From: Simon Pilgrim <llvm-dev@redking.me.uk>
Date: Thu, 8 Aug 2019 10:37:03 +0000 (+0000)
Subject: [TargetLowering] SimplifyDemandedBits - call SimplifyMultipleUseDemandedBits for... 
X-Git-Tag: llvmorg-11-init~12402
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=e2e366797ed51599c57e64417e616ccbd39c9901;p=platform%2Fupstream%2Fllvm.git

[TargetLowering] SimplifyDemandedBits - call SimplifyMultipleUseDemandedBits for ISD::EXTRACT_VECTOR_ELT

This patch attempts to peek through vectors based on the demanded bits/elt of a particular ISD::EXTRACT_VECTOR_ELT node, allowing us to avoid dependencies on ops that have no impact on the extract.

In particular this helps remove some unnecessary scalar->vector->scalar patterns.

The wasm shift patterns are annoying - @tlively has indicated that the wasm vector shift codegen are to be refactored in the near-term and isn't considered a major issue.

Differential Revision: https://reviews.llvm.org/D65887

llvm-svn: 368276
---

diff --git a/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp b/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
index 8884718..411d146 100644
--- a/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
@@ -1779,6 +1779,17 @@ bool TargetLowering::SimplifyDemandedBits(
                              Depth + 1))
       return true;
 
+    // Attempt to avoid multi-use ops if we don't need anything from them.
+    if (!DemandedSrcBits.isAllOnesValue() ||
+        !DemandedSrcElts.isAllOnesValue()) {
+      if (SDValue DemandedSrc = SimplifyMultipleUseDemandedBits(
+              Src, DemandedSrcBits, DemandedSrcElts, TLO.DAG, Depth + 1)) {
+        SDValue NewOp =
+            TLO.DAG.getNode(Op.getOpcode(), dl, VT, DemandedSrc, Idx);
+        return TLO.CombineTo(Op, NewOp);
+      }
+    }
+
     Known = Known2;
     if (BitWidth > EltBitWidth)
       Known = Known.zext(BitWidth, false /* => any extend */);
diff --git a/llvm/test/CodeGen/AArch64/vecreduce-and-legalization.ll b/llvm/test/CodeGen/AArch64/vecreduce-and-legalization.ll
index bf18626..375f220 100644
--- a/llvm/test/CodeGen/AArch64/vecreduce-and-legalization.ll
+++ b/llvm/test/CodeGen/AArch64/vecreduce-and-legalization.ll
@@ -105,17 +105,17 @@ define i8 @test_v9i8(<9 x i8> %a) nounwind {
 ; CHECK-NEXT:    mov v0.b[14], w8
 ; CHECK-NEXT:    mov v0.b[15], w8
 ; CHECK-NEXT:    ext v1.16b, v0.16b, v0.16b, #8
-; CHECK-NEXT:    and v0.8b, v0.8b, v1.8b
-; CHECK-NEXT:    umov w8, v0.b[1]
-; CHECK-NEXT:    umov w9, v0.b[0]
+; CHECK-NEXT:    and v1.8b, v0.8b, v1.8b
+; CHECK-NEXT:    umov w8, v1.b[1]
+; CHECK-NEXT:    umov w9, v1.b[0]
 ; CHECK-NEXT:    and w8, w9, w8
-; CHECK-NEXT:    umov w9, v0.b[2]
+; CHECK-NEXT:    umov w9, v1.b[2]
 ; CHECK-NEXT:    and w8, w8, w9
-; CHECK-NEXT:    umov w9, v0.b[3]
+; CHECK-NEXT:    umov w9, v1.b[3]
 ; CHECK-NEXT:    and w8, w8, w9
-; CHECK-NEXT:    umov w9, v0.b[4]
+; CHECK-NEXT:    umov w9, v1.b[4]
 ; CHECK-NEXT:    and w8, w8, w9
-; CHECK-NEXT:    umov w9, v0.b[5]
+; CHECK-NEXT:    umov w9, v1.b[5]
 ; CHECK-NEXT:    and w8, w8, w9
 ; CHECK-NEXT:    umov w9, v0.b[6]
 ; CHECK-NEXT:    and w8, w8, w9
@@ -132,9 +132,9 @@ define i32 @test_v3i32(<3 x i32> %a) nounwind {
 ; CHECK-NEXT:    mov w8, #-1
 ; CHECK-NEXT:    mov v0.s[3], w8
 ; CHECK-NEXT:    ext v1.16b, v0.16b, v0.16b, #8
-; CHECK-NEXT:    and v0.8b, v0.8b, v1.8b
+; CHECK-NEXT:    and v1.8b, v0.8b, v1.8b
 ; CHECK-NEXT:    mov w8, v0.s[1]
-; CHECK-NEXT:    fmov w9, s0
+; CHECK-NEXT:    fmov w9, s1
 ; CHECK-NEXT:    and w0, w9, w8
 ; CHECK-NEXT:    ret
   %b = call i32 @llvm.experimental.vector.reduce.and.v3i32(<3 x i32> %a)
diff --git a/llvm/test/CodeGen/ARM/dagcombine-anyexttozeroext.ll b/llvm/test/CodeGen/ARM/dagcombine-anyexttozeroext.ll
index de24681..08f6d21 100644
--- a/llvm/test/CodeGen/ARM/dagcombine-anyexttozeroext.ll
+++ b/llvm/test/CodeGen/ARM/dagcombine-anyexttozeroext.ll
@@ -1,3 +1,4 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple armv7 %s -o - | FileCheck %s
 
 define float @f(<4 x i16>* nocapture %in) {
@@ -63,8 +64,8 @@ define <4 x i32> @h(<4 x i8> *%in) {
   ret <4 x i32> %13
 }
 
+; FIXME: The vmov.u + sxt can convert to a vmov.s
 define float @i(<4 x i16>* nocapture %in) {
-  ; FIXME: The vmov.u + sxt can convert to a vmov.s
 ; CHECK-LABEL: i:
 ; CHECK:       @ %bb.0:
 ; CHECK-NEXT:    vldr d16, [r0]
@@ -95,8 +96,8 @@ define float @j(<8 x i8>* nocapture %in) {
   ret float %3
 }
 
-define float @k(<8 x i8>* nocapture %in) {
 ; FIXME: The vmov.u + sxt can convert to a vmov.s
+define float @k(<8 x i8>* nocapture %in) {
 ; CHECK-LABEL: k:
 ; CHECK:       @ %bb.0:
 ; CHECK-NEXT:    vldr d16, [r0]
@@ -113,17 +114,14 @@ define float @k(<8 x i8>* nocapture %in) {
 }
 
 define float @KnownUpperZero(<4 x i16> %v) {
-; FIXME: uxtb are not required
 ; CHECK-LABEL: KnownUpperZero:
 ; CHECK:       @ %bb.0:
-; CHECK-NEXT:    vmov.i16 d16, #0x3
-; CHECK-NEXT:    vmov d17, r0, r1
-; CHECK-NEXT:    vand d16, d17, d16
+; CHECK-NEXT:    vmov d16, r0, r1
 ; CHECK-NEXT:    vmov.u16 r0, d16[0]
 ; CHECK-NEXT:    vmov.u16 r1, d16[3]
-; CHECK-NEXT:    uxtb r0, r0
+; CHECK-NEXT:    and r0, r0, #3
 ; CHECK-NEXT:    vmov s0, r0
-; CHECK-NEXT:    uxtb r0, r1
+; CHECK-NEXT:    and r0, r1, #3
 ; CHECK-NEXT:    vmov s2, r0
 ; CHECK-NEXT:    vcvt.f32.s32 s0, s0
 ; CHECK-NEXT:    vcvt.f32.s32 s2, s2
diff --git a/llvm/test/CodeGen/WebAssembly/simd-arith.ll b/llvm/test/CodeGen/WebAssembly/simd-arith.ll
index 410b2e6..57bc86e 100644
--- a/llvm/test/CodeGen/WebAssembly/simd-arith.ll
+++ b/llvm/test/CodeGen/WebAssembly/simd-arith.ll
@@ -90,18 +90,18 @@ define <16 x i8> @shl_const_v16i8(<16 x i8> %v) {
 ; NO-SIMD128-NOT: i8x16
 ; SIMD128-NEXT: .functype shl_vec_v16i8 (v128, v128) -> (v128){{$}}
 ; SIMD128-NEXT: i8x16.extract_lane_s $push[[L0:[0-9]+]]=, $0, 0{{$}}
+; SIMD128-NEXT: i8x16.extract_lane_s $push[[L1:[0-9]+]]=, $1, 0{{$}}
 ; SIMD128-NEXT: i32.const $push[[M0:[0-9]+]]=, 7{{$}}
-; SIMD128-NEXT: i8x16.splat $push[[M1:[0-9]+]]=, $pop[[M0]]{{$}}
-; SIMD128-NEXT: v128.and $push[[M2:[0-9]+]]=, $1, $pop[[M1]]{{$}}
-; SIMD128-NEXT: local.tee $push[[M:[0-9]+]]=, $1=, $pop[[M2]]{{$}}
-; SIMD128-NEXT: i8x16.extract_lane_u $push[[L1:[0-9]+]]=, $pop[[M]], 0{{$}}
-; SIMD128-NEXT: i32.shl $push[[L2:[0-9]+]]=, $pop[[L0]], $pop[[L1]]{{$}}
-; SIMD128-NEXT: i8x16.splat $push[[L3:[0-9]+]]=, $pop[[L2]]{{$}}
+; SIMD128-NEXT: i32.and $push[[M1:[0-9]+]]=, $pop[[L1]], $pop[[M0]]{{$}}
+; SIMD128-NEXT: i32.shl $push[[M2:[0-9]+]]=, $pop[[L0]], $pop[[M1]]
+; SIMD128-NEXT: i8x16.splat $push[[M3:[0-9]+]]=, $pop[[M2]]
 ; Skip 14 lanes
 ; SIMD128:      i8x16.extract_lane_s $push[[L4:[0-9]+]]=, $0, 15{{$}}
-; SIMD128-NEXT: i8x16.extract_lane_u $push[[L5:[0-9]+]]=, $1, 15{{$}}
-; SIMD128-NEXT: i32.shl $push[[L6:[0-9]+]]=, $pop[[L4]], $pop[[L5]]{{$}}
-; SIMD128-NEXT: i8x16.replace_lane $push[[R:[0-9]+]]=, $pop[[L7:[0-9]+]], 15, $pop[[L6]]{{$}}
+; SIMD128-NEXT: i8x16.extract_lane_s $push[[L5:[0-9]+]]=, $1, 15{{$}}
+; SIMD128-NEXT: i32.const $push[[M4:[0-9]+]]=, 7{{$}}
+; SIMD128-NEXT: i32.and $push[[M5:[0-9]+]]=, $pop[[L5]], $pop[[M4]]{{$}}
+; SIMD128-NEXT: i32.shl $push[[M6:[0-9]+]]=, $pop[[L4]], $pop[[M5]]{{$}}
+; SIMD128-NEXT: i8x16.replace_lane $push[[R:[0-9]+]]=, $pop[[M7:[0-9]+]], 15, $pop[[M6]]{{$}}
 ; SIMD128-NEXT: return $pop[[R]]{{$}}
 define <16 x i8> @shl_vec_v16i8(<16 x i8> %v, <16 x i8> %x) {
   %a = shl <16 x i8> %v, %x
@@ -126,18 +126,18 @@ define <16 x i8> @shr_s_v16i8(<16 x i8> %v, i8 %x) {
 ; NO-SIMD128-NOT: i8x16
 ; SIMD128-NEXT: .functype shr_s_vec_v16i8 (v128, v128) -> (v128){{$}}
 ; SIMD128-NEXT: i8x16.extract_lane_s $push[[L0:[0-9]+]]=, $0, 0{{$}}
+; SIMD128-NEXT: i8x16.extract_lane_s $push[[L1:[0-9]+]]=, $1, 0{{$}}
 ; SIMD128-NEXT: i32.const $push[[M0:[0-9]+]]=, 7{{$}}
-; SIMD128-NEXT: i8x16.splat $push[[M1:[0-9]+]]=, $pop[[M0]]{{$}}
-; SIMD128-NEXT: v128.and $push[[M2:[0-9]+]]=, $1, $pop[[M1]]{{$}}
-; SIMD128-NEXT: local.tee $push[[M:[0-9]+]]=, $1=, $pop[[M2]]{{$}}
-; SIMD128-NEXT: i8x16.extract_lane_u $push[[L1:[0-9]+]]=, $pop[[M]], 0{{$}}
-; SIMD128-NEXT: i32.shr_s $push[[L2:[0-9]+]]=, $pop[[L0]], $pop[[L1]]{{$}}
-; SIMD128-NEXT: i8x16.splat $push[[L3:[0-9]+]]=, $pop[[L2]]{{$}}
+; SIMD128-NEXT: i32.and $push[[M1:[0-9]+]]=, $pop[[L1]], $pop[[M0]]{{$}}
+; SIMD128-NEXT: i32.shr_s $push[[M2:[0-9]+]]=, $pop[[L0]], $pop[[M1]]
+; SIMD128-NEXT: i8x16.splat $push[[M3:[0-9]+]]=, $pop[[M2]]
 ; Skip 14 lanes
-; SIMD128:      i8x16.extract_lane_s $push[[L0:[0-9]+]]=, $0, 15{{$}}
-; SIMD128-NEXT: i8x16.extract_lane_u $push[[L1:[0-9]+]]=, $1, 15{{$}}
-; SIMD128-NEXT: i32.shr_s $push[[L2:[0-9]+]]=, $pop[[L0]], $pop[[L1]]{{$}}
-; SIMD128-NEXT: i8x16.replace_lane $push[[R:[0-9]+]]=, $pop{{[0-9]+}}, 15, $pop[[L2]]{{$}}
+; SIMD128:      i8x16.extract_lane_s $push[[L4:[0-9]+]]=, $0, 15{{$}}
+; SIMD128-NEXT: i8x16.extract_lane_s $push[[L5:[0-9]+]]=, $1, 15{{$}}
+; SIMD128-NEXT: i32.const $push[[M4:[0-9]+]]=, 7{{$}}
+; SIMD128-NEXT: i32.and $push[[M5:[0-9]+]]=, $pop[[L5]], $pop[[M4]]{{$}}
+; SIMD128-NEXT: i32.shr_s $push[[M6:[0-9]+]]=, $pop[[L4]], $pop[[M5]]{{$}}
+; SIMD128-NEXT: i8x16.replace_lane $push[[R:[0-9]+]]=, $pop[[M7:[0-9]+]], 15, $pop[[M6]]{{$}}
 ; SIMD128-NEXT: return $pop[[R]]{{$}}
 define <16 x i8> @shr_s_vec_v16i8(<16 x i8> %v, <16 x i8> %x) {
   %a = ashr <16 x i8> %v, %x
@@ -162,18 +162,18 @@ define <16 x i8> @shr_u_v16i8(<16 x i8> %v, i8 %x) {
 ; NO-SIMD128-NOT: i8x16
 ; SIMD128-NEXT: .functype shr_u_vec_v16i8 (v128, v128) -> (v128){{$}}
 ; SIMD128-NEXT: i8x16.extract_lane_u $push[[L0:[0-9]+]]=, $0, 0{{$}}
+; SIMD128-NEXT: i8x16.extract_lane_s $push[[L1:[0-9]+]]=, $1, 0{{$}}
 ; SIMD128-NEXT: i32.const $push[[M0:[0-9]+]]=, 7{{$}}
-; SIMD128-NEXT: i8x16.splat $push[[M1:[0-9]+]]=, $pop[[M0]]{{$}}
-; SIMD128-NEXT: v128.and $push[[M2:[0-9]+]]=, $1, $pop[[M1]]{{$}}
-; SIMD128-NEXT: local.tee $push[[M:[0-9]+]]=, $1=, $pop[[M2]]{{$}}
-; SIMD128-NEXT: i8x16.extract_lane_u $push[[L1:[0-9]+]]=, $pop[[M]], 0{{$}}
-; SIMD128-NEXT: i32.shr_u $push[[L2:[0-9]+]]=, $pop[[L0]], $pop[[L1]]{{$}}
-; SIMD128-NEXT: i8x16.splat $push[[L3:[0-9]+]]=, $pop[[L2]]{{$}}
+; SIMD128-NEXT: i32.and $push[[M1:[0-9]+]]=, $pop[[L1]], $pop[[M0]]{{$}}
+; SIMD128-NEXT: i32.shr_u $push[[M2:[0-9]+]]=, $pop[[L0]], $pop[[M1]]
+; SIMD128-NEXT: i8x16.splat $push[[M3:[0-9]+]]=, $pop[[M2]]
 ; Skip 14 lanes
 ; SIMD128:      i8x16.extract_lane_u $push[[L4:[0-9]+]]=, $0, 15{{$}}
-; SIMD128-NEXT: i8x16.extract_lane_u $push[[L5:[0-9]+]]=, $1, 15{{$}}
-; SIMD128-NEXT: i32.shr_u $push[[L6:[0-9]+]]=, $pop[[L4]], $pop[[L5]]{{$}}
-; SIMD128-NEXT: i8x16.replace_lane $push[[R:[0-9]+]]=, $pop[[L7:[0-9]+]], 15, $pop[[L6]]{{$}}
+; SIMD128-NEXT: i8x16.extract_lane_s $push[[L5:[0-9]+]]=, $1, 15{{$}}
+; SIMD128-NEXT: i32.const $push[[M4:[0-9]+]]=, 7{{$}}
+; SIMD128-NEXT: i32.and $push[[M5:[0-9]+]]=, $pop[[L5]], $pop[[M4]]{{$}}
+; SIMD128-NEXT: i32.shr_u $push[[M6:[0-9]+]]=, $pop[[L4]], $pop[[M5]]{{$}}
+; SIMD128-NEXT: i8x16.replace_lane $push[[R:[0-9]+]]=, $pop[[M7:[0-9]+]], 15, $pop[[M6]]{{$}}
 ; SIMD128-NEXT: return $pop[[R]]{{$}}
 define <16 x i8> @shr_u_vec_v16i8(<16 x i8> %v, <16 x i8> %x) {
   %a = lshr <16 x i8> %v, %x
@@ -316,18 +316,18 @@ define <8 x i16> @shl_const_v8i16(<8 x i16> %v) {
 ; NO-SIMD128-NOT: i16x8
 ; SIMD128-NEXT: .functype shl_vec_v8i16 (v128, v128) -> (v128){{$}}
 ; SIMD128-NEXT: i16x8.extract_lane_s $push[[L0:[0-9]+]]=, $0, 0{{$}}
+; SIMD128-NEXT: i16x8.extract_lane_s $push[[L1:[0-9]+]]=, $1, 0{{$}}
 ; SIMD128-NEXT: i32.const $push[[M0:[0-9]+]]=, 15{{$}}
-; SIMD128-NEXT: i16x8.splat $push[[M1:[0-9]+]]=, $pop[[M0]]{{$}}
-; SIMD128-NEXT: v128.and $push[[M2:[0-9]+]]=, $1, $pop[[M1]]{{$}}
-; SIMD128-NEXT: local.tee $push[[M:[0-9]+]]=, $1=, $pop[[M2]]{{$}}
-; SIMD128-NEXT: i16x8.extract_lane_u $push[[L1:[0-9]+]]=, $pop[[M]], 0{{$}}
-; SIMD128-NEXT: i32.shl $push[[L2:[0-9]+]]=, $pop[[L0]], $pop[[L1]]{{$}}
-; SIMD128-NEXT: i16x8.splat $push[[L3:[0-9]+]]=, $pop[[L2]]{{$}}
+; SIMD128-NEXT: i32.and $push[[M1:[0-9]+]]=, $pop[[L1]], $pop[[M0]]{{$}}
+; SIMD128-NEXT: i32.shl $push[[M2:[0-9]+]]=, $pop[[L0]], $pop[[M1]]{{$}}
+; SIMD128-NEXT: i16x8.splat $push[[M3:[0-9]+]]=, $pop[[M2]]{{$}}
 ; Skip 6 lanes
 ; SIMD128:      i16x8.extract_lane_s $push[[L4:[0-9]+]]=, $0, 7{{$}}
-; SIMD128-NEXT: i16x8.extract_lane_u $push[[L5:[0-9]+]]=, $1, 7{{$}}
-; SIMD128-NEXT: i32.shl $push[[L6:[0-9]+]]=, $pop[[L4]], $pop[[L5]]{{$}}
-; SIMD128-NEXT: i16x8.replace_lane $push[[R:[0-9]+]]=, $pop[[L7:[0-9]+]], 7, $pop[[L6]]{{$}}
+; SIMD128-NEXT: i16x8.extract_lane_s $push[[L5:[0-9]+]]=, $1, 7{{$}}
+; SIMD128-NEXT: i32.const $push[[M4:[0-9]+]]=, 15{{$}}
+; SIMD128-NEXT: i32.and $push[[M5:[0-9]+]]=, $pop[[L5]], $pop[[M4]]{{$}}
+; SIMD128-NEXT: i32.shl $push[[M6:[0-9]+]]=, $pop[[L4]], $pop[[M5]]{{$}}
+; SIMD128-NEXT: i16x8.replace_lane $push[[R:[0-9]+]]=, $pop[[M7:[0-9]+]], 7, $pop[[M6]]{{$}}
 ; SIMD128-NEXT: return $pop[[R]]{{$}}
 define <8 x i16> @shl_vec_v8i16(<8 x i16> %v, <8 x i16> %x) {
   %a = shl <8 x i16> %v, %x
@@ -351,18 +351,18 @@ define <8 x i16> @shr_s_v8i16(<8 x i16> %v, i16 %x) {
 ; NO-SIMD128-NOT: i16x8
 ; SIMD128-NEXT: .functype shr_s_vec_v8i16 (v128, v128) -> (v128){{$}}
 ; SIMD128-NEXT: i16x8.extract_lane_s $push[[L0:[0-9]+]]=, $0, 0{{$}}
+; SIMD128-NEXT: i16x8.extract_lane_s $push[[L1:[0-9]+]]=, $1, 0{{$}}
 ; SIMD128-NEXT: i32.const $push[[M0:[0-9]+]]=, 15{{$}}
-; SIMD128-NEXT: i16x8.splat $push[[M1:[0-9]+]]=, $pop[[M0]]{{$}}
-; SIMD128-NEXT: v128.and $push[[M2:[0-9]+]]=, $1, $pop[[M1]]{{$}}
-; SIMD128-NEXT: local.tee $push[[M:[0-9]+]]=, $1=, $pop[[M2]]{{$}}
-; SIMD128-NEXT: i16x8.extract_lane_u $push[[L1:[0-9]+]]=, $pop[[M]], 0{{$}}
-; SIMD128-NEXT: i32.shr_s $push[[L2:[0-9]+]]=, $pop[[L0]], $pop[[L1]]{{$}}
-; SIMD128-NEXT: i16x8.splat $push[[L3:[0-9]+]]=, $pop[[L2]]{{$}}
+; SIMD128-NEXT: i32.and $push[[M1:[0-9]+]]=, $pop[[L1]], $pop[[M0]]{{$}}
+; SIMD128-NEXT: i32.shr_s $push[[M2:[0-9]+]]=, $pop[[L0]], $pop[[M1]]{{$}}
+; SIMD128-NEXT: i16x8.splat $push[[M3:[0-9]+]]=, $pop[[M2]]{{$}}
 ; Skip 6 lanes
-; SIMD128:      i16x8.extract_lane_s $push[[L0:[0-9]+]]=, $0, 7{{$}}
-; SIMD128-NEXT: i16x8.extract_lane_u $push[[L1:[0-9]+]]=, $1, 7{{$}}
-; SIMD128-NEXT: i32.shr_s $push[[L2:[0-9]+]]=, $pop[[L0]], $pop[[L1]]{{$}}
-; SIMD128-NEXT: i16x8.replace_lane $push[[R:[0-9]+]]=, $pop{{[0-9]+}}, 7, $pop[[L2]]{{$}}
+; SIMD128:      i16x8.extract_lane_s $push[[L4:[0-9]+]]=, $0, 7{{$}}
+; SIMD128-NEXT: i16x8.extract_lane_s $push[[L5:[0-9]+]]=, $1, 7{{$}}
+; SIMD128-NEXT: i32.const $push[[M4:[0-9]+]]=, 15{{$}}
+; SIMD128-NEXT: i32.and $push[[M5:[0-9]+]]=, $pop[[L5]], $pop[[M4]]{{$}}
+; SIMD128-NEXT: i32.shr_s $push[[M6:[0-9]+]]=, $pop[[L4]], $pop[[M5]]{{$}}
+; SIMD128-NEXT: i16x8.replace_lane $push[[R:[0-9]+]]=, $pop[[M7:[0-9]+]], 7, $pop[[M6]]{{$}}
 ; SIMD128-NEXT: return $pop[[R]]{{$}}
 define <8 x i16> @shr_s_vec_v8i16(<8 x i16> %v, <8 x i16> %x) {
   %a = ashr <8 x i16> %v, %x
@@ -386,18 +386,18 @@ define <8 x i16> @shr_u_v8i16(<8 x i16> %v, i16 %x) {
 ; NO-SIMD128-NOT: i16x8
 ; SIMD128-NEXT: .functype shr_u_vec_v8i16 (v128, v128) -> (v128){{$}}
 ; SIMD128-NEXT: i16x8.extract_lane_u $push[[L0:[0-9]+]]=, $0, 0{{$}}
+; SIMD128-NEXT: i16x8.extract_lane_s $push[[L1:[0-9]+]]=, $1, 0{{$}}
 ; SIMD128-NEXT: i32.const $push[[M0:[0-9]+]]=, 15{{$}}
-; SIMD128-NEXT: i16x8.splat $push[[M1:[0-9]+]]=, $pop[[M0]]{{$}}
-; SIMD128-NEXT: v128.and $push[[M2:[0-9]+]]=, $1, $pop[[M1]]{{$}}
-; SIMD128-NEXT: local.tee $push[[M:[0-9]+]]=, $1=, $pop[[M2]]{{$}}
-; SIMD128-NEXT: i16x8.extract_lane_u $push[[L1:[0-9]+]]=, $pop[[M]], 0{{$}}
-; SIMD128-NEXT: i32.shr_u $push[[L2:[0-9]+]]=, $pop[[L0]], $pop[[L1]]{{$}}
-; SIMD128-NEXT: i16x8.splat $push[[L3:[0-9]+]]=, $pop[[L2]]{{$}}
+; SIMD128-NEXT: i32.and $push[[M1:[0-9]+]]=, $pop[[L1]], $pop[[M0]]{{$}}
+; SIMD128-NEXT: i32.shr_u $push[[M2:[0-9]+]]=, $pop[[L0]], $pop[[M1]]{{$}}
+; SIMD128-NEXT: i16x8.splat $push[[M3:[0-9]+]]=, $pop[[M2]]{{$}}
 ; Skip 6 lanes
 ; SIMD128:      i16x8.extract_lane_u $push[[L4:[0-9]+]]=, $0, 7{{$}}
-; SIMD128-NEXT: i16x8.extract_lane_u $push[[L5:[0-9]+]]=, $1, 7{{$}}
-; SIMD128-NEXT: i32.shr_u $push[[L6:[0-9]+]]=, $pop[[L4]], $pop[[L5]]{{$}}
-; SIMD128-NEXT: i16x8.replace_lane $push[[R:[0-9]+]]=, $pop[[L7:[0-9]+]], 7, $pop[[L6]]{{$}}
+; SIMD128-NEXT: i16x8.extract_lane_s $push[[L5:[0-9]+]]=, $1, 7{{$}}
+; SIMD128-NEXT: i32.const $push[[M4:[0-9]+]]=, 15{{$}}
+; SIMD128-NEXT: i32.and $push[[M5:[0-9]+]]=, $pop[[L5]], $pop[[M4]]{{$}}
+; SIMD128-NEXT: i32.shr_u $push[[M6:[0-9]+]]=, $pop[[L4]], $pop[[M5]]{{$}}
+; SIMD128-NEXT: i16x8.replace_lane $push[[R:[0-9]+]]=, $pop[[M7:[0-9]+]], 7, $pop[[M6]]{{$}}
 ; SIMD128-NEXT: return $pop[[R]]{{$}}
 define <8 x i16> @shr_u_vec_v8i16(<8 x i16> %v, <8 x i16> %x) {
   %a = lshr <8 x i16> %v, %x
diff --git a/llvm/test/CodeGen/WebAssembly/simd-extended-extract.ll b/llvm/test/CodeGen/WebAssembly/simd-extended-extract.ll
index b00fcf6..149b184 100644
--- a/llvm/test/CodeGen/WebAssembly/simd-extended-extract.ll
+++ b/llvm/test/CodeGen/WebAssembly/simd-extended-extract.ll
@@ -21,16 +21,6 @@ define void @foo(<4 x i8>* %p) {
 ; CHECK:         .functype foo (i32) -> ()
 ; CHECK-NEXT:    i32.load8_u 0
 ; CHECK-NEXT:    i32x4.splat
-; CHECK-NEXT:    i32.load8_u 1
-; CHECK-NEXT:    i32x4.replace_lane 1
-; CHECK-NEXT:    i32.const 2
-; CHECK-NEXT:    i32.add
-; CHECK-NEXT:    i32.load8_u 0
-; CHECK-NEXT:    i32x4.replace_lane 2
-; CHECK-NEXT:    i32.const 3
-; CHECK-NEXT:    i32.add
-; CHECK-NEXT:    i32.load8_u 0
-; CHECK-NEXT:    i32x4.replace_lane 3
 ; CHECK-NEXT:    local.tee
 ; CHECK-NEXT:    i8x16.extract_lane_s 0
 ; CHECK-NEXT:    f64.convert_i32_s
@@ -40,6 +30,9 @@ define void @foo(<4 x i8>* %p) {
 ; CHECK-NEXT:    f64.add
 ; CHECK-NEXT:    f32.demote_f64
 ; CHECK-NEXT:    f32x4.splat
+; CHECK-NEXT:    i32.load8_u 1
+; CHECK-NEXT:    i32x4.replace_lane 1
+; CHECK-NEXT:    local.tee
 ; CHECK-NEXT:    i8x16.extract_lane_s 4
 ; CHECK-NEXT:    f64.convert_i32_s
 ; CHECK-NEXT:    f64.const 0x0p0
@@ -48,6 +41,11 @@ define void @foo(<4 x i8>* %p) {
 ; CHECK-NEXT:    f64.add
 ; CHECK-NEXT:    f32.demote_f64
 ; CHECK-NEXT:    f32x4.replace_lane 1
+; CHECK-NEXT:    i32.const 2
+; CHECK-NEXT:    i32.add
+; CHECK-NEXT:    i32.load8_u 0
+; CHECK-NEXT:    i32x4.replace_lane 2
+; CHECK-NEXT:    local.tee
 ; CHECK-NEXT:    i8x16.extract_lane_s 8
 ; CHECK-NEXT:    f64.convert_i32_s
 ; CHECK-NEXT:    f64.const 0x0p0
@@ -56,6 +54,10 @@ define void @foo(<4 x i8>* %p) {
 ; CHECK-NEXT:    f64.add
 ; CHECK-NEXT:    f32.demote_f64
 ; CHECK-NEXT:    f32x4.replace_lane 2
+; CHECK-NEXT:    i32.const 3
+; CHECK-NEXT:    i32.add
+; CHECK-NEXT:    i32.load8_u 0
+; CHECK-NEXT:    i32x4.replace_lane 3
 ; CHECK-NEXT:    i8x16.extract_lane_s 12
 ; CHECK-NEXT:    f64.convert_i32_s
 ; CHECK-NEXT:    f64.const 0x0p0
diff --git a/llvm/test/CodeGen/X86/promote-vec3.ll b/llvm/test/CodeGen/X86/promote-vec3.ll
index 61df546..19eda22 100644
--- a/llvm/test/CodeGen/X86/promote-vec3.ll
+++ b/llvm/test/CodeGen/X86/promote-vec3.ll
@@ -8,13 +8,9 @@
 define <3 x i16> @zext_i8(<3 x i8>) {
 ; SSE3-LABEL: zext_i8:
 ; SSE3:       # %bb.0:
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%esp), %eax
 ; SSE3-NEXT:    movzbl {{[0-9]+}}(%esp), %ecx
 ; SSE3-NEXT:    movzbl {{[0-9]+}}(%esp), %edx
-; SSE3-NEXT:    movzbl {{[0-9]+}}(%esp), %eax
-; SSE3-NEXT:    movd %eax, %xmm0
-; SSE3-NEXT:    pinsrw $1, %edx, %xmm0
-; SSE3-NEXT:    pinsrw $2, %ecx, %xmm0
-; SSE3-NEXT:    movd %xmm0, %eax
 ; SSE3-NEXT:    # kill: def $ax killed $ax killed $eax
 ; SSE3-NEXT:    # kill: def $dx killed $dx killed $edx
 ; SSE3-NEXT:    # kill: def $cx killed $cx killed $ecx
diff --git a/llvm/test/CodeGen/X86/vec_smulo.ll b/llvm/test/CodeGen/X86/vec_smulo.ll
index 385c556..ebe26c1 100644
--- a/llvm/test/CodeGen/X86/vec_smulo.ll
+++ b/llvm/test/CodeGen/X86/vec_smulo.ll
@@ -1790,7 +1790,6 @@ define <4 x i32> @smulo_v4i24(<4 x i24> %a0, <4 x i24> %a1, <4 x i24>* %p2) noun
 ; SSE2-NEXT:    psrad $8, %xmm1
 ; SSE2-NEXT:    pcmpeqd %xmm4, %xmm1
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[3,1,2,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm4[2,3,0,1]
 ; SSE2-NEXT:    psrad $31, %xmm4
 ; SSE2-NEXT:    pcmpeqd %xmm3, %xmm4
 ; SSE2-NEXT:    pcmpeqd %xmm3, %xmm3
@@ -1799,20 +1798,21 @@ define <4 x i32> @smulo_v4i24(<4 x i24> %a0, <4 x i24> %a1, <4 x i24>* %p2) noun
 ; SSE2-NEXT:    por %xmm4, %xmm1
 ; SSE2-NEXT:    movd %xmm0, %eax
 ; SSE2-NEXT:    movw %ax, (%rdi)
-; SSE2-NEXT:    movd %xmm2, %ecx
-; SSE2-NEXT:    movw %cx, 3(%rdi)
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; SSE2-NEXT:    movd %xmm0, %ecx
+; SSE2-NEXT:    movw %cx, 6(%rdi)
+; SSE2-NEXT:    movd %xmm2, %edx
+; SSE2-NEXT:    movw %dx, 3(%rdi)
 ; SSE2-NEXT:    shrl $16, %eax
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    shrl $16, %ecx
-; SSE2-NEXT:    movb %cl, 5(%rdi)
+; SSE2-NEXT:    movb %cl, 8(%rdi)
+; SSE2-NEXT:    shrl $16, %edx
+; SSE2-NEXT:    movb %dl, 5(%rdi)
 ; SSE2-NEXT:    movd %xmm5, %eax
 ; SSE2-NEXT:    movw %ax, 9(%rdi)
-; SSE2-NEXT:    movd %xmm6, %ecx
-; SSE2-NEXT:    movw %cx, 6(%rdi)
 ; SSE2-NEXT:    shrl $16, %eax
 ; SSE2-NEXT:    movb %al, 11(%rdi)
-; SSE2-NEXT:    shrl $16, %ecx
-; SSE2-NEXT:    movb %cl, 8(%rdi)
 ; SSE2-NEXT:    movdqa %xmm1, %xmm0
 ; SSE2-NEXT:    retq
 ;
@@ -1845,7 +1845,6 @@ define <4 x i32> @smulo_v4i24(<4 x i24> %a0, <4 x i24> %a1, <4 x i24>* %p2) noun
 ; SSSE3-NEXT:    psrad $8, %xmm1
 ; SSSE3-NEXT:    pcmpeqd %xmm4, %xmm1
 ; SSSE3-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[3,1,2,3]
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm6 = xmm4[2,3,0,1]
 ; SSSE3-NEXT:    psrad $31, %xmm4
 ; SSSE3-NEXT:    pcmpeqd %xmm3, %xmm4
 ; SSSE3-NEXT:    pcmpeqd %xmm3, %xmm3
@@ -1854,20 +1853,21 @@ define <4 x i32> @smulo_v4i24(<4 x i24> %a0, <4 x i24> %a1, <4 x i24>* %p2) noun
 ; SSSE3-NEXT:    por %xmm4, %xmm1
 ; SSSE3-NEXT:    movd %xmm0, %eax
 ; SSSE3-NEXT:    movw %ax, (%rdi)
-; SSSE3-NEXT:    movd %xmm2, %ecx
-; SSSE3-NEXT:    movw %cx, 3(%rdi)
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; SSSE3-NEXT:    movd %xmm0, %ecx
+; SSSE3-NEXT:    movw %cx, 6(%rdi)
+; SSSE3-NEXT:    movd %xmm2, %edx
+; SSSE3-NEXT:    movw %dx, 3(%rdi)
 ; SSSE3-NEXT:    shrl $16, %eax
 ; SSSE3-NEXT:    movb %al, 2(%rdi)
 ; SSSE3-NEXT:    shrl $16, %ecx
-; SSSE3-NEXT:    movb %cl, 5(%rdi)
+; SSSE3-NEXT:    movb %cl, 8(%rdi)
+; SSSE3-NEXT:    shrl $16, %edx
+; SSSE3-NEXT:    movb %dl, 5(%rdi)
 ; SSSE3-NEXT:    movd %xmm5, %eax
 ; SSSE3-NEXT:    movw %ax, 9(%rdi)
-; SSSE3-NEXT:    movd %xmm6, %ecx
-; SSSE3-NEXT:    movw %cx, 6(%rdi)
 ; SSSE3-NEXT:    shrl $16, %eax
 ; SSSE3-NEXT:    movb %al, 11(%rdi)
-; SSSE3-NEXT:    shrl $16, %ecx
-; SSSE3-NEXT:    movb %cl, 8(%rdi)
 ; SSSE3-NEXT:    movdqa %xmm1, %xmm0
 ; SSSE3-NEXT:    retq
 ;
diff --git a/llvm/test/CodeGen/X86/vec_umulo.ll b/llvm/test/CodeGen/X86/vec_umulo.ll
index c8146e3..455681c 100644
--- a/llvm/test/CodeGen/X86/vec_umulo.ll
+++ b/llvm/test/CodeGen/X86/vec_umulo.ll
@@ -1579,33 +1579,33 @@ define <4 x i32> @umulo_v4i24(<4 x i24> %a0, <4 x i24> %a1, <4 x i24>* %p2) noun
 ; SSE2-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
 ; SSE2-NEXT:    pxor %xmm4, %xmm4
 ; SSE2-NEXT:    pcmpeqd %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm5, %xmm5
+; SSE2-NEXT:    pxor %xmm5, %xmm3
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[0,2,2,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm2[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm1[3,1,2,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm1[2,3,0,1]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm2[0,2,2,3]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm6[0],xmm1[1],xmm6[1]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm1[3,1,2,3]
 ; SSE2-NEXT:    psrld $24, %xmm1
 ; SSE2-NEXT:    pcmpeqd %xmm4, %xmm1
-; SSE2-NEXT:    pcmpeqd %xmm4, %xmm4
-; SSE2-NEXT:    pxor %xmm4, %xmm3
-; SSE2-NEXT:    pxor %xmm4, %xmm1
+; SSE2-NEXT:    pxor %xmm5, %xmm1
 ; SSE2-NEXT:    por %xmm3, %xmm1
 ; SSE2-NEXT:    movd %xmm0, %eax
 ; SSE2-NEXT:    movw %ax, (%rdi)
-; SSE2-NEXT:    movd %xmm2, %ecx
-; SSE2-NEXT:    movw %cx, 3(%rdi)
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; SSE2-NEXT:    movd %xmm0, %ecx
+; SSE2-NEXT:    movw %cx, 6(%rdi)
+; SSE2-NEXT:    movd %xmm2, %edx
+; SSE2-NEXT:    movw %dx, 3(%rdi)
 ; SSE2-NEXT:    shrl $16, %eax
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    shrl $16, %ecx
-; SSE2-NEXT:    movb %cl, 5(%rdi)
-; SSE2-NEXT:    movd %xmm5, %eax
+; SSE2-NEXT:    movb %cl, 8(%rdi)
+; SSE2-NEXT:    shrl $16, %edx
+; SSE2-NEXT:    movb %dl, 5(%rdi)
+; SSE2-NEXT:    movd %xmm6, %eax
 ; SSE2-NEXT:    movw %ax, 9(%rdi)
-; SSE2-NEXT:    movd %xmm6, %ecx
-; SSE2-NEXT:    movw %cx, 6(%rdi)
 ; SSE2-NEXT:    shrl $16, %eax
 ; SSE2-NEXT:    movb %al, 11(%rdi)
-; SSE2-NEXT:    shrl $16, %ecx
-; SSE2-NEXT:    movb %cl, 8(%rdi)
 ; SSE2-NEXT:    movdqa %xmm1, %xmm0
 ; SSE2-NEXT:    retq
 ;
@@ -1623,33 +1623,33 @@ define <4 x i32> @umulo_v4i24(<4 x i24> %a0, <4 x i24> %a1, <4 x i24>* %p2) noun
 ; SSSE3-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
 ; SSSE3-NEXT:    pxor %xmm4, %xmm4
 ; SSSE3-NEXT:    pcmpeqd %xmm4, %xmm3
+; SSSE3-NEXT:    pcmpeqd %xmm5, %xmm5
+; SSSE3-NEXT:    pxor %xmm5, %xmm3
 ; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[0,2,2,3]
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm5 = xmm2[0,2,2,3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1]
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm5 = xmm1[3,1,2,3]
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm6 = xmm1[2,3,0,1]
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm6 = xmm2[0,2,2,3]
+; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm6[0],xmm1[1],xmm6[1]
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm6 = xmm1[3,1,2,3]
 ; SSSE3-NEXT:    psrld $24, %xmm1
 ; SSSE3-NEXT:    pcmpeqd %xmm4, %xmm1
-; SSSE3-NEXT:    pcmpeqd %xmm4, %xmm4
-; SSSE3-NEXT:    pxor %xmm4, %xmm3
-; SSSE3-NEXT:    pxor %xmm4, %xmm1
+; SSSE3-NEXT:    pxor %xmm5, %xmm1
 ; SSSE3-NEXT:    por %xmm3, %xmm1
 ; SSSE3-NEXT:    movd %xmm0, %eax
 ; SSSE3-NEXT:    movw %ax, (%rdi)
-; SSSE3-NEXT:    movd %xmm2, %ecx
-; SSSE3-NEXT:    movw %cx, 3(%rdi)
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; SSSE3-NEXT:    movd %xmm0, %ecx
+; SSSE3-NEXT:    movw %cx, 6(%rdi)
+; SSSE3-NEXT:    movd %xmm2, %edx
+; SSSE3-NEXT:    movw %dx, 3(%rdi)
 ; SSSE3-NEXT:    shrl $16, %eax
 ; SSSE3-NEXT:    movb %al, 2(%rdi)
 ; SSSE3-NEXT:    shrl $16, %ecx
-; SSSE3-NEXT:    movb %cl, 5(%rdi)
-; SSSE3-NEXT:    movd %xmm5, %eax
+; SSSE3-NEXT:    movb %cl, 8(%rdi)
+; SSSE3-NEXT:    shrl $16, %edx
+; SSSE3-NEXT:    movb %dl, 5(%rdi)
+; SSSE3-NEXT:    movd %xmm6, %eax
 ; SSSE3-NEXT:    movw %ax, 9(%rdi)
-; SSSE3-NEXT:    movd %xmm6, %ecx
-; SSSE3-NEXT:    movw %cx, 6(%rdi)
 ; SSSE3-NEXT:    shrl $16, %eax
 ; SSSE3-NEXT:    movb %al, 11(%rdi)
-; SSSE3-NEXT:    shrl $16, %ecx
-; SSSE3-NEXT:    movb %cl, 8(%rdi)
 ; SSSE3-NEXT:    movdqa %xmm1, %xmm0
 ; SSSE3-NEXT:    retq
 ;
diff --git a/llvm/test/CodeGen/X86/vector-reduce-mul.ll b/llvm/test/CodeGen/X86/vector-reduce-mul.ll
index 424e93f..5086889 100644
--- a/llvm/test/CodeGen/X86/vector-reduce-mul.ll
+++ b/llvm/test/CodeGen/X86/vector-reduce-mul.ll
@@ -1739,14 +1739,11 @@ define i8 @test_v16i8(<16 x i8> %a0) {
 ; SSE2-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[2,3,4,5,6,7,8,9,10,11,12,13,14,15],zero,zero
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; SSE2-NEXT:    pmullw %xmm2, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm1, %xmm2
-; SSE2-NEXT:    packuswb %xmm3, %xmm2
-; SSE2-NEXT:    psrlw $8, %xmm2
-; SSE2-NEXT:    pmullw %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm1, %xmm2
-; SSE2-NEXT:    packuswb %xmm0, %xmm2
-; SSE2-NEXT:    movd %xmm2, %eax
+; SSE2-NEXT:    pand %xmm0, %xmm1
+; SSE2-NEXT:    packuswb %xmm3, %xmm1
+; SSE2-NEXT:    psrlw $8, %xmm1
+; SSE2-NEXT:    pmullw %xmm0, %xmm1
+; SSE2-NEXT:    movd %xmm1, %eax
 ; SSE2-NEXT:    # kill: def $al killed $al killed $eax
 ; SSE2-NEXT:    retq
 ;
@@ -1954,14 +1951,11 @@ define i8 @test_v32i8(<32 x i8> %a0) {
 ; SSE2-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[2,3,4,5,6,7,8,9,10,11,12,13,14,15],zero,zero
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; SSE2-NEXT:    pmullw %xmm2, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm1, %xmm2
-; SSE2-NEXT:    packuswb %xmm3, %xmm2
-; SSE2-NEXT:    psrlw $8, %xmm2
-; SSE2-NEXT:    pmullw %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm1, %xmm2
-; SSE2-NEXT:    packuswb %xmm0, %xmm2
-; SSE2-NEXT:    movd %xmm2, %eax
+; SSE2-NEXT:    pand %xmm0, %xmm1
+; SSE2-NEXT:    packuswb %xmm3, %xmm1
+; SSE2-NEXT:    psrlw $8, %xmm1
+; SSE2-NEXT:    pmullw %xmm0, %xmm1
+; SSE2-NEXT:    movd %xmm1, %eax
 ; SSE2-NEXT:    # kill: def $al killed $al killed $eax
 ; SSE2-NEXT:    retq
 ;
@@ -2223,14 +2217,11 @@ define i8 @test_v64i8(<64 x i8> %a0) {
 ; SSE2-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[2,3,4,5,6,7,8,9,10,11,12,13,14,15],zero,zero
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; SSE2-NEXT:    pmullw %xmm2, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm1, %xmm2
-; SSE2-NEXT:    packuswb %xmm3, %xmm2
-; SSE2-NEXT:    psrlw $8, %xmm2
-; SSE2-NEXT:    pmullw %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm1, %xmm2
-; SSE2-NEXT:    packuswb %xmm0, %xmm2
-; SSE2-NEXT:    movd %xmm2, %eax
+; SSE2-NEXT:    pand %xmm0, %xmm1
+; SSE2-NEXT:    packuswb %xmm3, %xmm1
+; SSE2-NEXT:    psrlw $8, %xmm1
+; SSE2-NEXT:    pmullw %xmm0, %xmm1
+; SSE2-NEXT:    movd %xmm1, %eax
 ; SSE2-NEXT:    # kill: def $al killed $al killed $eax
 ; SSE2-NEXT:    retq
 ;
@@ -2595,14 +2586,11 @@ define i8 @test_v128i8(<128 x i8> %a0) {
 ; SSE2-NEXT:    psrldq {{.*#+}} xmm1 = xmm1[2,3,4,5,6,7,8,9,10,11,12,13,14,15],zero,zero
 ; 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:    pmullw %xmm2, %xmm1
-; SSE2-NEXT:    movdqa %xmm1, %xmm2
-; SSE2-NEXT:    pand %xmm0, %xmm2
-; SSE2-NEXT:    packuswb %xmm3, %xmm2
-; SSE2-NEXT:    psrlw $8, %xmm2
-; SSE2-NEXT:    pmullw %xmm1, %xmm2
-; SSE2-NEXT:    pand %xmm0, %xmm2
-; SSE2-NEXT:    packuswb %xmm0, %xmm2
-; SSE2-NEXT:    movd %xmm2, %eax
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    packuswb %xmm3, %xmm0
+; SSE2-NEXT:    psrlw $8, %xmm0
+; SSE2-NEXT:    pmullw %xmm1, %xmm0
+; SSE2-NEXT:    movd %xmm0, %eax
 ; SSE2-NEXT:    # kill: def $al killed $al killed $eax
 ; SSE2-NEXT:    retq
 ;
diff --git a/llvm/test/CodeGen/X86/xor.ll b/llvm/test/CodeGen/X86/xor.ll
index 654382f..e836b00 100644
--- a/llvm/test/CodeGen/X86/xor.ll
+++ b/llvm/test/CodeGen/X86/xor.ll
@@ -407,13 +407,10 @@ define i32 @PR17487(i1 %tobool) {
 ; X32-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; X32-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
 ; X32-NEXT:    pandn {{\.LCPI.*}}, %xmm0
-; X32-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[3,1,2,3]
-; X32-NEXT:    movd %xmm1, %ecx
 ; X32-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
-; X32-NEXT:    movd %xmm0, %edx
-; X32-NEXT:    xorl $1, %edx
+; X32-NEXT:    movd %xmm0, %ecx
 ; X32-NEXT:    xorl %eax, %eax
-; X32-NEXT:    orl %ecx, %edx
+; X32-NEXT:    cmpl $1, %ecx
 ; X32-NEXT:    setne %al
 ; X32-NEXT:    retl
 ;