Index: lib/Target/X86/X86ISelLowering.cpp
===================================================================
--- lib/Target/X86/X86ISelLowering.cpp
+++ lib/Target/X86/X86ISelLowering.cpp
@@ -13357,18 +13357,19 @@
           DL, MVT::v8i32, V1, V2, Mask, Subtarget, DAG))
     return Rotate;
 
+  // If the shuffle is from a single input, directly
+  // generate a cross-lane VPERMD instruction.
+  if (V2.isUndef()) {
+    SDValue VPermMask = getConstVector(Mask, MVT::v8i32, DAG, DL, true);
+    return DAG.getNode(X86ISD::VPERMV, DL, MVT::v8i32, VPermMask, V1);
+  }
+
   // Try to create an in-lane repeating shuffle mask and then shuffle the
   // results into the target lanes.
   if (SDValue V = lowerShuffleAsRepeatedMaskAndLanePermute(
           DL, MVT::v8i32, V1, V2, Mask, Subtarget, DAG))
     return V;
 
-  // If the shuffle patterns aren't repeated but it is a single input, directly
-  // generate a cross-lane VPERMD instruction.
-  if (V2.isUndef()) {
-    SDValue VPermMask = getConstVector(Mask, MVT::v8i32, DAG, DL, true);
-    return DAG.getNode(X86ISD::VPERMV, DL, MVT::v8i32, VPermMask, V1);
-  }
 
   // Assume that a single SHUFPS is faster than an alternative sequence of
   // multiple instructions (even if the CPU has a domain penalty).
Index: test/CodeGen/X86/avx2-conversions.ll
===================================================================
--- test/CodeGen/X86/avx2-conversions.ll
+++ test/CodeGen/X86/avx2-conversions.ll
@@ -5,16 +5,16 @@
 define <4 x i32> @trunc4(<4 x i64> %A) nounwind {
 ; X32-LABEL: trunc4:
 ; X32:       # BB#0:
-; X32-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; X32-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; X32-NEXT:    vmovaps {{.*#+}} ymm1 = <0,2,4,6,u,u,u,u>
+; X32-NEXT:    vpermps %ymm0, %ymm1, %ymm0
 ; X32-NEXT:    # kill: %xmm0<def> %xmm0<kill> %ymm0<kill>
 ; X32-NEXT:    vzeroupper
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: trunc4:
 ; X64:       # BB#0:
-; X64-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; X64-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; X64-NEXT:    vmovaps {{.*#+}} ymm1 = <0,2,4,6,u,u,u,u>
+; X64-NEXT:    vpermps %ymm0, %ymm1, %ymm0
 ; X64-NEXT:    # kill: %xmm0<def> %xmm0<kill> %ymm0<kill>
 ; X64-NEXT:    vzeroupper
 ; X64-NEXT:    retq
Index: test/CodeGen/X86/avx2-vector-shifts.ll
===================================================================
--- test/CodeGen/X86/avx2-vector-shifts.ll
+++ test/CodeGen/X86/avx2-vector-shifts.ll
@@ -374,8 +374,8 @@
 define <4 x i32> @srl_trunc_and_v4i64(<4 x i32> %x, <4 x i64> %y) nounwind {
 ; X32-LABEL: srl_trunc_and_v4i64:
 ; X32:       # BB#0:
-; X32-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; X32-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; X32-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; X32-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; X32-NEXT:    vpbroadcastd {{.*#+}} xmm2 = [8,8,8,8]
 ; X32-NEXT:    vpand %xmm2, %xmm1, %xmm1
 ; X32-NEXT:    vpsrlvd %xmm1, %xmm0, %xmm0
@@ -384,8 +384,8 @@
 ;
 ; X64-LABEL: srl_trunc_and_v4i64:
 ; X64:       # BB#0:
-; X64-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; X64-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; X64-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; X64-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; X64-NEXT:    vpbroadcastd {{.*#+}} xmm2 = [8,8,8,8]
 ; X64-NEXT:    vpand %xmm2, %xmm1, %xmm1
 ; X64-NEXT:    vpsrlvd %xmm1, %xmm0, %xmm0
Index: test/CodeGen/X86/avx512-shuffles/broadcast-vector-int.ll
===================================================================
--- test/CodeGen/X86/avx512-shuffles/broadcast-vector-int.ll
+++ test/CodeGen/X86/avx512-shuffles/broadcast-vector-int.ll
@@ -460,8 +460,8 @@
 ; CHECK-LABEL: test_2xi32_to_8xi32_mem:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    vpmovzxdq {{.*#+}} xmm0 = mem[0],zero,mem[1],zero
-; CHECK-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; CHECK-NEXT:    vpbroadcastq %xmm0, %ymm0
+; CHECK-NEXT:    vmovdqa {{.*#+}} ymm1 = [0,2,0,2,0,2,0,2]
+; CHECK-NEXT:    vpermd %ymm0, %ymm1, %ymm0
 ; CHECK-NEXT:    retq
   %vec = load <2 x i32>, <2 x i32>* %vp
   %res = shufflevector <2 x i32> %vec, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
@@ -471,10 +471,10 @@
 ; CHECK-LABEL: test_masked_2xi32_to_8xi32_mem_mask0:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    vpmovzxdq {{.*#+}} xmm2 = mem[0],zero,mem[1],zero
-; CHECK-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
-; CHECK-NEXT:    vpxor %xmm3, %xmm3, %xmm3
-; CHECK-NEXT:    vpcmpeqd %ymm3, %ymm1, %k1
-; CHECK-NEXT:    vbroadcasti32x2 {{.*#+}} ymm0 {%k1} = xmm2[0,1,0,1,0,1,0,1]
+; CHECK-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,2,0,2,0,2,0,2]
+; CHECK-NEXT:    vpxor %xmm4, %xmm4, %xmm4
+; CHECK-NEXT:    vpcmpeqd %ymm4, %ymm1, %k1
+; CHECK-NEXT:    vpermd %ymm2, %ymm3, %ymm0 {%k1}
 ; CHECK-NEXT:    retq
   %vec = load <2 x i32>, <2 x i32>* %vp
   %shuf = shufflevector <2 x i32> %vec, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
@@ -487,10 +487,10 @@
 ; CHECK-LABEL: test_masked_z_2xi32_to_8xi32_mem_mask0:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    vpmovzxdq {{.*#+}} xmm1 = mem[0],zero,mem[1],zero
-; CHECK-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
-; CHECK-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; CHECK-NEXT:    vpcmpeqd %ymm2, %ymm0, %k1
-; CHECK-NEXT:    vbroadcasti32x2 {{.*#+}} ymm0 {%k1} {z} = xmm1[0,1,0,1,0,1,0,1]
+; CHECK-NEXT:    vmovdqa {{.*#+}} ymm2 = [0,2,0,2,0,2,0,2]
+; CHECK-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; CHECK-NEXT:    vpcmpeqd %ymm3, %ymm0, %k1
+; CHECK-NEXT:    vpermd %ymm1, %ymm2, %ymm0 {%k1} {z}
 ; CHECK-NEXT:    retq
   %vec = load <2 x i32>, <2 x i32>* %vp
   %shuf = shufflevector <2 x i32> %vec, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
@@ -502,10 +502,10 @@
 ; CHECK-LABEL: test_masked_2xi32_to_8xi32_mem_mask1:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    vpmovzxdq {{.*#+}} xmm2 = mem[0],zero,mem[1],zero
-; CHECK-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
-; CHECK-NEXT:    vpxor %xmm3, %xmm3, %xmm3
-; CHECK-NEXT:    vpcmpeqd %ymm3, %ymm1, %k1
-; CHECK-NEXT:    vbroadcasti32x2 {{.*#+}} ymm0 {%k1} = xmm2[0,1,0,1,0,1,0,1]
+; CHECK-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,2,0,2,0,2,0,2]
+; CHECK-NEXT:    vpxor %xmm4, %xmm4, %xmm4
+; CHECK-NEXT:    vpcmpeqd %ymm4, %ymm1, %k1
+; CHECK-NEXT:    vpermd %ymm2, %ymm3, %ymm0 {%k1}
 ; CHECK-NEXT:    retq
   %vec = load <2 x i32>, <2 x i32>* %vp
   %shuf = shufflevector <2 x i32> %vec, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
@@ -518,10 +518,10 @@
 ; CHECK-LABEL: test_masked_z_2xi32_to_8xi32_mem_mask1:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    vpmovzxdq {{.*#+}} xmm1 = mem[0],zero,mem[1],zero
-; CHECK-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
-; CHECK-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; CHECK-NEXT:    vpcmpeqd %ymm2, %ymm0, %k1
-; CHECK-NEXT:    vbroadcasti32x2 {{.*#+}} ymm0 {%k1} {z} = xmm1[0,1,0,1,0,1,0,1]
+; CHECK-NEXT:    vmovdqa {{.*#+}} ymm2 = [0,2,0,2,0,2,0,2]
+; CHECK-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; CHECK-NEXT:    vpcmpeqd %ymm3, %ymm0, %k1
+; CHECK-NEXT:    vpermd %ymm1, %ymm2, %ymm0 {%k1} {z}
 ; CHECK-NEXT:    retq
   %vec = load <2 x i32>, <2 x i32>* %vp
   %shuf = shufflevector <2 x i32> %vec, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
@@ -533,10 +533,10 @@
 ; CHECK-LABEL: test_masked_2xi32_to_8xi32_mem_mask2:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    vpmovzxdq {{.*#+}} xmm2 = mem[0],zero,mem[1],zero
-; CHECK-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
-; CHECK-NEXT:    vpxor %xmm3, %xmm3, %xmm3
-; CHECK-NEXT:    vpcmpeqd %ymm3, %ymm1, %k1
-; CHECK-NEXT:    vbroadcasti32x2 {{.*#+}} ymm0 {%k1} = xmm2[0,1,0,1,0,1,0,1]
+; CHECK-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,2,0,2,0,2,0,2]
+; CHECK-NEXT:    vpxor %xmm4, %xmm4, %xmm4
+; CHECK-NEXT:    vpcmpeqd %ymm4, %ymm1, %k1
+; CHECK-NEXT:    vpermd %ymm2, %ymm3, %ymm0 {%k1}
 ; CHECK-NEXT:    retq
   %vec = load <2 x i32>, <2 x i32>* %vp
   %shuf = shufflevector <2 x i32> %vec, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
@@ -549,10 +549,10 @@
 ; CHECK-LABEL: test_masked_z_2xi32_to_8xi32_mem_mask2:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    vpmovzxdq {{.*#+}} xmm1 = mem[0],zero,mem[1],zero
-; CHECK-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
-; CHECK-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; CHECK-NEXT:    vpcmpeqd %ymm2, %ymm0, %k1
-; CHECK-NEXT:    vbroadcasti32x2 {{.*#+}} ymm0 {%k1} {z} = xmm1[0,1,0,1,0,1,0,1]
+; CHECK-NEXT:    vmovdqa {{.*#+}} ymm2 = [0,2,0,2,0,2,0,2]
+; CHECK-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; CHECK-NEXT:    vpcmpeqd %ymm3, %ymm0, %k1
+; CHECK-NEXT:    vpermd %ymm1, %ymm2, %ymm0 {%k1} {z}
 ; CHECK-NEXT:    retq
   %vec = load <2 x i32>, <2 x i32>* %vp
   %shuf = shufflevector <2 x i32> %vec, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
@@ -564,10 +564,10 @@
 ; CHECK-LABEL: test_masked_2xi32_to_8xi32_mem_mask3:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    vpmovzxdq {{.*#+}} xmm2 = mem[0],zero,mem[1],zero
-; CHECK-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
-; CHECK-NEXT:    vpxor %xmm3, %xmm3, %xmm3
-; CHECK-NEXT:    vpcmpeqd %ymm3, %ymm1, %k1
-; CHECK-NEXT:    vbroadcasti32x2 {{.*#+}} ymm0 {%k1} = xmm2[0,1,0,1,0,1,0,1]
+; CHECK-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,2,0,2,0,2,0,2]
+; CHECK-NEXT:    vpxor %xmm4, %xmm4, %xmm4
+; CHECK-NEXT:    vpcmpeqd %ymm4, %ymm1, %k1
+; CHECK-NEXT:    vpermd %ymm2, %ymm3, %ymm0 {%k1}
 ; CHECK-NEXT:    retq
   %vec = load <2 x i32>, <2 x i32>* %vp
   %shuf = shufflevector <2 x i32> %vec, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
@@ -580,10 +580,10 @@
 ; CHECK-LABEL: test_masked_z_2xi32_to_8xi32_mem_mask3:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    vpmovzxdq {{.*#+}} xmm1 = mem[0],zero,mem[1],zero
-; CHECK-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
-; CHECK-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; CHECK-NEXT:    vpcmpeqd %ymm2, %ymm0, %k1
-; CHECK-NEXT:    vbroadcasti32x2 {{.*#+}} ymm0 {%k1} {z} = xmm1[0,1,0,1,0,1,0,1]
+; CHECK-NEXT:    vmovdqa {{.*#+}} ymm2 = [0,2,0,2,0,2,0,2]
+; CHECK-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; CHECK-NEXT:    vpcmpeqd %ymm3, %ymm0, %k1
+; CHECK-NEXT:    vpermd %ymm1, %ymm2, %ymm0 {%k1} {z}
 ; CHECK-NEXT:    retq
   %vec = load <2 x i32>, <2 x i32>* %vp
   %shuf = shufflevector <2 x i32> %vec, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
Index: test/CodeGen/X86/combine-shl.ll
===================================================================
--- test/CodeGen/X86/combine-shl.ll
+++ test/CodeGen/X86/combine-shl.ll
@@ -115,8 +115,8 @@
 ;
 ; AVX-LABEL: combine_vec_shl_trunc_and:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
 ; AVX-NEXT:    vpsllvd %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    vzeroupper
Index: test/CodeGen/X86/combine-sra.ll
===================================================================
--- test/CodeGen/X86/combine-sra.ll
+++ test/CodeGen/X86/combine-sra.ll
@@ -182,8 +182,8 @@
 ;
 ; AVX-LABEL: combine_vec_ashr_trunc_and:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
 ; AVX-NEXT:    vpsravd %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    vzeroupper
@@ -215,9 +215,8 @@
 ;
 ; AVX-LABEL: combine_vec_ashr_trunc_lshr:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vpsrlq $32, %ymm0, %ymm0
-; AVX-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX-NEXT:    vmovdqa {{.*#+}} ymm1 = <1,3,5,7,u,u,u,u>
+; AVX-NEXT:    vpermd %ymm0, %ymm1, %ymm0
 ; AVX-NEXT:    vpsravd {{.*}}(%rip), %xmm0, %xmm0
 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
@@ -249,8 +248,8 @@
 ;
 ; AVX-LABEL: combine_vec_ashr_trunc_ashr:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[1,3,2,3,5,7,6,7]
-; AVX-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX-NEXT:    vmovdqa {{.*#+}} ymm1 = <1,3,5,7,u,u,u,u>
+; AVX-NEXT:    vpermd %ymm0, %ymm1, %ymm0
 ; AVX-NEXT:    vpsravd {{.*}}(%rip), %xmm0, %xmm0
 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
Index: test/CodeGen/X86/combine-srl.ll
===================================================================
--- test/CodeGen/X86/combine-srl.ll
+++ test/CodeGen/X86/combine-srl.ll
@@ -218,8 +218,8 @@
 ; AVX-LABEL: combine_vec_lshr_trunc_lshr1:
 ; AVX:       # BB#0:
 ; AVX-NEXT:    vpsrlvq {{.*}}(%rip), %ymm0, %ymm0
-; AVX-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX-NEXT:    vmovdqa {{.*#+}} ymm1 = <0,2,4,6,u,u,u,u>
+; AVX-NEXT:    vpermd %ymm0, %ymm1, %ymm0
 ; AVX-NEXT:    vpsrlvd {{.*}}(%rip), %xmm0, %xmm0
 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
@@ -273,8 +273,8 @@
 ; AVX-LABEL: combine_vec_lshr_trunc_lshr_zero1:
 ; AVX:       # BB#0:
 ; AVX-NEXT:    vpsrlvq {{.*}}(%rip), %ymm0, %ymm0
-; AVX-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX-NEXT:    vmovdqa {{.*#+}} ymm1 = <0,2,4,6,u,u,u,u>
+; AVX-NEXT:    vpermd %ymm0, %ymm1, %ymm0
 ; AVX-NEXT:    vpsrlvd {{.*}}(%rip), %xmm0, %xmm0
 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
@@ -448,8 +448,8 @@
 ;
 ; AVX-LABEL: combine_vec_lshr_trunc_and:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
 ; AVX-NEXT:    vpsrlvd %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    vzeroupper
Index: test/CodeGen/X86/oddshuffles.ll
===================================================================
--- test/CodeGen/X86/oddshuffles.ll
+++ test/CodeGen/X86/oddshuffles.ll
@@ -639,16 +639,16 @@
 ;
 ; AVX2-LABEL: v12i32:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpermilps {{.*#+}} xmm2 = xmm1[2,3,2,3]
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm3 = ymm0[3,3,2,3,7,7,6,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm3 = ymm3[0,2,2,3]
-; AVX2-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0],xmm3[1,2],xmm2[3]
-; AVX2-NEXT:    vmovaps {{.*#+}} ymm3 = <0,4,u,1,5,u,2,6>
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm2 = <0,4,u,1,5,u,2,6>
+; AVX2-NEXT:    vpermps %ymm0, %ymm2, %ymm2
+; AVX2-NEXT:    vbroadcastsd %xmm1, %ymm3
+; AVX2-NEXT:    vblendps {{.*#+}} ymm2 = ymm2[0,1],ymm3[2],ymm2[3,4],ymm3[5],ymm2[6,7]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm3 = <u,3,7,u,u,u,u,u>
 ; AVX2-NEXT:    vpermps %ymm0, %ymm3, %ymm0
-; AVX2-NEXT:    vbroadcastsd %xmm1, %ymm1
-; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3,4],ymm1[5],ymm0[6,7]
-; AVX2-NEXT:    vmovaps %ymm0, (%rdi)
-; AVX2-NEXT:    vmovaps %xmm2, 32(%rdi)
+; AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm1[2,3,2,3]
+; AVX2-NEXT:    vblendps {{.*#+}} xmm0 = xmm1[0],xmm0[1,2],xmm1[3]
+; AVX2-NEXT:    vmovaps %xmm0, 32(%rdi)
+; AVX2-NEXT:    vmovaps %ymm2, (%rdi)
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
@@ -1417,12 +1417,12 @@
 ; AVX2-NEXT:    vmovaps {{.*#+}} ymm6 = <1,4,7,2,5,u,u,u>
 ; AVX2-NEXT:    vpermps %ymm5, %ymm6, %ymm5
 ; AVX2-NEXT:    vblendps {{.*#+}} ymm4 = ymm5[0,1,2,3,4],ymm4[5,6,7]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm5 = <u,u,u,u,u,1,4,7>
+; AVX2-NEXT:    vpermps %ymm2, %ymm5, %ymm2
 ; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2],ymm1[3,4],ymm0[5],ymm1[6,7]
 ; AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = <2,5,0,3,6,u,u,u>
 ; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm1 = ymm2[0,1,0,3,4,5,4,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm1 = ymm1[0,1,0,3]
-; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7]
+; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm2[5,6,7]
 ; AVX2-NEXT:    vmovups %ymm3, (%rsi)
 ; AVX2-NEXT:    vmovups %ymm4, (%rdx)
 ; AVX2-NEXT:    vmovups %ymm0, (%rcx)
@@ -1608,25 +1608,25 @@
 ; AVX2-NEXT:    vmovups (%rsi), %ymm0
 ; AVX2-NEXT:    vmovups (%rdx), %ymm1
 ; AVX2-NEXT:    vmovups (%rcx), %ymm2
-; AVX2-NEXT:    vpermilps {{.*#+}} xmm3 = xmm1[1,0,2,2]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm3 = ymm3[0,1,0,1]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm3 = <u,0,u,u,1,u,u,2>
+; AVX2-NEXT:    vpermps %ymm1, %ymm3, %ymm3
 ; AVX2-NEXT:    vpermpd {{.*#+}} ymm4 = ymm0[0,0,2,1]
 ; AVX2-NEXT:    vblendps {{.*#+}} ymm3 = ymm4[0],ymm3[1],ymm4[2,3],ymm3[4],ymm4[5,6],ymm3[7]
 ; AVX2-NEXT:    vbroadcastsd %xmm2, %ymm4
 ; AVX2-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1],ymm4[2],ymm3[3,4],ymm4[5],ymm3[6,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm4 = ymm2[2,1,3,3]
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm5 = ymm1[1,2,3,3,5,6,7,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm5 = ymm5[2,2,2,3]
-; AVX2-NEXT:    vblendps {{.*#+}} ymm4 = ymm5[0],ymm4[1],ymm5[2,3],ymm4[4],ymm5[5,6],ymm4[7]
-; AVX2-NEXT:    vbroadcastsd 24(%rsi), %ymm5
-; AVX2-NEXT:    vblendps {{.*#+}} ymm4 = ymm4[0,1],ymm5[2],ymm4[3,4],ymm5[5],ymm4[6,7]
 ; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[1,1,2,2]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm2 = ymm2[1,1,2,2]
-; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm2[0],ymm0[1],ymm2[2,3],ymm0[4],ymm2[5,6],ymm0[7]
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm1 = ymm1[0,0,3,3,4,4,7,7]
-; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3,4],ymm1[5],ymm0[6,7]
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm4 = ymm2[1,1,2,2]
+; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm4[0],ymm0[1],ymm4[2,3],ymm0[4],ymm4[5,6],ymm0[7]
+; AVX2-NEXT:    vpermilps {{.*#+}} ymm4 = ymm1[0,0,3,3,4,4,7,7]
+; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm4[2],ymm0[3,4],ymm4[5],ymm0[6,7]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm4 = <5,u,u,6,u,u,7,u>
+; AVX2-NEXT:    vpermps %ymm1, %ymm4, %ymm1
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm2 = ymm2[2,1,3,3]
+; AVX2-NEXT:    vblendps {{.*#+}} ymm1 = ymm1[0],ymm2[1],ymm1[2,3],ymm2[4],ymm1[5,6],ymm2[7]
+; AVX2-NEXT:    vbroadcastsd 24(%rsi), %ymm2
+; AVX2-NEXT:    vblendps {{.*#+}} ymm1 = ymm1[0,1],ymm2[2],ymm1[3,4],ymm2[5],ymm1[6,7]
+; AVX2-NEXT:    vmovups %ymm1, 64(%rdi)
 ; AVX2-NEXT:    vmovups %ymm0, 32(%rdi)
-; AVX2-NEXT:    vmovups %ymm4, 64(%rdi)
 ; AVX2-NEXT:    vmovups %ymm3, (%rdi)
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
Index: test/CodeGen/X86/psubus.ll
===================================================================
--- test/CodeGen/X86/psubus.ll
+++ test/CodeGen/X86/psubus.ll
@@ -1918,10 +1918,9 @@
 ; AVX2-NEXT:    vblendvpd %ymm5, %ymm2, %ymm0, %ymm0
 ; AVX2-NEXT:    vpsubq %ymm2, %ymm0, %ymm0
 ; AVX2-NEXT:    vpsubq %ymm1, %ymm3, %ymm1
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
 ; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
Index: test/CodeGen/X86/reduce-trunc-shl.ll
===================================================================
--- test/CodeGen/X86/reduce-trunc-shl.ll
+++ test/CodeGen/X86/reduce-trunc-shl.ll
@@ -13,8 +13,8 @@
 ;
 ; AVX2-LABEL: trunc_shl_7_v4i32_v4i64:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = mem[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm0 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd (%rsi), %ymm0, %ymm0
 ; AVX2-NEXT:    vpslld $7, %xmm0, %xmm0
 ; AVX2-NEXT:    vmovdqa %xmm0, (%rdi)
 ; AVX2-NEXT:    vzeroupper
Index: test/CodeGen/X86/shuffle-vs-trunc-256.ll
===================================================================
--- test/CodeGen/X86/shuffle-vs-trunc-256.ll
+++ test/CodeGen/X86/shuffle-vs-trunc-256.ll
@@ -292,8 +292,8 @@
 ;
 ; AVX2-LABEL: trunc_v4i64_to_v4i32:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = mem[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm0 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermps (%rdi), %ymm0, %ymm0
 ; AVX2-NEXT:    vmovaps %xmm0, (%rsi)
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -565,8 +565,8 @@
 ;
 ; AVX2-LABEL: trunc_v4i64_to_v4i16:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = mem[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm0 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd (%rdi), %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
 ; AVX2-NEXT:    vmovq %xmm0, (%rsi)
 ; AVX2-NEXT:    vzeroupper
@@ -695,8 +695,8 @@
 ;
 ; AVX2-LABEL: trunc_v4i64_to_v4i8:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = mem[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm0 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd (%rdi), %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
 ; AVX2-NEXT:    vmovd %xmm0, (%rsi)
 ; AVX2-NEXT:    vzeroupper
Index: test/CodeGen/X86/vector-shuffle-256-v8.ll
===================================================================
--- test/CodeGen/X86/vector-shuffle-256-v8.ll
+++ test/CodeGen/X86/vector-shuffle-256-v8.ll
@@ -944,8 +944,8 @@
 ;
 ; AVX2OR512VL-LABEL: shuffle_v8i32_00000010:
 ; AVX2OR512VL:       # BB#0:
-; AVX2OR512VL-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,0,1,0]
-; AVX2OR512VL-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,0,0,1]
+; AVX2OR512VL-NEXT:    vmovaps {{.*#+}} ymm1 = [0,0,0,0,0,0,1,0]
+; AVX2OR512VL-NEXT:    vpermps %ymm0, %ymm1, %ymm0
 ; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 0>
   ret <8 x i32> %shuffle
@@ -961,8 +961,8 @@
 ;
 ; AVX2OR512VL-LABEL: shuffle_v8i32_00000200:
 ; AVX2OR512VL:       # BB#0:
-; AVX2OR512VL-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,0,0,2]
-; AVX2OR512VL-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,0,1,0]
+; AVX2OR512VL-NEXT:    vmovaps {{.*#+}} ymm1 = [0,0,0,0,0,2,0,0]
+; AVX2OR512VL-NEXT:    vpermps %ymm0, %ymm1, %ymm0
 ; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 2, i32 0, i32 0>
   ret <8 x i32> %shuffle
@@ -978,8 +978,8 @@
 ;
 ; AVX2OR512VL-LABEL: shuffle_v8i32_00003000:
 ; AVX2OR512VL:       # BB#0:
-; AVX2OR512VL-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,0,3,0]
-; AVX2OR512VL-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,0,1,0]
+; AVX2OR512VL-NEXT:    vmovaps {{.*#+}} ymm1 = [0,0,0,0,3,0,0,0]
+; AVX2OR512VL-NEXT:    vpermps %ymm0, %ymm1, %ymm0
 ; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 3, i32 0, i32 0, i32 0>
   ret <8 x i32> %shuffle
@@ -1096,8 +1096,8 @@
 ;
 ; AVX2OR512VL-LABEL: shuffle_v8i32_00001111:
 ; AVX2OR512VL:       # BB#0:
-; AVX2OR512VL-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,0,1,1]
-; AVX2OR512VL-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,0,1,1]
+; AVX2OR512VL-NEXT:    vmovaps {{.*#+}} ymm1 = [0,0,0,0,1,1,1,1]
+; AVX2OR512VL-NEXT:    vpermps %ymm0, %ymm1, %ymm0
 ; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1>
   ret <8 x i32> %shuffle
@@ -1221,18 +1221,18 @@
 ;
 ; AVX2-LABEL: shuffle_v8i32_08991abb:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <u,0,1,1,u,2,3,3>
-; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
-; AVX2-NEXT:    vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
-; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm2 = <0,u,u,u,1,u,u,u>
+; AVX2-NEXT:    vpermps %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm2 = <u,0,1,1,u,2,3,3>
+; AVX2-NEXT:    vpermps %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v8i32_08991abb:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpmovzxdq {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero
-; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm0 = [8,0,1,1,10,2,3,3]
-; AVX512VL-NEXT:    vpermi2d %ymm2, %ymm1, %ymm0
+; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm2 = [8,0,1,1,9,2,3,3]
+; AVX512VL-NEXT:    vpermi2d %ymm0, %ymm1, %ymm2
+; AVX512VL-NEXT:    vmovdqa %ymm2, %ymm0
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 8, i32 9, i32 9, i32 1, i32 10, i32 11, i32 11>
   ret <8 x i32> %shuffle
@@ -1264,12 +1264,19 @@
 ; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
 ; AVX1-NEXT:    retq
 ;
-; AVX2OR512VL-LABEL: shuffle_v8i32_09ab1def:
-; AVX2OR512VL:       # BB#0:
-; AVX2OR512VL-NEXT:    vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
-; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
-; AVX2OR512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
-; AVX2OR512VL-NEXT:    retq
+; AVX2-LABEL: shuffle_v8i32_09ab1def:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm2 = <0,u,u,u,1,u,u,u>
+; AVX2-NEXT:    vpermps %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-NEXT:    retq
+;
+; AVX512VL-LABEL: shuffle_v8i32_09ab1def:
+; AVX512VL:       # BB#0:
+; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm2 = [8,1,2,3,9,5,6,7]
+; AVX512VL-NEXT:    vpermi2d %ymm0, %ymm1, %ymm2
+; AVX512VL-NEXT:    vmovdqa %ymm2, %ymm0
+; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 9, i32 10, i32 11, i32 1, i32 13, i32 14, i32 15>
   ret <8 x i32> %shuffle
 }
@@ -1698,9 +1705,9 @@
 ;
 ; AVX2-LABEL: shuffle_v8i32_6caa87e5:
 ; AVX2:       # BB#0:
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm2 = <u,4,2,2,0,u,6,u>
+; AVX2-NEXT:    vpermps %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[3,1,3,2]
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm1 = ymm1[0,0,2,2,4,4,6,6]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm1 = ymm1[2,1,0,3]
 ; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3,4],ymm0[5],ymm1[6],ymm0[7]
 ; AVX2-NEXT:    retq
 ;
@@ -1723,8 +1730,8 @@
 ;
 ; AVX2OR512VL-LABEL: shuffle_v8i32_32103210:
 ; AVX2OR512VL:       # BB#0:
-; AVX2OR512VL-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[3,2,1,0]
-; AVX2OR512VL-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2OR512VL-NEXT:    vmovaps {{.*#+}} ymm1 = [3,2,1,0,3,2,1,0]
+; AVX2OR512VL-NEXT:    vpermps %ymm0, %ymm1, %ymm0
 ; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 3, i32 2, i32 1, i32 0>
   ret <8 x i32> %shuffle
@@ -1739,8 +1746,8 @@
 ;
 ; AVX2OR512VL-LABEL: shuffle_v8i32_76547654:
 ; AVX2OR512VL:       # BB#0:
-; AVX2OR512VL-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX2OR512VL-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX2OR512VL-NEXT:    vmovaps {{.*#+}} ymm1 = [7,6,5,4,7,6,5,4]
+; AVX2OR512VL-NEXT:    vpermps %ymm0, %ymm1, %ymm0
 ; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 7, i32 6, i32 5, i32 4>
   ret <8 x i32> %shuffle
@@ -1755,8 +1762,8 @@
 ;
 ; AVX2OR512VL-LABEL: shuffle_v8i32_76543210:
 ; AVX2OR512VL:       # BB#0:
-; AVX2OR512VL-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX2OR512VL-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX2OR512VL-NEXT:    vmovaps {{.*#+}} ymm1 = [7,6,5,4,3,2,1,0]
+; AVX2OR512VL-NEXT:    vpermps %ymm0, %ymm1, %ymm0
 ; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
   ret <8 x i32> %shuffle
Index: test/CodeGen/X86/vector-shuffle-combining.ll
===================================================================
--- test/CodeGen/X86/vector-shuffle-combining.ll
+++ test/CodeGen/X86/vector-shuffle-combining.ll
@@ -2539,8 +2539,8 @@
 ; AVX2-LABEL: combine_unneeded_subvector1:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpaddd {{.*}}(%rip), %ymm0, %ymm0
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = [7,6,5,4,7,6,5,4]
+; AVX2-NEXT:    vpermd %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
   %b = add <8 x i32> %a, <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
   %c = shufflevector <8 x i32> %b, <8 x i32> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 7, i32 6, i32 5, i32 4>
Index: test/CodeGen/X86/vector-trunc-math.ll
===================================================================
--- test/CodeGen/X86/vector-trunc-math.ll
+++ test/CodeGen/X86/vector-trunc-math.ll
@@ -31,8 +31,8 @@
 ; AVX2-LABEL: trunc_add_v4i64_v4i32:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpaddq %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    # kill: %xmm0<def> %xmm0<kill> %ymm0<kill>
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -94,10 +94,9 @@
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpaddq %ymm3, %ymm1, %ymm1
 ; AVX2-NEXT:    vpaddq %ymm2, %ymm0, %ymm0
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
@@ -234,24 +233,21 @@
 ; AVX2-NEXT:    vpaddq %ymm4, %ymm0, %ymm0
 ; AVX2-NEXT:    vpaddq %ymm7, %ymm3, %ymm3
 ; AVX2-NEXT:    vpaddq %ymm6, %ymm2, %ymm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm2 = ymm2[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm2, %ymm4, %ymm2
+; AVX2-NEXT:    vpermd %ymm3, %ymm4, %ymm3
 ; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm4 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX2-NEXT:    vpshufb %xmm4, %xmm2, %xmm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm5 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX2-NEXT:    vpshufb %xmm5, %xmm2, %xmm2
+; AVX2-NEXT:    vpermd %ymm0, %ymm4, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm4, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufb %xmm4, %xmm0, %xmm0
+; AVX2-NEXT:    vpshufb %xmm5, %xmm0, %xmm0
 ; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -469,8 +465,8 @@
 ;
 ; AVX2-LABEL: trunc_add_const_v4i64_v4i32:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpaddd {{.*}}(%rip), %xmm0, %xmm0
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -522,10 +518,9 @@
 ;
 ; AVX2-LABEL: trunc_add_const_v8i64_v8i16:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
@@ -636,24 +631,21 @@
 ;
 ; AVX2-LABEL: trunc_add_const_v16i64_v16i8:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm2 = ymm2[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm2, %ymm4, %ymm2
+; AVX2-NEXT:    vpermd %ymm3, %ymm4, %ymm3
 ; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm4 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX2-NEXT:    vpshufb %xmm4, %xmm2, %xmm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm5 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX2-NEXT:    vpshufb %xmm5, %xmm2, %xmm2
+; AVX2-NEXT:    vpermd %ymm0, %ymm4, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm4, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufb %xmm4, %xmm0, %xmm0
+; AVX2-NEXT:    vpshufb %xmm5, %xmm0, %xmm0
 ; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
 ; AVX2-NEXT:    vpaddb {{.*}}(%rip), %xmm0, %xmm0
 ; AVX2-NEXT:    vzeroupper
@@ -814,8 +806,8 @@
 ; AVX2-LABEL: trunc_sub_v4i64_v4i32:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpsubq %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    # kill: %xmm0<def> %xmm0<kill> %ymm0<kill>
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -877,10 +869,9 @@
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpsubq %ymm3, %ymm1, %ymm1
 ; AVX2-NEXT:    vpsubq %ymm2, %ymm0, %ymm0
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
@@ -1017,24 +1008,21 @@
 ; AVX2-NEXT:    vpsubq %ymm4, %ymm0, %ymm0
 ; AVX2-NEXT:    vpsubq %ymm7, %ymm3, %ymm3
 ; AVX2-NEXT:    vpsubq %ymm6, %ymm2, %ymm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm2 = ymm2[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm2, %ymm4, %ymm2
+; AVX2-NEXT:    vpermd %ymm3, %ymm4, %ymm3
 ; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm4 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX2-NEXT:    vpshufb %xmm4, %xmm2, %xmm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm5 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX2-NEXT:    vpshufb %xmm5, %xmm2, %xmm2
+; AVX2-NEXT:    vpermd %ymm0, %ymm4, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm4, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufb %xmm4, %xmm0, %xmm0
+; AVX2-NEXT:    vpshufb %xmm5, %xmm0, %xmm0
 ; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -1212,8 +1200,8 @@
 ; AVX2-LABEL: trunc_sub_const_v4i64_v4i32:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpsubq {{.*}}(%rip), %ymm0, %ymm0
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    # kill: %xmm0<def> %xmm0<kill> %ymm0<kill>
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -1280,10 +1268,9 @@
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpsubq {{.*}}(%rip), %ymm1, %ymm1
 ; AVX2-NEXT:    vpsubq {{.*}}(%rip), %ymm0, %ymm0
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
@@ -1421,24 +1408,21 @@
 ; AVX2-NEXT:    vpsubq {{.*}}(%rip), %ymm0, %ymm0
 ; AVX2-NEXT:    vpsubq {{.*}}(%rip), %ymm3, %ymm3
 ; AVX2-NEXT:    vpsubq {{.*}}(%rip), %ymm2, %ymm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm2 = ymm2[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm2, %ymm4, %ymm2
+; AVX2-NEXT:    vpermd %ymm3, %ymm4, %ymm3
 ; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm4 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX2-NEXT:    vpshufb %xmm4, %xmm2, %xmm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm5 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX2-NEXT:    vpshufb %xmm5, %xmm2, %xmm2
+; AVX2-NEXT:    vpermd %ymm0, %ymm4, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm4, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufb %xmm4, %xmm0, %xmm0
+; AVX2-NEXT:    vpshufb %xmm5, %xmm0, %xmm0
 ; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -1625,10 +1609,9 @@
 ;
 ; AVX2-LABEL: trunc_mul_v4i64_v4i32:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
 ; AVX2-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -1722,18 +1705,15 @@
 ;
 ; AVX2-LABEL: trunc_mul_v8i64_v8i16:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm2 = ymm2[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm2, %ymm4, %ymm2
+; AVX2-NEXT:    vpermd %ymm3, %ymm4, %ymm3
 ; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vpermd %ymm0, %ymm4, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm4, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
@@ -2030,15 +2010,12 @@
 ;
 ; AVX2-LABEL: trunc_mul_v16i64_v16i8:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm7 = ymm7[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm7 = ymm7[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm8 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm7, %ymm8, %ymm7
+; AVX2-NEXT:    vpermd %ymm3, %ymm8, %ymm3
 ; AVX2-NEXT:    vpmulld %xmm7, %xmm3, %xmm3
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm6 = ymm6[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm6 = ymm6[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm2 = ymm2[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
+; AVX2-NEXT:    vpermd %ymm6, %ymm8, %ymm6
+; AVX2-NEXT:    vpermd %ymm2, %ymm8, %ymm2
 ; AVX2-NEXT:    vpmulld %xmm6, %xmm2, %xmm2
 ; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
@@ -2046,15 +2023,11 @@
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} xmm6 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
 ; AVX2-NEXT:    vpshufb %xmm6, %xmm2, %xmm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm5 = ymm5[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm5 = ymm5[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vpermd %ymm5, %ymm8, %ymm5
+; AVX2-NEXT:    vpermd %ymm1, %ymm8, %ymm1
 ; AVX2-NEXT:    vpmulld %xmm5, %xmm1, %xmm1
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm4 = ymm4[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm4 = ymm4[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vpermd %ymm4, %ymm8, %ymm4
+; AVX2-NEXT:    vpermd %ymm0, %ymm8, %ymm0
 ; AVX2-NEXT:    vpmulld %xmm4, %xmm0, %xmm0
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
@@ -2342,8 +2315,8 @@
 ;
 ; AVX2-LABEL: trunc_mul_const_v4i64_v4i32:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpmulld {{.*}}(%rip), %xmm0, %xmm0
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -2395,10 +2368,9 @@
 ;
 ; AVX2-LABEL: trunc_mul_const_v8i64_v8i16:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
@@ -2615,28 +2587,25 @@
 ;
 ; AVX2-LABEL: trunc_mul_const_v16i64_v16i8:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm2 = ymm2[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm2, %ymm4, %ymm2
 ; AVX2-NEXT:    vpmulld {{.*}}(%rip), %xmm2, %xmm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,2,2,3]
+; AVX2-NEXT:    vpermd %ymm3, %ymm4, %ymm3
 ; AVX2-NEXT:    vpmulld {{.*}}(%rip), %xmm3, %xmm3
 ; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm4 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX2-NEXT:    vpshufb %xmm4, %xmm2, %xmm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm5 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX2-NEXT:    vpshufb %xmm5, %xmm2, %xmm2
+; AVX2-NEXT:    vpermd %ymm0, %ymm4, %ymm0
 ; AVX2-NEXT:    vpmulld {{.*}}(%rip), %xmm0, %xmm0
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vpermd %ymm1, %ymm4, %ymm1
 ; AVX2-NEXT:    vpmulld {{.*}}(%rip), %xmm1, %xmm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufb %xmm4, %xmm0, %xmm0
+; AVX2-NEXT:    vpshufb %xmm5, %xmm0, %xmm0
 ; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -2832,8 +2801,8 @@
 ; AVX2-LABEL: trunc_and_v4i64_v4i32:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vandps %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    # kill: %xmm0<def> %xmm0<kill> %ymm0<kill>
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -2891,10 +2860,9 @@
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpand %ymm3, %ymm1, %ymm1
 ; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
@@ -3021,24 +2989,21 @@
 ; AVX2-NEXT:    vpand %ymm4, %ymm0, %ymm0
 ; AVX2-NEXT:    vpand %ymm7, %ymm3, %ymm3
 ; AVX2-NEXT:    vpand %ymm6, %ymm2, %ymm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm2 = ymm2[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm2, %ymm4, %ymm2
+; AVX2-NEXT:    vpermd %ymm3, %ymm4, %ymm3
 ; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm4 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX2-NEXT:    vpshufb %xmm4, %xmm2, %xmm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm5 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX2-NEXT:    vpshufb %xmm5, %xmm2, %xmm2
+; AVX2-NEXT:    vpermd %ymm0, %ymm4, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm4, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufb %xmm4, %xmm0, %xmm0
+; AVX2-NEXT:    vpshufb %xmm5, %xmm0, %xmm0
 ; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -3201,8 +3166,8 @@
 ;
 ; AVX2-LABEL: trunc_and_const_v4i64_v4i32:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vandps {{.*}}(%rip), %xmm0, %xmm0
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -3254,10 +3219,9 @@
 ;
 ; AVX2-LABEL: trunc_and_const_v8i64_v8i16:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
@@ -3368,24 +3332,21 @@
 ;
 ; AVX2-LABEL: trunc_and_const_v16i64_v16i8:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm2 = ymm2[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm2, %ymm4, %ymm2
+; AVX2-NEXT:    vpermd %ymm3, %ymm4, %ymm3
 ; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm4 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX2-NEXT:    vpshufb %xmm4, %xmm2, %xmm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm5 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX2-NEXT:    vpshufb %xmm5, %xmm2, %xmm2
+; AVX2-NEXT:    vpermd %ymm0, %ymm4, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm4, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufb %xmm4, %xmm0, %xmm0
+; AVX2-NEXT:    vpshufb %xmm5, %xmm0, %xmm0
 ; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
 ; AVX2-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
 ; AVX2-NEXT:    vzeroupper
@@ -3544,8 +3505,8 @@
 ; AVX2-LABEL: trunc_xor_v4i64_v4i32:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vxorps %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    # kill: %xmm0<def> %xmm0<kill> %ymm0<kill>
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -3603,10 +3564,9 @@
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpxor %ymm3, %ymm1, %ymm1
 ; AVX2-NEXT:    vpxor %ymm2, %ymm0, %ymm0
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
@@ -3733,24 +3693,21 @@
 ; AVX2-NEXT:    vpxor %ymm4, %ymm0, %ymm0
 ; AVX2-NEXT:    vpxor %ymm7, %ymm3, %ymm3
 ; AVX2-NEXT:    vpxor %ymm6, %ymm2, %ymm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm2 = ymm2[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm2, %ymm4, %ymm2
+; AVX2-NEXT:    vpermd %ymm3, %ymm4, %ymm3
 ; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm4 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX2-NEXT:    vpshufb %xmm4, %xmm2, %xmm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm5 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX2-NEXT:    vpshufb %xmm5, %xmm2, %xmm2
+; AVX2-NEXT:    vpermd %ymm0, %ymm4, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm4, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufb %xmm4, %xmm0, %xmm0
+; AVX2-NEXT:    vpshufb %xmm5, %xmm0, %xmm0
 ; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -3913,8 +3870,8 @@
 ;
 ; AVX2-LABEL: trunc_xor_const_v4i64_v4i32:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vxorps {{.*}}(%rip), %xmm0, %xmm0
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -3966,10 +3923,9 @@
 ;
 ; AVX2-LABEL: trunc_xor_const_v8i64_v8i16:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
@@ -4080,24 +4036,21 @@
 ;
 ; AVX2-LABEL: trunc_xor_const_v16i64_v16i8:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm2 = ymm2[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm2, %ymm4, %ymm2
+; AVX2-NEXT:    vpermd %ymm3, %ymm4, %ymm3
 ; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm4 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX2-NEXT:    vpshufb %xmm4, %xmm2, %xmm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm5 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX2-NEXT:    vpshufb %xmm5, %xmm2, %xmm2
+; AVX2-NEXT:    vpermd %ymm0, %ymm4, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm4, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufb %xmm4, %xmm0, %xmm0
+; AVX2-NEXT:    vpshufb %xmm5, %xmm0, %xmm0
 ; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
 ; AVX2-NEXT:    vpxor {{.*}}(%rip), %xmm0, %xmm0
 ; AVX2-NEXT:    vzeroupper
@@ -4256,8 +4209,8 @@
 ; AVX2-LABEL: trunc_or_v4i64_v4i32:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vorps %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    # kill: %xmm0<def> %xmm0<kill> %ymm0<kill>
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -4315,10 +4268,9 @@
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpor %ymm3, %ymm1, %ymm1
 ; AVX2-NEXT:    vpor %ymm2, %ymm0, %ymm0
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
@@ -4445,24 +4397,21 @@
 ; AVX2-NEXT:    vpor %ymm4, %ymm0, %ymm0
 ; AVX2-NEXT:    vpor %ymm7, %ymm3, %ymm3
 ; AVX2-NEXT:    vpor %ymm6, %ymm2, %ymm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm2 = ymm2[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm2, %ymm4, %ymm2
+; AVX2-NEXT:    vpermd %ymm3, %ymm4, %ymm3
 ; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm4 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX2-NEXT:    vpshufb %xmm4, %xmm2, %xmm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm5 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX2-NEXT:    vpshufb %xmm5, %xmm2, %xmm2
+; AVX2-NEXT:    vpermd %ymm0, %ymm4, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm4, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufb %xmm4, %xmm0, %xmm0
+; AVX2-NEXT:    vpshufb %xmm5, %xmm0, %xmm0
 ; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -4625,8 +4574,8 @@
 ;
 ; AVX2-LABEL: trunc_or_const_v4i64_v4i32:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vorps {{.*}}(%rip), %xmm0, %xmm0
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -4678,10 +4627,9 @@
 ;
 ; AVX2-LABEL: trunc_or_const_v8i64_v8i16:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
@@ -4792,24 +4740,21 @@
 ;
 ; AVX2-LABEL: trunc_or_const_v16i64_v16i8:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm2 = ymm2[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm2, %ymm4, %ymm2
+; AVX2-NEXT:    vpermd %ymm3, %ymm4, %ymm3
 ; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm4 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX2-NEXT:    vpshufb %xmm4, %xmm2, %xmm2
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm5 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX2-NEXT:    vpshufb %xmm5, %xmm2, %xmm2
+; AVX2-NEXT:    vpermd %ymm0, %ymm4, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm4, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufb %xmm4, %xmm0, %xmm0
+; AVX2-NEXT:    vpshufb %xmm5, %xmm0, %xmm0
 ; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
 ; AVX2-NEXT:    vpor {{.*}}(%rip), %xmm0, %xmm0
 ; AVX2-NEXT:    vzeroupper
Index: test/CodeGen/X86/vector-trunc.ll
===================================================================
--- test/CodeGen/X86/vector-trunc.ll
+++ test/CodeGen/X86/vector-trunc.ll
@@ -28,10 +28,9 @@
 ;
 ; AVX2-LABEL: trunc8i64_8i32:
 ; AVX2:       # BB#0: # %entry
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermps %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermps %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    retq
 ;
@@ -105,10 +104,9 @@
 ;
 ; AVX2-LABEL: trunc8i64_8i32_ashr:
 ; AVX2:       # BB#0: # %entry
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[1,3,2,3,5,7,6,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm1 = ymm1[1,3,2,3,5,7,6,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm2 = <1,3,5,7,u,u,u,u>
+; AVX2-NEXT:    vpermps %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermps %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    retq
 ;
@@ -150,13 +148,10 @@
 ;
 ; AVX2-LABEL: trunc8i64_8i32_lshr:
 ; AVX2:       # BB#0: # %entry
-; AVX2-NEXT:    vpsrlq $32, %ymm1, %ymm1
-; AVX2-NEXT:    vpsrlq $32, %ymm0, %ymm0
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
-; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm2 = <1,3,5,7,u,u,u,u>
+; AVX2-NEXT:    vpermps %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermps %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: trunc8i64_8i32_lshr:
@@ -230,10 +225,9 @@
 ;
 ; AVX2-LABEL: trunc8i64_8i16:
 ; AVX2:       # BB#0: # %entry
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
@@ -285,10 +279,9 @@
 ;
 ; AVX2-LABEL: trunc8i64_8i8:
 ; AVX2:       # BB#0: # %entry
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
 ; AVX2-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
@@ -1370,10 +1363,9 @@
 ;
 ; AVX2-LABEL: trunc2x4i64_8i32:
 ; AVX2:       # BB#0: # %entry
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermps %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermps %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    retq
 ;
@@ -1476,10 +1468,9 @@
 ;
 ; AVX2-LABEL: trunc2x4i64_8i16:
 ; AVX2:       # BB#0: # %entry
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,2,4,6,u,u,u,u>
+; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
 ; AVX2-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufb %xmm2, %xmm0, %xmm0