diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index 6d30d0c26e0a..859ad3ecd9bb 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -5936,50 +5936,31 @@ Value *CodeGenFunction::EmitX86BuiltinExpr(unsigned BuiltinID,
     unsigned NumLanes = NumElts / 16;
     unsigned NumLaneElts = NumElts / NumLanes;
 
-    // If palignr is shifting the pair of input vectors less than the size of
-    // a lane, emit a shuffle instruction.
-    if (ShiftVal <= NumLaneElts) {
-      SmallVector<llvm::Constant*, 32> Indices;
-      // 256-bit palignr operates on 128-bit lanes so we need to handle that
-      for (unsigned l = 0; l != NumElts; l += NumLaneElts) {
-        for (unsigned i = 0; i != NumLaneElts; ++i) {
-          unsigned Idx = ShiftVal + i;
-          if (Idx >= NumLaneElts)
-            Idx += NumElts - NumLaneElts; // End of lane, switch operand.
-          Indices.push_back(llvm::ConstantInt::get(Int32Ty, Idx + l));
-        }
-      }
-
-      Value* SV = llvm::ConstantVector::get(Indices);
-      return Builder.CreateShuffleVector(Ops[1], Ops[0], SV, "palignr");
-    }
-
     // If palignr is shifting the pair of vectors more than the size of two
     // lanes, emit zero.
     if (ShiftVal >= (2 * NumLaneElts))
       return llvm::Constant::getNullValue(ConvertType(E->getType()));
 
     // If palignr is shifting the pair of input vectors more than one lane,
-    // but less than two lanes, emit a shift.
-    llvm::Type *VecTy = llvm::VectorType::get(Int64Ty, NumElts/8);
-
-    Ops[0] = Builder.CreateBitCast(Ops[0], VecTy, "cast");
-    Ops[1] = llvm::ConstantInt::get(Int32Ty, (ShiftVal-NumLaneElts) * 8);
-
-    Intrinsic::ID ID;
-    switch (BuiltinID) {
-    default: llvm_unreachable("Unsupported intrinsic!");
-    case X86::BI__builtin_ia32_palignr128:
-      ID = Intrinsic::x86_sse2_psrl_dq;
-      break;
-    case X86::BI__builtin_ia32_palignr256:
-      ID = Intrinsic::x86_avx2_psrl_dq;
-      break;
+    // but less than two lanes, convert to shifting in zeroes.
+    if (ShiftVal > NumLaneElts) {
+      ShiftVal -= NumLaneElts;
+      Ops[0] = llvm::Constant::getNullValue(Ops[0]->getType());
     }
 
-    // create i32 constant
-    llvm::Function *F = CGM.getIntrinsic(ID);
-    return Builder.CreateCall(F, makeArrayRef(Ops.data(), 2), "palignr");
+    SmallVector<llvm::Constant*, 32> Indices;
+    // 256-bit palignr operates on 128-bit lanes so we need to handle that
+    for (unsigned l = 0; l != NumElts; l += NumLaneElts) {
+      for (unsigned i = 0; i != NumLaneElts; ++i) {
+        unsigned Idx = ShiftVal + i;
+        if (Idx >= NumLaneElts)
+          Idx += NumElts - NumLaneElts; // End of lane, switch operand.
+        Indices.push_back(llvm::ConstantInt::get(Int32Ty, Idx + l));
+      }
+    }
+
+    Value* SV = llvm::ConstantVector::get(Indices);
+    return Builder.CreateShuffleVector(Ops[1], Ops[0], SV, "palignr");
   }
   case X86::BI__builtin_ia32_pslldqi256: {
     // Shift value is in bits so divide by 8.
diff --git a/clang/test/CodeGen/avx2-builtins.c b/clang/test/CodeGen/avx2-builtins.c
index a30f590ad169..69781d51d6c5 100644
--- a/clang/test/CodeGen/avx2-builtins.c
+++ b/clang/test/CodeGen/avx2-builtins.c
@@ -96,7 +96,7 @@ __m256i test_mm256_alignr_epi8(__m256i a, __m256i b) {
 }
 
 __m256i test2_mm256_alignr_epi8(__m256i a, __m256i b) {
-  // CHECK: @llvm.x86.avx2.psrl.dq({{.*}}, i32 8)
+  // CHECK: shufflevector <32 x i8> %5, <32 x i8> zeroinitializer, <32 x i32> <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 32, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 48>
   return _mm256_alignr_epi8(a, b, 17);
 }
 
diff --git a/clang/test/CodeGen/sse-builtins.c b/clang/test/CodeGen/sse-builtins.c
index 267deead700a..2893e611b193 100644
--- a/clang/test/CodeGen/sse-builtins.c
+++ b/clang/test/CodeGen/sse-builtins.c
@@ -574,6 +574,6 @@ __m128i test_mm_alignr_epi8(__m128i a, __m128i b) {
 }
 
 __m128i test2_mm_alignr_epi8(__m128i a, __m128i b) {
-  // CHECK: @llvm.x86.sse2.psrl.dq({{.*}}, i32 8)
+  // CHECK: shufflevector <16 x i8> %5, <16 x i8> zeroinitializer, <16 x i32> <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16>
   return _mm_alignr_epi8(a, b, 17);
 }