diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index 91fc26cd1e3..81416341658 100644
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -4304,8 +4304,18 @@ X86TargetLowering::LowerBUILD_VECTOR(SDValue Op, SelectionDAG &DAG) const {
     //   Step 2: unpcklps X, Y ==>    <3, 2, 1, 0>
     unsigned EltStride = NumElems >> 1;
     while (EltStride != 0) {
-      for (unsigned i = 0; i < EltStride; ++i)
+      for (unsigned i = 0; i < EltStride; ++i) {
+        // If V[i+EltStride] is undef and this is the first round of mixing,
+        // then it is safe to just drop this shuffle: V[i] is already in the
+        // right place, the one element (since it's the first round) being
+        // inserted as undef can be dropped.  This isn't safe for successive
+        // rounds because they will permute elements within both vectors.
+        if (V[i+EltStride].getOpcode() == ISD::UNDEF &&
+            EltStride == NumElems/2)
+          continue;
+        
         V[i] = getUnpackl(DAG, dl, VT, V[i], V[i + EltStride]);
+      }
       EltStride >>= 1;
     }
     return V[0];
diff --git a/test/CodeGen/X86/sse1.ll b/test/CodeGen/X86/sse1.ll
index 6e7aad84de6..73f88aec643 100644
--- a/test/CodeGen/X86/sse1.ll
+++ b/test/CodeGen/X86/sse1.ll
@@ -18,3 +18,28 @@ define <8 x i16> @test2(<8 x i32> %a) nounwind {
 ;  %c = sext <4 x i16> %a to <4 x i32>             ; <<4 x i32>> [#uses=1]
 ;  ret <4 x i32> %c
 ;}
+
+; This should not emit shuffles to populate the top 2 elements of the 4-element
+; vector that this ends up returning.
+; rdar://8368414
+define <2 x float> @test4(<2 x float> %A, <2 x float> %B) nounwind {
+entry:
+  %tmp7 = extractelement <2 x float> %A, i32 0
+  %tmp5 = extractelement <2 x float> %A, i32 1
+  %tmp3 = extractelement <2 x float> %B, i32 0
+  %tmp1 = extractelement <2 x float> %B, i32 1
+  %add.r = fadd float %tmp7, %tmp3
+  %add.i = fsub float %tmp5, %tmp1
+  %tmp11 = insertelement <2 x float> undef, float %add.r, i32 0
+  %tmp9 = insertelement <2 x float> %tmp11, float %add.i, i32 1
+  ret <2 x float> %tmp9
+; CHECK: test4:
+; CHECK-NOT: shufps	$16
+; CHECK: shufps	$1, 
+; CHECK-NOT: shufps	$16
+; CHECK: shufps	$1, 
+; CHECK-NOT: shufps	$16
+; CHECK: unpcklps
+; CHECK-NOT: shufps	$16
+; CHECK: ret
+}