diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index fadc818..74d9bae 100644
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -2389,6 +2389,56 @@
   return ::isPSHUFLWMask(M, N->getValueType(0));
 }
 
+/// isPALIGNRMask - Return true if the node specifies a shuffle of elements that
+/// is suitable for input to PALIGNR.
+static bool isPALIGNRMask(const SmallVectorImpl<int> &Mask, EVT VT,
+                          bool hasSSSE3) {
+  int i, e = VT.getVectorNumElements();
+  
+  // Do not handle v2i64 / v2f64 shuffles with palignr.
+  if (e < 4 || !hasSSSE3)
+    return false;
+  
+  for (i = 0; i != e; ++i)
+    if (Mask[i] >= 0)
+      break;
+  
+  // All undef, not a palignr.
+  if (i == e)
+    return false;
+
+  // Determine if it's ok to perform a palignr with only the LHS, since we
+  // don't have access to the actual shuffle elements to see if RHS is undef.
+  bool Unary = Mask[i] < (int)e;
+  bool NeedsUnary = false;
+
+  int s = Mask[i] - i;
+  
+  // Check the rest of the elements to see if they are consecutive.
+  for (++i; i != e; ++i) {
+    int m = Mask[i];
+    if (m < 0) 
+      continue;
+    
+    Unary = Unary && (m < (int)e);
+    NeedsUnary = NeedsUnary || (m < s);
+
+    if (NeedsUnary && !Unary)
+      return false;
+    if (Unary && m != ((s+i) & (e-1)))
+      return false;
+    if (!Unary && m != (s+i))
+      return false;
+  }
+  return true;
+}
+
+bool X86::isPALIGNRMask(ShuffleVectorSDNode *N) {
+  SmallVector<int, 8> M;
+  N->getMask(M);
+  return ::isPALIGNRMask(M, N->getValueType(0), true);
+}
+
 /// isSHUFPMask - Return true if the specified VECTOR_SHUFFLE operand
 /// specifies a shuffle of elements that is suitable for input to SHUFP*.
 static bool isSHUFPMask(const SmallVectorImpl<int> &Mask, EVT VT) {
@@ -2733,8 +2783,7 @@
 }
 
 /// getShuffleSHUFImmediate - Return the appropriate immediate to shuffle
-/// the specified isShuffleMask VECTOR_SHUFFLE mask with PSHUF* and SHUFP*
-/// instructions.
+/// the specified VECTOR_SHUFFLE mask with PSHUF* and SHUFP* instructions.
 unsigned X86::getShuffleSHUFImmediate(SDNode *N) {
   ShuffleVectorSDNode *SVOp = cast<ShuffleVectorSDNode>(N);
   int NumOperands = SVOp->getValueType(0).getVectorNumElements();
@@ -2753,8 +2802,7 @@
 }
 
 /// getShufflePSHUFHWImmediate - Return the appropriate immediate to shuffle
-/// the specified isShuffleMask VECTOR_SHUFFLE mask with PSHUFHW
-/// instructions.
+/// the specified VECTOR_SHUFFLE mask with the PSHUFHW instruction.
 unsigned X86::getShufflePSHUFHWImmediate(SDNode *N) {
   ShuffleVectorSDNode *SVOp = cast<ShuffleVectorSDNode>(N);
   unsigned Mask = 0;
@@ -2770,8 +2818,7 @@
 }
 
 /// getShufflePSHUFLWImmediate - Return the appropriate immediate to shuffle
-/// the specified isShuffleMask VECTOR_SHUFFLE mask with PSHUFLW
-/// instructions.
+/// the specified VECTOR_SHUFFLE mask with the PSHUFLW instruction.
 unsigned X86::getShufflePSHUFLWImmediate(SDNode *N) {
   ShuffleVectorSDNode *SVOp = cast<ShuffleVectorSDNode>(N);
   unsigned Mask = 0;
@@ -2786,6 +2833,23 @@
   return Mask;
 }
 
+/// getShufflePALIGNRImmediate - Return the appropriate immediate to shuffle
+/// the specified VECTOR_SHUFFLE mask with the PALIGNR instruction.
+unsigned X86::getShufflePALIGNRImmediate(SDNode *N) {
+  ShuffleVectorSDNode *SVOp = cast<ShuffleVectorSDNode>(N);
+  EVT VVT = N->getValueType(0);
+  unsigned EltSize = VVT.getVectorElementType().getSizeInBits() >> 3;
+  int Val = 0;
+
+  unsigned i, e;
+  for (i = 0, e = VVT.getVectorNumElements(); i != e; ++i) {
+    Val = SVOp->getMaskElt(i);
+    if (Val >= 0)
+      break;
+  }
+  return (Val - i) * EltSize;
+}
+
 /// isZeroNode - Returns true if Elt is a constant zero or a floating point
 /// constant +0.0.
 bool X86::isZeroNode(SDValue Elt) {
@@ -7274,7 +7338,7 @@
   if (VT.getSizeInBits() == 64)
     return false;
 
-  // FIXME: pshufb, blends, palignr, shifts.
+  // FIXME: pshufb, blends, shifts.
   return (VT.getVectorNumElements() == 2 ||
           ShuffleVectorSDNode::isSplatMask(&M[0], VT) ||
           isMOVLMask(M, VT) ||
@@ -7282,6 +7346,7 @@
           isPSHUFDMask(M, VT) ||
           isPSHUFHWMask(M, VT) ||
           isPSHUFLWMask(M, VT) ||
+          isPALIGNRMask(M, VT, Subtarget->hasSSSE3()) ||
           isUNPCKLMask(M, VT) ||
           isUNPCKHMask(M, VT) ||
           isUNPCKL_v_undef_Mask(M, VT) ||
diff --git a/lib/Target/X86/X86ISelLowering.h b/lib/Target/X86/X86ISelLowering.h
index 2f7b8ba..66a9107 100644
--- a/lib/Target/X86/X86ISelLowering.h
+++ b/lib/Target/X86/X86ISelLowering.h
@@ -323,21 +323,27 @@
     /// specifies a shuffle of elements that is suitable for input to MOVDDUP.
     bool isMOVDDUPMask(ShuffleVectorSDNode *N);
 
+    /// isPALIGNRMask - Return true if the specified VECTOR_SHUFFLE operand
+    /// specifies a shuffle of elements that is suitable for input to PALIGNR.
+    bool isPALIGNRMask(ShuffleVectorSDNode *N);
+
     /// getShuffleSHUFImmediate - Return the appropriate immediate to shuffle
     /// the specified isShuffleMask VECTOR_SHUFFLE mask with PSHUF* and SHUFP*
     /// instructions.
     unsigned getShuffleSHUFImmediate(SDNode *N);
 
     /// getShufflePSHUFHWImmediate - Return the appropriate immediate to shuffle
-    /// the specified isShuffleMask VECTOR_SHUFFLE mask with PSHUFHW
-    /// instructions.
+    /// the specified VECTOR_SHUFFLE mask with PSHUFHW instruction.
     unsigned getShufflePSHUFHWImmediate(SDNode *N);
 
-    /// getShufflePSHUFKWImmediate - Return the appropriate immediate to shuffle
-    /// the specified isShuffleMask VECTOR_SHUFFLE mask with PSHUFLW
-    /// instructions.
+    /// getShufflePSHUFLWImmediate - Return the appropriate immediate to shuffle
+    /// the specified VECTOR_SHUFFLE mask with PSHUFLW instruction.
     unsigned getShufflePSHUFLWImmediate(SDNode *N);
 
+    /// getShufflePALIGNRImmediate - Return the appropriate immediate to shuffle
+    /// the specified VECTOR_SHUFFLE mask with the PALIGNR instruction.
+    unsigned getShufflePALIGNRImmediate(SDNode *N);
+
     /// isZeroNode - Returns true if Elt is a constant zero or a floating point
     /// constant +0.0.
     bool isZeroNode(SDValue Elt);
diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td
index 96fc932..f4e97c9 100644
--- a/lib/Target/X86/X86InstrSSE.td
+++ b/lib/Target/X86/X86InstrSSE.td
@@ -197,6 +197,12 @@
   return getI8Imm(X86::getShufflePSHUFLWImmediate(N));
 }]>;
 
+// SHUFFLE_get_palign_imm xform function: convert vector_shuffle mask to
+// a PALIGNR imm.
+def SHUFFLE_get_palign_imm : SDNodeXForm<vector_shuffle, [{
+  return getI8Imm(X86::getShufflePALIGNRImmediate(N));
+}]>;
+
 def splat_lo : PatFrag<(ops node:$lhs, node:$rhs),
                        (vector_shuffle node:$lhs, node:$rhs), [{
   ShuffleVectorSDNode *SVOp = cast<ShuffleVectorSDNode>(N);
@@ -283,6 +289,11 @@
   return X86::isPSHUFLWMask(cast<ShuffleVectorSDNode>(N));
 }], SHUFFLE_get_pshuflw_imm>;
 
+def palign : PatFrag<(ops node:$lhs, node:$rhs),
+                     (vector_shuffle node:$lhs, node:$rhs), [{
+  return X86::isPALIGNRMask(cast<ShuffleVectorSDNode>(N));
+}], SHUFFLE_get_palign_imm>;
+
 //===----------------------------------------------------------------------===//
 // SSE scalar FP Instructions
 //===----------------------------------------------------------------------===//
@@ -2062,6 +2073,7 @@
 defm PACKUSWB : PDI_binop_rm_int<0x67, "packuswb", int_x86_sse2_packuswb_128>;
 
 // Shuffle and unpack instructions
+let AddedComplexity = 5 in {
 def PSHUFDri : PDIi8<0x70, MRMSrcReg,
                      (outs VR128:$dst), (ins VR128:$src1, i8imm:$src2),
                      "pshufd\t{$src2, $src1, $dst|$dst, $src1, $src2}",
@@ -2073,6 +2085,7 @@
                      [(set VR128:$dst, (v4i32 (pshufd:$src2
                                              (bc_v4i32(memopv2i64 addr:$src1)),
                                              (undef))))]>;
+}                                             
 
 // SSE2 with ImmT == Imm8 and XS prefix.
 def PSHUFHWri : Ii8<0x70, MRMSrcReg,
@@ -2839,6 +2852,26 @@
                               imm:$src3))]>, OpSize;
 }
 
+// palignr patterns.
+let AddedComplexity = 5 in {
+def : Pat<(v4i32 (palign:$src3 VR128:$src1, VR128:$src2)),
+          (PALIGNR128rr VR128:$src2, VR128:$src1,
+                        (SHUFFLE_get_palign_imm VR128:$src3))>,
+      Requires<[HasSSSE3]>;
+def : Pat<(v4f32 (palign:$src3 VR128:$src1, VR128:$src2)),
+          (PALIGNR128rr VR128:$src2, VR128:$src1,
+                        (SHUFFLE_get_palign_imm VR128:$src3))>,
+      Requires<[HasSSSE3]>;
+def : Pat<(v8i16 (palign:$src3 VR128:$src1, VR128:$src2)),
+          (PALIGNR128rr VR128:$src2, VR128:$src1,
+                        (SHUFFLE_get_palign_imm VR128:$src3))>,
+      Requires<[HasSSSE3]>;
+def : Pat<(v16i8 (palign:$src3 VR128:$src1, VR128:$src2)),
+          (PALIGNR128rr VR128:$src2, VR128:$src1,
+                        (SHUFFLE_get_palign_imm VR128:$src3))>,
+      Requires<[HasSSSE3]>;
+}      
+
 def : Pat<(X86pshufb VR128:$src, VR128:$mask),
           (PSHUFBrr128 VR128:$src, VR128:$mask)>, Requires<[HasSSSE3]>;
 def : Pat<(X86pshufb VR128:$src, (bc_v16i8 (memopv2i64 addr:$mask))),
diff --git a/test/CodeGen/X86/palignr.ll b/test/CodeGen/X86/palignr.ll
new file mode 100644
index 0000000..3812c72
--- /dev/null
+++ b/test/CodeGen/X86/palignr.ll
@@ -0,0 +1,58 @@
+; RUN: llc < %s -march=x86 -mcpu=core2 | FileCheck %s
+; RUN: llc < %s -march=x86 -mcpu=yonah | FileCheck --check-prefix=YONAH %s
+
+define <4 x i32> @test1(<4 x i32> %A, <4 x i32> %B) nounwind {
+; CHECK: pshufd
+; CHECK-YONAH: pshufd
+  %C = shufflevector <4 x i32> %A, <4 x i32> undef, <4 x i32> < i32 1, i32 2, i32 3, i32 0 >
+	ret <4 x i32> %C
+}
+
+define <4 x i32> @test2(<4 x i32> %A, <4 x i32> %B) nounwind {
+; CHECK: palignr
+; CHECK-YONAH: shufps
+  %C = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> < i32 1, i32 2, i32 3, i32 4 >
+	ret <4 x i32> %C
+}
+
+define <4 x i32> @test3(<4 x i32> %A, <4 x i32> %B) nounwind {
+; CHECK: palignr
+  %C = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> < i32 1, i32 2, i32 undef, i32 4 >
+	ret <4 x i32> %C
+}
+
+define <4 x i32> @test4(<4 x i32> %A, <4 x i32> %B) nounwind {
+; CHECK: palignr
+  %C = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> < i32 6, i32 7, i32 undef, i32 1 >
+	ret <4 x i32> %C
+}
+
+define <4 x float> @test5(<4 x float> %A, <4 x float> %B) nounwind {
+; CHECK: palignr
+  %C = shufflevector <4 x float> %A, <4 x float> %B, <4 x i32> < i32 6, i32 7, i32 undef, i32 1 >
+	ret <4 x float> %C
+}
+
+define <8 x i16> @test6(<8 x i16> %A, <8 x i16> %B) nounwind {
+; CHECK: palignr
+  %C = shufflevector <8 x i16> %A, <8 x i16> %B, <8 x i32> < i32 3, i32 4, i32 undef, i32 6, i32 7, i32 8, i32 9, i32 10 >
+	ret <8 x i16> %C
+}
+
+define <8 x i16> @test7(<8 x i16> %A, <8 x i16> %B) nounwind {
+; CHECK: palignr
+  %C = shufflevector <8 x i16> %A, <8 x i16> %B, <8 x i32> < i32 undef, i32 6, i32 undef, i32 8, i32 9, i32 10, i32 11, i32 12 >
+	ret <8 x i16> %C
+}
+
+define <8 x i16> @test8(<8 x i16> %A, <8 x i16> %B) nounwind {
+; CHECK: palignr
+  %C = shufflevector <8 x i16> %A, <8 x i16> %B, <8 x i32> < i32 undef, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0 >
+	ret <8 x i16> %C
+}
+
+define <16 x i8> @test9(<16 x i8> %A, <16 x i8> %B) nounwind {
+; CHECK: palignr
+  %C = shufflevector <16 x i8> %A, <16 x i8> %B, <16 x i32> < i32 5, i32 6, i32 7, i32 undef, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20 >
+	ret <16 x i8> %C
+}
diff --git a/test/CodeGen/X86/vec_shuffle-22.ll b/test/CodeGen/X86/vec_shuffle-22.ll
index 5307ced..1cf37d4 100644
--- a/test/CodeGen/X86/vec_shuffle-22.ll
+++ b/test/CodeGen/X86/vec_shuffle-22.ll
@@ -1,19 +1,15 @@
-; RUN: llc < %s -march=x86 -mcpu=pentium-m -o %t
-; RUN: grep movlhps %t | count 1
-; RUN: grep pshufd %t | count 1
-; RUN: llc < %s -march=x86 -mcpu=core2 -o %t
-; RUN: grep movlhps %t | count 1
-; RUN: grep movddup %t | count 1
+; RUN: llc < %s -march=x86 -mcpu=pentium-m  | FileCheck %s
 
 define <4 x float> @t1(<4 x float> %a) nounwind  {
-entry:
-        %tmp1 = shufflevector <4 x float> %a, <4 x float> undef, <4 x i32> < i32 0, i32 1, i32 0, i32 1 >       ; <<4 x float>> [#uses=1]
-        ret <4 x float> %tmp1
+; CHECK: movlhps
+  %tmp1 = shufflevector <4 x float> %a, <4 x float> undef, <4 x i32> < i32 0, i32 1, i32 0, i32 1 >       ; <<4 x float>> [#uses=1]
+  ret <4 x float> %tmp1
 }
 
 define <4 x i32> @t2(<4 x i32>* %a) nounwind {
-entry:
-        %tmp1 = load <4 x i32>* %a;
+; CHECK: pshufd
+; CHECK: ret
+  %tmp1 = load <4 x i32>* %a;
 	%tmp2 = shufflevector <4 x i32> %tmp1, <4 x i32> undef, <4 x i32> < i32 0, i32 1, i32 0, i32 1 >		; <<4 x i32>> [#uses=1]
 	ret <4 x i32> %tmp2
 }
diff --git a/test/CodeGen/X86/vec_shuffle-9.ll b/test/CodeGen/X86/vec_shuffle-9.ll
index 2bef24d..fc16a26 100644
--- a/test/CodeGen/X86/vec_shuffle-9.ll
+++ b/test/CodeGen/X86/vec_shuffle-9.ll
@@ -1,9 +1,10 @@
-; RUN: llc < %s -march=x86 -mattr=+sse2 -o %t
-; RUN: grep punpck %t | count 2
-; RUN: not grep pextrw %t
+; RUN: llc < %s -march=x86 -mattr=+sse2 | FileCheck %s
 
 define <4 x i32> @test(i8** %ptr) {
-entry:
+; CHECK: xorps
+; CHECK: punpcklbw
+; CHECK: punpcklwd
+
 	%tmp = load i8** %ptr		; <i8*> [#uses=1]
 	%tmp.upgrd.1 = bitcast i8* %tmp to float*		; <float*> [#uses=1]
 	%tmp.upgrd.2 = load float* %tmp.upgrd.1		; <float> [#uses=1]