:    movl %eax, %ecx
-; CHECK-NEXT:    andl $224, %ecx
+; CHECK-NEXT:    andl $-32, %ecx
 ; CHECK-NEXT:    orl $63, %eax
 ; CHECK-NEXT:    addl %ecx, %eax
 ; CHECK-NEXT:    retl
@@ -41,7 +41,7 @@
 ; CHECK-NEXT:    movzbl {{[0-9]+}}(%esp), %eax
 ; CHECK-NEXT:    subl $8, %esp
 ; CHECK-NEXT:    pushl %eax
-; CHECK-NEXT:    andl $224, %eax
+; CHECK-NEXT:    andl $-32, %eax
 ; CHECK-NEXT:    pushl %eax
 ; CHECK-NEXT:    calll use
 ; CHECK-NEXT:    addl $28, %esp