Index: llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
===================================================================
--- llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
+++ llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
@@ -4382,6 +4382,7 @@
   case X86ISD::VPERMILPI:
   case X86ISD::VPERMILPV:
   case X86ISD::VPERM2X128:
+  case X86ISD::SHUF128:
   case X86ISD::VPERMIL2:
   case X86ISD::VPERMI:
   case X86ISD::VPPERM:
@@ -5907,6 +5908,15 @@
                          Mask);
     IsUnary = IsFakeUnary = N->getOperand(0) == N->getOperand(1);
     break;
+  case X86ISD::SHUF128:
+    assert(N->getOperand(0).getValueType() == VT && "Unexpected value type");
+    assert(N->getOperand(1).getValueType() == VT && "Unexpected value type");
+    ImmN = N->getOperand(N->getNumOperands()-1);
+    decodeVSHUF64x2FamilyMask(NumElems, VT.getScalarSizeInBits(),
+                              cast<ConstantSDNode>(ImmN)->getZExtValue(),
+                              Mask);
+    IsUnary = IsFakeUnary = N->getOperand(0) == N->getOperand(1);
+    break;
   case X86ISD::MOVSLDUP:
     assert(N->getOperand(0).getValueType() == VT && "Unexpected value type");
     DecodeMOVSLDUPMask(NumElems, Mask);
@@ -39432,6 +39442,7 @@
   case X86ISD::VPERMILPI:
   case X86ISD::VPERMILPV:
   case X86ISD::VPERM2X128:
+  case X86ISD::SHUF128:
   case X86ISD::VZEXT_MOVL:
   case ISD::VECTOR_SHUFFLE: return combineShuffle(N, DAG, DCI,Subtarget);
   case X86ISD::FMADD_RND:
Index: llvm/trunk/test/CodeGen/X86/vector-shuffle-256-v4.ll
===================================================================
--- llvm/trunk/test/CodeGen/X86/vector-shuffle-256-v4.ll
+++ llvm/trunk/test/CodeGen/X86/vector-shuffle-256-v4.ll
@@ -1,7 +1,7 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=ALL --check-prefix=AVX1
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX2 --check-prefix=AVX2-SLOW
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2,+fast-variable-shuffle | FileCheck %s --check-prefix=ALL --check-prefix=AVX2 --check-prefix=AVX2-FAST
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=ALL --check-prefix=AVX1OR2 --check-prefix=AVX1
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX1OR2 --check-prefix=AVX2 --check-prefix=AVX2-SLOW
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2,+fast-variable-shuffle | FileCheck %s --check-prefix=ALL --check-prefix=AVX1OR2 --check-prefix=AVX2 --check-prefix=AVX2-FAST
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl | FileCheck %s --check-prefix=ALL --check-prefix=AVX512VL --check-prefix=AVX512VL-SLOW
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl,+fast-variable-shuffle | FileCheck %s --check-prefix=ALL --check-prefix=AVX512VL --check-prefix=AVX512VL-FAST
 
@@ -420,21 +420,45 @@
 }
 
 define <4 x double> @shuffle_v4f64_3254(<4 x double> %a, <4 x double> %b) {
-; ALL-LABEL: shuffle_v4f64_3254:
-; ALL:       # %bb.0:
-; ALL-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
-; ALL-NEXT:    vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
-; ALL-NEXT:    retq
+; AVX1OR2-LABEL: shuffle_v4f64_3254:
+; AVX1OR2:       # %bb.0:
+; AVX1OR2-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
+; AVX1OR2-NEXT:    vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
+; AVX1OR2-NEXT:    retq
+;
+; AVX512VL-SLOW-LABEL: shuffle_v4f64_3254:
+; AVX512VL-SLOW:       # %bb.0:
+; AVX512VL-SLOW-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
+; AVX512VL-SLOW-NEXT:    vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
+; AVX512VL-SLOW-NEXT:    retq
+;
+; AVX512VL-FAST-LABEL: shuffle_v4f64_3254:
+; AVX512VL-FAST:       # %bb.0:
+; AVX512VL-FAST-NEXT:    vmovapd {{.*#+}} ymm2 = [3,2,5,4]
+; AVX512VL-FAST-NEXT:    vpermt2pd %ymm1, %ymm2, %ymm0
+; AVX512VL-FAST-NEXT:    retq
   %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
   ret <4 x double> %shuffle
 }
 
 define <4 x double> @shuffle_v4f64_3276(<4 x double> %a, <4 x double> %b) {
-; ALL-LABEL: shuffle_v4f64_3276:
-; ALL:       # %bb.0:
-; ALL-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
-; ALL-NEXT:    vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
-; ALL-NEXT:    retq
+; AVX1OR2-LABEL: shuffle_v4f64_3276:
+; AVX1OR2:       # %bb.0:
+; AVX1OR2-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
+; AVX1OR2-NEXT:    vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
+; AVX1OR2-NEXT:    retq
+;
+; AVX512VL-SLOW-LABEL: shuffle_v4f64_3276:
+; AVX512VL-SLOW:       # %bb.0:
+; AVX512VL-SLOW-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
+; AVX512VL-SLOW-NEXT:    vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
+; AVX512VL-SLOW-NEXT:    retq
+;
+; AVX512VL-FAST-LABEL: shuffle_v4f64_3276:
+; AVX512VL-FAST:       # %bb.0:
+; AVX512VL-FAST-NEXT:    vmovapd {{.*#+}} ymm2 = [3,2,7,6]
+; AVX512VL-FAST-NEXT:    vpermt2pd %ymm1, %ymm2, %ymm0
+; AVX512VL-FAST-NEXT:    retq
   %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
   ret <4 x double> %shuffle
 }
@@ -1095,11 +1119,17 @@
 ; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
 ; AVX2-NEXT:    retq
 ;
-; AVX512VL-LABEL: shuffle_v4i64_3254:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
-; AVX512VL-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
-; AVX512VL-NEXT:    retq
+; AVX512VL-SLOW-LABEL: shuffle_v4i64_3254:
+; AVX512VL-SLOW:       # %bb.0:
+; AVX512VL-SLOW-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
+; AVX512VL-SLOW-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
+; AVX512VL-SLOW-NEXT:    retq
+;
+; AVX512VL-FAST-LABEL: shuffle_v4i64_3254:
+; AVX512VL-FAST:       # %bb.0:
+; AVX512VL-FAST-NEXT:    vmovdqa {{.*#+}} ymm2 = [3,2,5,4]
+; AVX512VL-FAST-NEXT:    vpermt2q %ymm1, %ymm2, %ymm0
+; AVX512VL-FAST-NEXT:    retq
   %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
   ret <4 x i64> %shuffle
 }
@@ -1117,11 +1147,17 @@
 ; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
 ; AVX2-NEXT:    retq
 ;
-; AVX512VL-LABEL: shuffle_v4i64_3276:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
-; AVX512VL-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
-; AVX512VL-NEXT:    retq
+; AVX512VL-SLOW-LABEL: shuffle_v4i64_3276:
+; AVX512VL-SLOW:       # %bb.0:
+; AVX512VL-SLOW-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
+; AVX512VL-SLOW-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
+; AVX512VL-SLOW-NEXT:    retq
+;
+; AVX512VL-FAST-LABEL: shuffle_v4i64_3276:
+; AVX512VL-FAST:       # %bb.0:
+; AVX512VL-FAST-NEXT:    vmovdqa {{.*#+}} ymm2 = [3,2,7,6]
+; AVX512VL-FAST-NEXT:    vpermt2q %ymm1, %ymm2, %ymm0
+; AVX512VL-FAST-NEXT:    retq
   %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
   ret <4 x i64> %shuffle
 }
@@ -1336,19 +1372,12 @@
 }
 
 define <4 x double> @insert_reg_and_zero_v4f64(double %a) {
-; AVX1-LABEL: insert_reg_and_zero_v4f64:
-; AVX1:       # %bb.0:
-; AVX1-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
-; AVX1-NEXT:    vxorps %xmm1, %xmm1, %xmm1
-; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
-; AVX1-NEXT:    retq
-;
-; AVX2-LABEL: insert_reg_and_zero_v4f64:
-; AVX2:       # %bb.0:
-; AVX2-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
-; AVX2-NEXT:    vxorps %xmm1, %xmm1, %xmm1
-; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
-; AVX2-NEXT:    retq
+; AVX1OR2-LABEL: insert_reg_and_zero_v4f64:
+; AVX1OR2:       # %bb.0:
+; AVX1OR2-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
+; AVX1OR2-NEXT:    vxorps %xmm1, %xmm1, %xmm1
+; AVX1OR2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
+; AVX1OR2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: insert_reg_and_zero_v4f64:
 ; AVX512VL:       # %bb.0:
@@ -1446,15 +1475,10 @@
 }
 
 define <4 x i64> @splat128_mem_v4i64_from_v2i64(<2 x i64>* %ptr) {
-; AVX1-LABEL: splat128_mem_v4i64_from_v2i64:
-; AVX1:       # %bb.0:
-; AVX1-NEXT:    vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1]
-; AVX1-NEXT:    retq
-;
-; AVX2-LABEL: splat128_mem_v4i64_from_v2i64:
-; AVX2:       # %bb.0:
-; AVX2-NEXT:    vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1]
-; AVX2-NEXT:    retq
+; AVX1OR2-LABEL: splat128_mem_v4i64_from_v2i64:
+; AVX1OR2:       # %bb.0:
+; AVX1OR2-NEXT:    vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1]
+; AVX1OR2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: splat128_mem_v4i64_from_v2i64:
 ; AVX512VL:       # %bb.0:
Index: llvm/trunk/test/CodeGen/X86/vector-shuffle-256-v8.ll
===================================================================
--- llvm/trunk/test/CodeGen/X86/vector-shuffle-256-v8.ll
+++ llvm/trunk/test/CodeGen/X86/vector-shuffle-256-v8.ll
@@ -838,21 +838,46 @@
 }
 
 define <8 x float> @shuffle_v8f32_7654fedc(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: shuffle_v8f32_7654fedc:
-; ALL:       # %bb.0:
-; ALL-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
-; ALL-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; ALL-NEXT:    retq
+; AVX1OR2-LABEL: shuffle_v8f32_7654fedc:
+; AVX1OR2:       # %bb.0:
+; AVX1OR2-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
+; AVX1OR2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX1OR2-NEXT:    retq
+;
+; AVX512VL-SLOW-LABEL: shuffle_v8f32_7654fedc:
+; AVX512VL-SLOW:       # %bb.0:
+; AVX512VL-SLOW-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
+; AVX512VL-SLOW-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX512VL-SLOW-NEXT:    retq
+;
+; AVX512VL-FAST-LABEL: shuffle_v8f32_7654fedc:
+; AVX512VL-FAST:       # %bb.0:
+; AVX512VL-FAST-NEXT:    vmovaps {{.*#+}} ymm2 = [7,6,5,4,15,14,13,12]
+; AVX512VL-FAST-NEXT:    vpermt2ps %ymm1, %ymm2, %ymm0
+; AVX512VL-FAST-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 15, i32 14, i32 13, i32 12>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_fedc7654(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: shuffle_v8f32_fedc7654:
-; ALL:       # %bb.0:
-; ALL-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm1[2,3],ymm0[2,3]
-; ALL-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; ALL-NEXT:    retq
+; AVX1OR2-LABEL: shuffle_v8f32_fedc7654:
+; AVX1OR2:       # %bb.0:
+; AVX1OR2-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm1[2,3],ymm0[2,3]
+; AVX1OR2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX1OR2-NEXT:    retq
+;
+; AVX512VL-SLOW-LABEL: shuffle_v8f32_fedc7654:
+; AVX512VL-SLOW:       # %bb.0:
+; AVX512VL-SLOW-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm1[2,3],ymm0[2,3]
+; AVX512VL-SLOW-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX512VL-SLOW-NEXT:    retq
+;
+; AVX512VL-FAST-LABEL: shuffle_v8f32_fedc7654:
+; AVX512VL-FAST:       # %bb.0:
+; AVX512VL-FAST-NEXT:    vmovaps {{.*#+}} ymm2 = [7,6,5,4,15,14,13,12]
+; AVX512VL-FAST-NEXT:    vpermi2ps %ymm0, %ymm1, %ymm2
+; AVX512VL-FAST-NEXT:    vmovaps %ymm2, %ymm0
+; AVX512VL-FAST-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 15, i32 14, i32 13, i32 12, i32 7, i32 6, i32 5, i32 4>
   ret <8 x float> %shuffle
 }
@@ -1980,11 +2005,17 @@
 ; AVX1OR2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
 ; AVX1OR2-NEXT:    retq
 ;
-; AVX512VL-LABEL: shuffle_v8i32_7654fedc:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
-; AVX512VL-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX512VL-NEXT:    retq
+; AVX512VL-SLOW-LABEL: shuffle_v8i32_7654fedc:
+; AVX512VL-SLOW:       # %bb.0:
+; AVX512VL-SLOW-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
+; AVX512VL-SLOW-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX512VL-SLOW-NEXT:    retq
+;
+; AVX512VL-FAST-LABEL: shuffle_v8i32_7654fedc:
+; AVX512VL-FAST:       # %bb.0:
+; AVX512VL-FAST-NEXT:    vmovdqa {{.*#+}} ymm2 = [7,6,5,4,15,14,13,12]
+; AVX512VL-FAST-NEXT:    vpermt2d %ymm1, %ymm2, %ymm0
+; AVX512VL-FAST-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 15, i32 14, i32 13, i32 12>
   ret <8 x i32> %shuffle
 }
@@ -1996,11 +2027,18 @@
 ; AVX1OR2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
 ; AVX1OR2-NEXT:    retq
 ;
-; AVX512VL-LABEL: shuffle_v8i32_fedc7654:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm1[2,3],ymm0[2,3]
-; AVX512VL-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX512VL-NEXT:    retq
+; AVX512VL-SLOW-LABEL: shuffle_v8i32_fedc7654:
+; AVX512VL-SLOW:       # %bb.0:
+; AVX512VL-SLOW-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm1[2,3],ymm0[2,3]
+; AVX512VL-SLOW-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX512VL-SLOW-NEXT:    retq
+;
+; AVX512VL-FAST-LABEL: shuffle_v8i32_fedc7654:
+; AVX512VL-FAST:       # %bb.0:
+; AVX512VL-FAST-NEXT:    vmovdqa {{.*#+}} ymm2 = [7,6,5,4,15,14,13,12]
+; AVX512VL-FAST-NEXT:    vpermi2d %ymm0, %ymm1, %ymm2
+; AVX512VL-FAST-NEXT:    vmovdqa %ymm2, %ymm0
+; AVX512VL-FAST-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 15, i32 14, i32 13, i32 12, i32 7, i32 6, i32 5, i32 4>
   ret <8 x i32> %shuffle
 }