diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -13332,12 +13332,20 @@
   SmallVector<int, 32> PermuteMask(NumElts, -1);
   for (int Elt = 0; Elt != NumElts; ++Elt) {
     int M = Mask[Elt];
+    if (M < 0)
+      continue;
+    int NormM = M;
     if (NumElts <= M)
-      PermuteMask[Elt] = NumLaneElts * ((M - NumElts) / NumLaneElts) +
-                         (2 * (M % NumHalfLaneElts)) + 1;
-    else if (0 <= M)
-      PermuteMask[Elt] =
-          NumLaneElts * (M / NumLaneElts) + (2 * (M % NumHalfLaneElts));
+      NormM -= NumElts;
+    bool IsFirstOp = M < NumElts;
+    int BaseMaskElt =
+        NumLaneElts * (NormM / NumLaneElts) + (2 * (NormM % NumHalfLaneElts));
+    if ((IsFirstOp && V1 == Ops[0]) || (!IsFirstOp && V2 == Ops[0]))
+      PermuteMask[Elt] = BaseMaskElt;
+    else if ((IsFirstOp && V1 == Ops[1]) || (!IsFirstOp && V2 == Ops[1]))
+      PermuteMask[Elt] = BaseMaskElt + 1;
+    assert(PermuteMask[Elt] != -1 &&
+           "Input mask element is defined but failed to assign permute mask");
   }
 
   unsigned UnpckOp = MatchLo ? X86ISD::UNPCKL : X86ISD::UNPCKH;
diff --git a/llvm/test/CodeGen/X86/pr62242.ll b/llvm/test/CodeGen/X86/pr62242.ll
new file mode 100644
--- /dev/null
+++ b/llvm/test/CodeGen/X86/pr62242.ll
@@ -0,0 +1,14 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefixes=AVX2
+
+define <4 x i32> @unpck_permute_canonicalized_mask(<4 x i32> %a, <4 x i32> %b) {
+; AVX2-LABEL: unpck_permute_canonicalized_mask:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vunpckhps {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; AVX2-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,3,2,1]
+; AVX2-NEXT:    retq
+  ; This mask will get canonicalized to vector_shuffle<6, 3, -1, 2> %b, %a.
+  ; Make sure the generated permute masks are still correct.
+  %shuffle = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 2, i32 7, i32 undef, i32 6>
+  ret <4 x i32> %shuffle
+}
diff --git a/llvm/test/CodeGen/X86/vector-interleaved-store-i32-stride-5.ll b/llvm/test/CodeGen/X86/vector-interleaved-store-i32-stride-5.ll
--- a/llvm/test/CodeGen/X86/vector-interleaved-store-i32-stride-5.ll
+++ b/llvm/test/CodeGen/X86/vector-interleaved-store-i32-stride-5.ll
@@ -165,7 +165,7 @@
 ; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm3, %ymm2, %ymm5
 ; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm2, %ymm3, %ymm6
 ; AVX1-ONLY-NEXT:    vunpcklps {{.*#+}} ymm5 = ymm5[0],ymm6[0],ymm5[1],ymm6[1],ymm5[4],ymm6[4],ymm5[5],ymm6[5]
-; AVX1-ONLY-NEXT:    vpermilps {{.*#+}} ymm5 = ymm5[u,u,1,0,u,u,u,6]
+; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} ymm5 = ymm5[0,0,2,3]
 ; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm1
 ; AVX1-ONLY-NEXT:    vunpcklps {{.*#+}} ymm7 = ymm4[0],ymm1[0],ymm4[1],ymm1[1],ymm4[4],ymm1[4],ymm4[5],ymm1[5]
 ; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} ymm7 = ymm7[0,0,3,3]
diff --git a/llvm/test/CodeGen/X86/vector-interleaved-store-i8-stride-5.ll b/llvm/test/CodeGen/X86/vector-interleaved-store-i8-stride-5.ll
--- a/llvm/test/CodeGen/X86/vector-interleaved-store-i8-stride-5.ll
+++ b/llvm/test/CodeGen/X86/vector-interleaved-store-i8-stride-5.ll
@@ -205,12 +205,12 @@
 define void @store_i8_stride5_vf8(ptr %in.vecptr0, ptr %in.vecptr1, ptr %in.vecptr2, ptr %in.vecptr3, ptr %in.vecptr4, ptr %out.vec) nounwind {
 ; SSE-LABEL: store_i8_stride5_vf8:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movq {{.*#+}} xmm2 = mem[0],zero
 ; SSE-NEXT:    movq {{.*#+}} xmm1 = mem[0],zero
+; SSE-NEXT:    movq {{.*#+}} xmm2 = mem[0],zero
 ; SSE-NEXT:    movq {{.*#+}} xmm4 = mem[0],zero
 ; SSE-NEXT:    movq {{.*#+}} xmm3 = mem[0],zero
 ; SSE-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
-; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm2[1,1,1,1]
+; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm1[1,1,1,1]
 ; SSE-NEXT:    movdqa {{.*#+}} xmm8 = [65535,65535,0,65535,0,65535,65535,0]
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm3[2,1,2,3]
@@ -225,8 +225,8 @@
 ; SSE-NEXT:    movdqa {{.*#+}} xmm6 = [255,0,255,255,255,255,0,255,255,255,255,0,255,255,255,255]
 ; SSE-NEXT:    movdqa %xmm6, %xmm10
 ; SSE-NEXT:    pandn %xmm5, %xmm10
-; SSE-NEXT:    movdqa %xmm1, %xmm7
-; SSE-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm1[0],xmm7[1],xmm1[1],xmm7[2],xmm1[2],xmm7[3],xmm1[3],xmm7[4],xmm1[4],xmm7[5],xmm1[5],xmm7[6],xmm1[6],xmm7[7],xmm1[7]
+; SSE-NEXT:    movdqa %xmm2, %xmm7
+; SSE-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm2[0],xmm7[1],xmm2[1],xmm7[2],xmm2[2],xmm7[3],xmm2[3],xmm7[4],xmm2[4],xmm7[5],xmm2[5],xmm7[6],xmm2[6],xmm7[7],xmm2[7]
 ; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm7[2,1,2,3]
 ; SSE-NEXT:    pshuflw {{.*#+}} xmm5 = xmm5[3,1,0,3,4,5,6,7]
 ; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,4,5,6,6]
@@ -249,7 +249,7 @@
 ; SSE-NEXT:    pandn %xmm10, %xmm8
 ; SSE-NEXT:    por %xmm9, %xmm8
 ; SSE-NEXT:    movdqa {{.*#+}} xmm9 = [255,255,0,0,255,255,255,0,0,255,255,255,0,0,255,255]
-; SSE-NEXT:    pshufd {{.*#+}} xmm10 = xmm2[0,0,0,0]
+; SSE-NEXT:    pshufd {{.*#+}} xmm10 = xmm1[0,0,0,0]
 ; SSE-NEXT:    pand %xmm6, %xmm10
 ; SSE-NEXT:    pshuflw {{.*#+}} xmm7 = xmm7[0,1,2,2,4,5,6,7]
 ; SSE-NEXT:    pshufd {{.*#+}} xmm7 = xmm7[0,0,1,3]
@@ -271,8 +271,9 @@
 ; SSE-NEXT:    por %xmm4, %xmm7
 ; SSE-NEXT:    movdqa {{.*#+}} xmm3 = [255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255]
 ; SSE-NEXT:    pand %xmm3, %xmm7
-; 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:    psrldq {{.*#+}} xmm1 = xmm1[11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; 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:    pshufhw {{.*#+}} xmm1 = xmm2[0,1,2,3,7,7,7,7]
+; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,2,2,2]
 ; SSE-NEXT:    pandn %xmm1, %xmm3
 ; SSE-NEXT:    por %xmm7, %xmm3
 ; SSE-NEXT:    movdqa {{.*#+}} xmm1 = [255,255,0,255,255,255,255,0,255,255,255,255,255,255,255,255]
diff --git a/llvm/test/CodeGen/X86/vector-interleaved-store-i8-stride-7.ll b/llvm/test/CodeGen/X86/vector-interleaved-store-i8-stride-7.ll
--- a/llvm/test/CodeGen/X86/vector-interleaved-store-i8-stride-7.ll
+++ b/llvm/test/CodeGen/X86/vector-interleaved-store-i8-stride-7.ll
@@ -288,35 +288,34 @@
 ; SSE-LABEL: store_i8_stride7_vf8:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movq {{[0-9]+}}(%rsp), %rax
-; SSE-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
-; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT:    movq {{.*#+}} xmm4 = mem[0],zero
 ; SSE-NEXT:    movq {{.*#+}} xmm10 = mem[0],zero
-; SSE-NEXT:    movq {{.*#+}} xmm1 = mem[0],zero
-; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE-NEXT:    movq {{.*#+}} xmm5 = mem[0],zero
+; SSE-NEXT:    movdqa %xmm10, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT:    movq {{.*#+}} xmm3 = mem[0],zero
+; SSE-NEXT:    movq {{.*#+}} xmm7 = mem[0],zero
+; SSE-NEXT:    movdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
 ; SSE-NEXT:    movq {{.*#+}} xmm2 = mem[0],zero
 ; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
 ; SSE-NEXT:    movq {{.*#+}} xmm6 = mem[0],zero
-; SSE-NEXT:    movq {{.*#+}} xmm4 = mem[0],zero
-; SSE-NEXT:    punpcklqdq {{.*#+}} xmm4 = xmm4[0],xmm6[0]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm6 = xmm1[0,0,0,0,4,5,6,7]
+; SSE-NEXT:    movq {{.*#+}} xmm5 = mem[0],zero
+; SSE-NEXT:    punpcklqdq {{.*#+}} xmm5 = xmm5[0],xmm6[0]
+; SSE-NEXT:    pshuflw {{.*#+}} xmm6 = xmm3[0,0,0,0,4,5,6,7]
+; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
 ; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[0,0,0,0]
 ; SSE-NEXT:    movdqa {{.*#+}} xmm11 = [255,255,255,0,255,255,255,255,255,255,0,255,255,255,255,255]
 ; SSE-NEXT:    pand %xmm11, %xmm6
-; SSE-NEXT:    movdqa %xmm5, %xmm7
-; SSE-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm5[0],xmm7[1],xmm5[1],xmm7[2],xmm5[2],xmm7[3],xmm5[3],xmm7[4],xmm5[4],xmm7[5],xmm5[5],xmm7[6],xmm5[6],xmm7[7],xmm5[7]
+; SSE-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; SSE-NEXT:    pshuflw {{.*#+}} xmm8 = xmm7[0,0,2,1,4,5,6,7]
 ; SSE-NEXT:    pshufd {{.*#+}} xmm8 = xmm8[0,1,1,3]
 ; SSE-NEXT:    pandn %xmm8, %xmm11
 ; SSE-NEXT:    por %xmm6, %xmm11
 ; SSE-NEXT:    movdqa {{.*#+}} xmm9 = [255,255,0,0,255,255,255,255,255,0,0,255,255,255,255,255]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm6 = xmm0[0,0,2,1,4,5,6,7]
+; SSE-NEXT:    pshuflw {{.*#+}} xmm6 = xmm4[0,0,2,1,4,5,6,7]
 ; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[0,0,2,1]
 ; SSE-NEXT:    movdqa {{.*#+}} xmm8 = [255,0,255,255,255,255,255,255,0,255,255,255,255,255,255,0]
 ; SSE-NEXT:    pand %xmm8, %xmm6
-; SSE-NEXT:    movdqa %xmm10, %xmm3
-; SSE-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm10[0],xmm3[1],xmm10[1],xmm3[2],xmm10[2],xmm3[3],xmm10[3],xmm3[4],xmm10[4],xmm3[5],xmm10[5],xmm3[6],xmm10[6],xmm3[7],xmm10[7]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm12 = xmm3[0,2,1,3,4,5,6,7]
+; SSE-NEXT:    punpcklbw {{.*#+}} xmm10 = xmm10[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE-NEXT:    pshuflw {{.*#+}} xmm12 = xmm10[0,2,1,3,4,5,6,7]
 ; SSE-NEXT:    pshufd {{.*#+}} xmm12 = xmm12[0,1,1,0]
 ; SSE-NEXT:    movdqa %xmm8, %xmm13
 ; SSE-NEXT:    pandn %xmm12, %xmm13
@@ -325,9 +324,8 @@
 ; SSE-NEXT:    pandn %xmm11, %xmm9
 ; SSE-NEXT:    por %xmm13, %xmm9
 ; SSE-NEXT:    pxor %xmm6, %xmm6
-; SSE-NEXT:    movdqa %xmm4, %xmm15
-; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE-NEXT:    movdqa %xmm4, %xmm12
+; SSE-NEXT:    movdqa %xmm5, %xmm12
+; SSE-NEXT:    movdqa %xmm5, %xmm15
 ; SSE-NEXT:    punpckhbw {{.*#+}} xmm15 = xmm15[8],xmm6[8],xmm15[9],xmm6[9],xmm15[10],xmm6[10],xmm15[11],xmm6[11],xmm15[12],xmm6[12],xmm15[13],xmm6[13],xmm15[14],xmm6[14],xmm15[15],xmm6[15]
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm12 = xmm12[0],xmm6[0],xmm12[1],xmm6[1],xmm12[2],xmm6[2],xmm12[3],xmm6[3],xmm12[4],xmm6[4],xmm12[5],xmm6[5],xmm12[6],xmm6[6],xmm12[7],xmm6[7]
 ; SSE-NEXT:    movdqa %xmm12, %xmm13
@@ -351,20 +349,18 @@
 ; SSE-NEXT:    pshufd {{.*#+}} xmm9 = xmm7[2,2,3,3]
 ; SSE-NEXT:    movdqa %xmm1, %xmm11
 ; SSE-NEXT:    pandn %xmm9, %xmm11
-; SSE-NEXT:    pshuflw $246, {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Folded Reload
-; SSE-NEXT:    # xmm9 = mem[2,1,3,3,4,5,6,7]
+; SSE-NEXT:    pshuflw {{.*#+}} xmm9 = xmm3[2,1,3,3,4,5,6,7]
 ; SSE-NEXT:    pshufd {{.*#+}} xmm14 = xmm9[0,0,2,1]
 ; SSE-NEXT:    pand %xmm1, %xmm14
 ; SSE-NEXT:    por %xmm11, %xmm14
 ; SSE-NEXT:    movdqa {{.*#+}} xmm9 = [255,255,255,255,255,0,0,255,255,255,255,255,0,0,255,255]
 ; SSE-NEXT:    movdqa %xmm9, %xmm11
 ; SSE-NEXT:    pandn %xmm14, %xmm11
-; SSE-NEXT:    pshufhw {{.*#+}} xmm14 = xmm3[0,1,2,3,5,6,6,7]
+; SSE-NEXT:    pshufhw {{.*#+}} xmm14 = xmm10[0,1,2,3,5,6,6,7]
 ; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm14[2,2,2,2]
 ; SSE-NEXT:    movdqa {{.*#+}} xmm0 = [255,255,255,255,0,255,255,255,255,255,255,0,255,255,255,255]
 ; SSE-NEXT:    movdqa %xmm0, %xmm14
 ; SSE-NEXT:    pandn %xmm2, %xmm14
-; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
 ; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm4[0,2,2,3,4,5,6,7]
 ; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,1,1,3]
 ; SSE-NEXT:    pand %xmm0, %xmm2
@@ -399,19 +395,19 @@
 ; SSE-NEXT:    pshufd {{.*#+}} xmm11 = xmm11[0,0,0,0]
 ; SSE-NEXT:    pandn %xmm11, %xmm0
 ; SSE-NEXT:    por %xmm2, %xmm0
-; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[0,1,2,2]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm3 = xmm4[1,1,2,3,4,5,6,7]
-; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,0,2,1]
-; SSE-NEXT:    pand %xmm1, %xmm3
+; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm10[0,1,2,2]
+; SSE-NEXT:    pshuflw {{.*#+}} xmm10 = xmm4[1,1,2,3,4,5,6,7]
+; SSE-NEXT:    pshufd {{.*#+}} xmm10 = xmm10[0,0,2,1]
+; SSE-NEXT:    pand %xmm1, %xmm10
 ; SSE-NEXT:    pandn %xmm2, %xmm1
-; SSE-NEXT:    por %xmm3, %xmm1
-; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
-; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm11[1,1,2,2,4,5,6,7]
+; SSE-NEXT:    por %xmm10, %xmm1
+; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
+; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm3[1,1,2,2,4,5,6,7]
 ; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,0,2,1]
 ; SSE-NEXT:    pand %xmm8, %xmm2
-; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm7[1,1,2,1]
-; SSE-NEXT:    pshufhw {{.*#+}} xmm3 = xmm3[0,1,2,3,7,5,6,4]
-; SSE-NEXT:    pandn %xmm3, %xmm8
+; SSE-NEXT:    pshufd {{.*#+}} xmm7 = xmm7[1,1,2,1]
+; SSE-NEXT:    pshufhw {{.*#+}} xmm7 = xmm7[0,1,2,3,7,5,6,4]
+; SSE-NEXT:    pandn %xmm7, %xmm8
 ; SSE-NEXT:    por %xmm2, %xmm8
 ; SSE-NEXT:    pand %xmm9, %xmm8
 ; SSE-NEXT:    pandn %xmm1, %xmm9
@@ -420,27 +416,31 @@
 ; SSE-NEXT:    pand %xmm1, %xmm9
 ; SSE-NEXT:    pandn %xmm0, %xmm1
 ; SSE-NEXT:    por %xmm1, %xmm9
-; SSE-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm11[0],xmm5[1],xmm11[1],xmm5[2],xmm11[2],xmm5[3],xmm11[3],xmm5[4],xmm11[4],xmm5[5],xmm11[5],xmm5[6],xmm11[6],xmm5[7],xmm11[7]
-; SSE-NEXT:    psrldq {{.*#+}} xmm5 = xmm5[11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; SSE-NEXT:    movdqa {{.*#+}} xmm0 = [255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255]
-; SSE-NEXT:    pandn %xmm5, %xmm0
-; SSE-NEXT:    punpcklbw {{.*#+}} xmm10 = xmm10[0],xmm4[0],xmm10[1],xmm4[1],xmm10[2],xmm4[2],xmm10[3],xmm4[3],xmm10[4],xmm4[4],xmm10[5],xmm4[5],xmm10[6],xmm4[6],xmm10[7],xmm4[7]
-; SSE-NEXT:    pshufd $231, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; SSE-NEXT:    # xmm1 = mem[3,1,2,3]
-; SSE-NEXT:    psrldq {{.*#+}} xmm10 = xmm10[13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; SSE-NEXT:    por %xmm0, %xmm10
-; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm13[3,3,3,3,4,5,6,7]
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255]
-; SSE-NEXT:    pand %xmm2, %xmm0
-; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[1,3,2,3,4,5,6,7]
-; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,1,3,1,4,5,6,7]
-; SSE-NEXT:    pandn %xmm1, %xmm2
-; SSE-NEXT:    por %xmm0, %xmm2
+; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; 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:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,7,7,7]
+; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,2,2,2]
+; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
+; SSE-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]
+; SSE-NEXT:    movdqa {{.*#+}} xmm1 = [255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255]
+; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,7,7,7,7]
+; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[2,2,2,2]
+; SSE-NEXT:    pand %xmm1, %xmm2
+; SSE-NEXT:    pandn %xmm0, %xmm1
+; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm5[3,1,2,3]
+; SSE-NEXT:    por %xmm2, %xmm1
+; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm13[3,3,3,3,4,5,6,7]
+; SSE-NEXT:    movdqa {{.*#+}} xmm3 = [255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255]
+; SSE-NEXT:    pand %xmm3, %xmm2
+; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[1,3,2,3,4,5,6,7]
+; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,1,3,1,4,5,6,7]
+; SSE-NEXT:    pandn %xmm0, %xmm3
+; SSE-NEXT:    por %xmm2, %xmm3
 ; SSE-NEXT:    movdqa {{.*#+}} xmm0 = [0,255,255,255,255,0,0,0,255,255,255,255,255,255,255,255]
-; SSE-NEXT:    pand %xmm0, %xmm10
-; SSE-NEXT:    pandn %xmm2, %xmm0
-; SSE-NEXT:    por %xmm10, %xmm0
+; SSE-NEXT:    pand %xmm0, %xmm1
+; SSE-NEXT:    pandn %xmm3, %xmm0
+; SSE-NEXT:    por %xmm1, %xmm0
 ; SSE-NEXT:    movq {{[0-9]+}}(%rsp), %rax
 ; SSE-NEXT:    movq %xmm0, 48(%rax)
 ; SSE-NEXT:    movdqa %xmm9, 16(%rax)