Index: lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp
===================================================================
--- lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp
+++ lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp
@@ -2422,11 +2422,6 @@
     Res = WidenVecRes_Convert(N);
     break;
 
-  case ISD::BITREVERSE:
-  case ISD::BSWAP:
-  case ISD::CTLZ:
-  case ISD::CTPOP:
-  case ISD::CTTZ:
   case ISD::FABS:
   case ISD::FCEIL:
   case ISD::FCOS:
@@ -2437,12 +2432,36 @@
   case ISD::FLOG10:
   case ISD::FLOG2:
   case ISD::FNEARBYINT:
-  case ISD::FNEG:
   case ISD::FRINT:
   case ISD::FROUND:
   case ISD::FSIN:
   case ISD::FSQRT:
-  case ISD::FTRUNC:
+  case ISD::FTRUNC: {
+    // We're going to widen this vector op to a legal type by padding with undef
+    // elements. If the wide vector op is eventually going to be expanded to
+    // scalar libcalls, then unroll into scalar ops now to avoid unnecessary
+    // libcalls on the undef elements. We are assuming that if the scalar op
+    // requires expanding, then the vector op needs expanding too.
+    EVT VT = N->getValueType(0);
+    if (TLI.isOperationExpand(N->getOpcode(), VT.getScalarType())) {
+      EVT WideVecVT = TLI.getTypeToTransformTo(*DAG.getContext(), VT);
+      assert(!TLI.isOperationLegalOrCustom(N->getOpcode(), WideVecVT) &&
+             "Target supports vector op, but scalar requires expansion?");
+      Res = DAG.UnrollVectorOp(N, WideVecVT.getVectorNumElements());
+      break;
+    }
+  }
+  // If the target has custom/legal support for the scalar FP intrinsic ops
+  // (they are probably not destined to become libcalls), then widen those like
+  // any other unary ops.
+  LLVM_FALLTHROUGH;
+
+  case ISD::BITREVERSE:
+  case ISD::BSWAP:
+  case ISD::CTLZ:
+  case ISD::CTPOP:
+  case ISD::CTTZ:
+  case ISD::FNEG:
   case ISD::FCANONICALIZE:
     Res = WidenVecRes_Unary(N);
     break;
Index: test/CodeGen/AArch64/vec-libcalls.ll
===================================================================
--- test/CodeGen/AArch64/vec-libcalls.ll
+++ test/CodeGen/AArch64/vec-libcalls.ll
@@ -35,22 +35,12 @@
 define <1 x float> @sin_v1f32(<1 x float> %x) nounwind {
 ; CHECK-LABEL: sin_v1f32:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    sub sp, sp, #48 // =48
+; CHECK-NEXT:    str x30, [sp, #-16]! // 8-byte Folded Spill
 ; CHECK-NEXT:    // kill: def $d0 killed $d0 def $q0
-; CHECK-NEXT:    str q0, [sp] // 16-byte Folded Spill
-; CHECK-NEXT:    mov s0, v0.s[1]
-; CHECK-NEXT:    str x30, [sp, #32] // 8-byte Folded Spill
-; CHECK-NEXT:    bl sinf
-; CHECK-NEXT:    str d0, [sp, #16] // 16-byte Folded Spill
-; CHECK-NEXT:    ldr q0, [sp] // 16-byte Folded Reload
 ; CHECK-NEXT:    // kill: def $s0 killed $s0 killed $q0
 ; CHECK-NEXT:    bl sinf
-; CHECK-NEXT:    ldr q1, [sp, #16] // 16-byte Folded Reload
-; CHECK-NEXT:    ldr x30, [sp, #32] // 8-byte Folded Reload
-; CHECK-NEXT:    // kill: def $s0 killed $s0 def $q0
-; CHECK-NEXT:    mov v0.s[1], v1.s[0]
-; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $q0
-; CHECK-NEXT:    add sp, sp, #48 // =48
+; CHECK-NEXT:    // kill: def $s0 killed $s0 def $d0
+; CHECK-NEXT:    ldr x30, [sp], #16 // 8-byte Folded Reload
 ; CHECK-NEXT:    ret
   %r = call <1 x float> @llvm.sin.v1f32(<1 x float> %x)
   ret <1 x float> %r
@@ -100,16 +90,9 @@
 ; CHECK-NEXT:    mov s0, v0.s[2]
 ; CHECK-NEXT:    bl sinf
 ; CHECK-NEXT:    ldr q1, [sp] // 16-byte Folded Reload
-; CHECK-NEXT:    // kill: def $s0 killed $s0 def $q0
-; CHECK-NEXT:    mov v1.s[2], v0.s[0]
-; CHECK-NEXT:    ldr q0, [sp, #16] // 16-byte Folded Reload
-; CHECK-NEXT:    str q1, [sp] // 16-byte Folded Spill
-; CHECK-NEXT:    mov s0, v0.s[3]
-; CHECK-NEXT:    bl sinf
-; CHECK-NEXT:    ldr q1, [sp] // 16-byte Folded Reload
 ; CHECK-NEXT:    ldr x30, [sp, #32] // 8-byte Folded Reload
 ; CHECK-NEXT:    // kill: def $s0 killed $s0 def $q0
-; CHECK-NEXT:    mov v1.s[3], v0.s[0]
+; CHECK-NEXT:    mov v1.s[2], v0.s[0]
 ; CHECK-NEXT:    mov v0.16b, v1.16b
 ; CHECK-NEXT:    add sp, sp, #48 // =48
 ; CHECK-NEXT:    ret
@@ -195,28 +178,20 @@
 define <6 x float> @sin_v6f32(<6 x float> %x) nounwind {
 ; CHECK-LABEL: sin_v6f32:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    sub sp, sp, #80 // =80
-; CHECK-NEXT:    str d12, [sp, #32] // 8-byte Folded Spill
-; CHECK-NEXT:    mov v12.16b, v0.16b
-; CHECK-NEXT:    mov v0.16b, v5.16b
-; CHECK-NEXT:    stp d11, d10, [sp, #40] // 8-byte Folded Spill
-; CHECK-NEXT:    stp d9, d8, [sp, #56] // 8-byte Folded Spill
-; CHECK-NEXT:    str x30, [sp, #72] // 8-byte Folded Spill
-; CHECK-NEXT:    mov v8.16b, v4.16b
-; CHECK-NEXT:    mov v9.16b, v3.16b
-; CHECK-NEXT:    mov v10.16b, v2.16b
-; CHECK-NEXT:    mov v11.16b, v1.16b
-; CHECK-NEXT:    bl sinf
-; CHECK-NEXT:    str d0, [sp, #16] // 16-byte Folded Spill
-; CHECK-NEXT:    mov v0.16b, v8.16b
+; CHECK-NEXT:    stp d13, d12, [sp, #-64]! // 8-byte Folded Spill
+; CHECK-NEXT:    stp d11, d10, [sp, #16] // 8-byte Folded Spill
+; CHECK-NEXT:    stp d9, d8, [sp, #32] // 8-byte Folded Spill
+; CHECK-NEXT:    str x30, [sp, #48] // 8-byte Folded Spill
+; CHECK-NEXT:    mov v8.16b, v5.16b
+; CHECK-NEXT:    mov v9.16b, v4.16b
+; CHECK-NEXT:    mov v10.16b, v3.16b
+; CHECK-NEXT:    mov v11.16b, v2.16b
+; CHECK-NEXT:    mov v12.16b, v1.16b
 ; CHECK-NEXT:    bl sinf
-; CHECK-NEXT:    mov v1.16b, v0.16b
-; CHECK-NEXT:    ldr q0, [sp, #16] // 16-byte Folded Reload
-; CHECK-NEXT:    mov v1.s[1], v0.s[0]
+; CHECK-NEXT:    mov v13.16b, v0.16b
 ; CHECK-NEXT:    mov v0.16b, v12.16b
-; CHECK-NEXT:    str q1, [sp] // 16-byte Folded Spill
 ; CHECK-NEXT:    bl sinf
-; CHECK-NEXT:    mov v8.16b, v0.16b
+; CHECK-NEXT:    mov v12.16b, v0.16b
 ; CHECK-NEXT:    mov v0.16b, v11.16b
 ; CHECK-NEXT:    bl sinf
 ; CHECK-NEXT:    mov v11.16b, v0.16b
@@ -225,18 +200,19 @@
 ; CHECK-NEXT:    mov v10.16b, v0.16b
 ; CHECK-NEXT:    mov v0.16b, v9.16b
 ; CHECK-NEXT:    bl sinf
-; CHECK-NEXT:    ldp q4, q5, [sp] // 16-byte Folded Reload
-; CHECK-NEXT:    mov v3.16b, v0.16b
+; CHECK-NEXT:    mov v9.16b, v0.16b
 ; CHECK-NEXT:    mov v0.16b, v8.16b
-; CHECK-NEXT:    mov v1.16b, v11.16b
-; CHECK-NEXT:    mov v2.16b, v10.16b
-; CHECK-NEXT:    ldr x30, [sp, #72] // 8-byte Folded Reload
-; CHECK-NEXT:    ldp d9, d8, [sp, #56] // 8-byte Folded Reload
-; CHECK-NEXT:    ldp d11, d10, [sp, #40] // 8-byte Folded Reload
-; CHECK-NEXT:    ldr d12, [sp, #32] // 8-byte Folded Reload
-; CHECK-NEXT:    // kill: def $s4 killed $s4 killed $q4
-; CHECK-NEXT:    // kill: def $s5 killed $s5 killed $q5
-; CHECK-NEXT:    add sp, sp, #80 // =80
+; CHECK-NEXT:    bl sinf
+; CHECK-NEXT:    mov v2.16b, v11.16b
+; CHECK-NEXT:    mov v3.16b, v10.16b
+; CHECK-NEXT:    mov v4.16b, v9.16b
+; CHECK-NEXT:    ldr x30, [sp, #48] // 8-byte Folded Reload
+; CHECK-NEXT:    ldp d9, d8, [sp, #32] // 8-byte Folded Reload
+; CHECK-NEXT:    ldp d11, d10, [sp, #16] // 8-byte Folded Reload
+; CHECK-NEXT:    mov v5.16b, v0.16b
+; CHECK-NEXT:    mov v0.16b, v13.16b
+; CHECK-NEXT:    mov v1.16b, v12.16b
+; CHECK-NEXT:    ldp d13, d12, [sp], #64 // 8-byte Folded Reload
 ; CHECK-NEXT:    ret
   %r = call <6 x float> @llvm.sin.v6f32(<6 x float> %x)
   ret <6 x float> %r
@@ -306,16 +282,9 @@
 ; CHECK-NEXT:    mov s0, v0.s[2]
 ; CHECK-NEXT:    bl cosf
 ; CHECK-NEXT:    ldr q1, [sp] // 16-byte Folded Reload
-; CHECK-NEXT:    // kill: def $s0 killed $s0 def $q0
-; CHECK-NEXT:    mov v1.s[2], v0.s[0]
-; CHECK-NEXT:    ldr q0, [sp, #16] // 16-byte Folded Reload
-; CHECK-NEXT:    str q1, [sp] // 16-byte Folded Spill
-; CHECK-NEXT:    mov s0, v0.s[3]
-; CHECK-NEXT:    bl cosf
-; CHECK-NEXT:    ldr q1, [sp] // 16-byte Folded Reload
 ; CHECK-NEXT:    ldr x30, [sp, #32] // 8-byte Folded Reload
 ; CHECK-NEXT:    // kill: def $s0 killed $s0 def $q0
-; CHECK-NEXT:    mov v1.s[3], v0.s[0]
+; CHECK-NEXT:    mov v1.s[2], v0.s[0]
 ; CHECK-NEXT:    mov v0.16b, v1.16b
 ; CHECK-NEXT:    add sp, sp, #48 // =48
 ; CHECK-NEXT:    ret
@@ -343,16 +312,9 @@
 ; CHECK-NEXT:    mov s0, v0.s[2]
 ; CHECK-NEXT:    bl expf
 ; CHECK-NEXT:    ldr q1, [sp] // 16-byte Folded Reload
-; CHECK-NEXT:    // kill: def $s0 killed $s0 def $q0
-; CHECK-NEXT:    mov v1.s[2], v0.s[0]
-; CHECK-NEXT:    ldr q0, [sp, #16] // 16-byte Folded Reload
-; CHECK-NEXT:    str q1, [sp] // 16-byte Folded Spill
-; CHECK-NEXT:    mov s0, v0.s[3]
-; CHECK-NEXT:    bl expf
-; CHECK-NEXT:    ldr q1, [sp] // 16-byte Folded Reload
 ; CHECK-NEXT:    ldr x30, [sp, #32] // 8-byte Folded Reload
 ; CHECK-NEXT:    // kill: def $s0 killed $s0 def $q0
-; CHECK-NEXT:    mov v1.s[3], v0.s[0]
+; CHECK-NEXT:    mov v1.s[2], v0.s[0]
 ; CHECK-NEXT:    mov v0.16b, v1.16b
 ; CHECK-NEXT:    add sp, sp, #48 // =48
 ; CHECK-NEXT:    ret
@@ -380,16 +342,9 @@
 ; CHECK-NEXT:    mov s0, v0.s[2]
 ; CHECK-NEXT:    bl exp2f
 ; CHECK-NEXT:    ldr q1, [sp] // 16-byte Folded Reload
-; CHECK-NEXT:    // kill: def $s0 killed $s0 def $q0
-; CHECK-NEXT:    mov v1.s[2], v0.s[0]
-; CHECK-NEXT:    ldr q0, [sp, #16] // 16-byte Folded Reload
-; CHECK-NEXT:    str q1, [sp] // 16-byte Folded Spill
-; CHECK-NEXT:    mov s0, v0.s[3]
-; CHECK-NEXT:    bl exp2f
-; CHECK-NEXT:    ldr q1, [sp] // 16-byte Folded Reload
 ; CHECK-NEXT:    ldr x30, [sp, #32] // 8-byte Folded Reload
 ; CHECK-NEXT:    // kill: def $s0 killed $s0 def $q0
-; CHECK-NEXT:    mov v1.s[3], v0.s[0]
+; CHECK-NEXT:    mov v1.s[2], v0.s[0]
 ; CHECK-NEXT:    mov v0.16b, v1.16b
 ; CHECK-NEXT:    add sp, sp, #48 // =48
 ; CHECK-NEXT:    ret
@@ -426,16 +381,9 @@
 ; CHECK-NEXT:    mov s0, v0.s[2]
 ; CHECK-NEXT:    bl logf
 ; CHECK-NEXT:    ldr q1, [sp] // 16-byte Folded Reload
-; CHECK-NEXT:    // kill: def $s0 killed $s0 def $q0
-; CHECK-NEXT:    mov v1.s[2], v0.s[0]
-; CHECK-NEXT:    ldr q0, [sp, #16] // 16-byte Folded Reload
-; CHECK-NEXT:    str q1, [sp] // 16-byte Folded Spill
-; CHECK-NEXT:    mov s0, v0.s[3]
-; CHECK-NEXT:    bl logf
-; CHECK-NEXT:    ldr q1, [sp] // 16-byte Folded Reload
 ; CHECK-NEXT:    ldr x30, [sp, #32] // 8-byte Folded Reload
 ; CHECK-NEXT:    // kill: def $s0 killed $s0 def $q0
-; CHECK-NEXT:    mov v1.s[3], v0.s[0]
+; CHECK-NEXT:    mov v1.s[2], v0.s[0]
 ; CHECK-NEXT:    mov v0.16b, v1.16b
 ; CHECK-NEXT:    add sp, sp, #48 // =48
 ; CHECK-NEXT:    ret
@@ -463,16 +411,9 @@
 ; CHECK-NEXT:    mov s0, v0.s[2]
 ; CHECK-NEXT:    bl log10f
 ; CHECK-NEXT:    ldr q1, [sp] // 16-byte Folded Reload
-; CHECK-NEXT:    // kill: def $s0 killed $s0 def $q0
-; CHECK-NEXT:    mov v1.s[2], v0.s[0]
-; CHECK-NEXT:    ldr q0, [sp, #16] // 16-byte Folded Reload
-; CHECK-NEXT:    str q1, [sp] // 16-byte Folded Spill
-; CHECK-NEXT:    mov s0, v0.s[3]
-; CHECK-NEXT:    bl log10f
-; CHECK-NEXT:    ldr q1, [sp] // 16-byte Folded Reload
 ; CHECK-NEXT:    ldr x30, [sp, #32] // 8-byte Folded Reload
 ; CHECK-NEXT:    // kill: def $s0 killed $s0 def $q0
-; CHECK-NEXT:    mov v1.s[3], v0.s[0]
+; CHECK-NEXT:    mov v1.s[2], v0.s[0]
 ; CHECK-NEXT:    mov v0.16b, v1.16b
 ; CHECK-NEXT:    add sp, sp, #48 // =48
 ; CHECK-NEXT:    ret
@@ -500,16 +441,9 @@
 ; CHECK-NEXT:    mov s0, v0.s[2]
 ; CHECK-NEXT:    bl log2f
 ; CHECK-NEXT:    ldr q1, [sp] // 16-byte Folded Reload
-; CHECK-NEXT:    // kill: def $s0 killed $s0 def $q0
-; CHECK-NEXT:    mov v1.s[2], v0.s[0]
-; CHECK-NEXT:    ldr q0, [sp, #16] // 16-byte Folded Reload
-; CHECK-NEXT:    str q1, [sp] // 16-byte Folded Spill
-; CHECK-NEXT:    mov s0, v0.s[3]
-; CHECK-NEXT:    bl log2f
-; CHECK-NEXT:    ldr q1, [sp] // 16-byte Folded Reload
 ; CHECK-NEXT:    ldr x30, [sp, #32] // 8-byte Folded Reload
 ; CHECK-NEXT:    // kill: def $s0 killed $s0 def $q0
-; CHECK-NEXT:    mov v1.s[3], v0.s[0]
+; CHECK-NEXT:    mov v1.s[2], v0.s[0]
 ; CHECK-NEXT:    mov v0.16b, v1.16b
 ; CHECK-NEXT:    add sp, sp, #48 // =48
 ; CHECK-NEXT:    ret
Index: test/CodeGen/X86/vec-libcalls.ll
===================================================================
--- test/CodeGen/X86/vec-libcalls.ll
+++ test/CodeGen/X86/vec-libcalls.ll
@@ -51,26 +51,14 @@
 ; CHECK-LABEL: sin_v2f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    subq $40, %rsp
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    callq sinf
 ; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    callq sinf
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
 ; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
 ; CHECK-NEXT:    callq sinf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
 ; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq sinf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilps $231, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,1,2,3]
-; CHECK-NEXT:    callq sinf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
 ; CHECK-NEXT:    addq $40, %rsp
 ; CHECK-NEXT:    retq
   %r = call <2 x float> @llvm.sin.v2f32(<2 x float> %x)
@@ -95,12 +83,6 @@
 ; CHECK-NEXT:    callq sinf
 ; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
 ; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilps $231, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,1,2,3]
-; CHECK-NEXT:    callq sinf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
 ; CHECK-NEXT:    addq $40, %rsp
 ; CHECK-NEXT:    retq
   %r = call <3 x float> @llvm.sin.v3f32(<3 x float> %x)
@@ -142,8 +124,7 @@
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    subq $88, %rsp
 ; CHECK-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; CHECK-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    # kill: def $xmm0 killed $xmm0 killed $ymm0
 ; CHECK-NEXT:    vzeroupper
 ; CHECK-NEXT:    callq sinf
 ; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
@@ -164,30 +145,13 @@
 ; CHECK-NEXT:    callq sinf
 ; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
 ; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovups %ymm0, (%rsp) # 32-byte Spill
 ; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; CHECK-NEXT:    # kill: def $xmm0 killed $xmm0 killed $ymm0
+; CHECK-NEXT:    vextractf128 $1, %ymm0, %xmm0
 ; CHECK-NEXT:    vzeroupper
 ; CHECK-NEXT:    callq sinf
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
-; CHECK-NEXT:    callq sinf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq sinf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilps $231, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,1,2,3]
-; CHECK-NEXT:    callq sinf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
-; CHECK-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vmovups (%rsp), %ymm1 # 32-byte Reload
+; CHECK-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; CHECK-NEXT:    addq $88, %rsp
 ; CHECK-NEXT:    retq
   %r = call <5 x float> @llvm.sin.v5f32(<5 x float> %x)
@@ -200,27 +164,15 @@
 ; CHECK-NEXT:    subq $88, %rsp
 ; CHECK-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
 ; CHECK-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
 ; CHECK-NEXT:    vzeroupper
 ; CHECK-NEXT:    callq sinf
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
 ; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
 ; CHECK-NEXT:    callq sinf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
 ; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq sinf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilps $231, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,1,2,3]
-; CHECK-NEXT:    callq sinf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
 ; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
 ; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
 ; CHECK-NEXT:    # kill: def $xmm0 killed $xmm0 killed $ymm0
@@ -256,18 +208,6 @@
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    subq $88, %rsp
 ; CHECK-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; CHECK-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq sin
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq sin
-; CHECK-NEXT:    vmovapd {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; CHECK-NEXT:    vmovapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
 ; CHECK-NEXT:    # kill: def $xmm0 killed $xmm0 killed $ymm0
 ; CHECK-NEXT:    vzeroupper
 ; CHECK-NEXT:    callq sin
@@ -277,7 +217,13 @@
 ; CHECK-NEXT:    callq sin
 ; CHECK-NEXT:    vmovapd (%rsp), %xmm1 # 16-byte Reload
 ; CHECK-NEXT:    vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; CHECK-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vmovupd %ymm0, (%rsp) # 32-byte Spill
+; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
+; CHECK-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; CHECK-NEXT:    vzeroupper
+; CHECK-NEXT:    callq sin
+; CHECK-NEXT:    vmovups (%rsp), %ymm1 # 32-byte Reload
+; CHECK-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; CHECK-NEXT:    addq $88, %rsp
 ; CHECK-NEXT:    retq
   %r = call <3 x double> @llvm.sin.v3f64(<3 x double> %x)
@@ -306,26 +252,14 @@
 ; CHECK-LABEL: cos_v2f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    subq $40, %rsp
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    callq cosf
 ; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    callq cosf
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
 ; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
 ; CHECK-NEXT:    callq cosf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
 ; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq cosf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilps $231, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,1,2,3]
-; CHECK-NEXT:    callq cosf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
 ; CHECK-NEXT:    addq $40, %rsp
 ; CHECK-NEXT:    retq
   %r = call <2 x float> @llvm.cos.v2f32(<2 x float> %x)
@@ -336,26 +270,14 @@
 ; CHECK-LABEL: exp_v2f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    subq $40, %rsp
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    callq expf
 ; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    callq expf
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
 ; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
 ; CHECK-NEXT:    callq expf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
 ; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq expf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilps $231, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,1,2,3]
-; CHECK-NEXT:    callq expf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
 ; CHECK-NEXT:    addq $40, %rsp
 ; CHECK-NEXT:    retq
   %r = call <2 x float> @llvm.exp.v2f32(<2 x float> %x)
@@ -366,26 +288,14 @@
 ; CHECK-LABEL: exp2_v2f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    subq $40, %rsp
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    callq exp2f
 ; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    callq exp2f
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
 ; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
 ; CHECK-NEXT:    callq exp2f
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
 ; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq exp2f
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilps $231, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,1,2,3]
-; CHECK-NEXT:    callq exp2f
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
 ; CHECK-NEXT:    addq $40, %rsp
 ; CHECK-NEXT:    retq
   %r = call <2 x float> @llvm.exp2.v2f32(<2 x float> %x)
@@ -405,26 +315,14 @@
 ; CHECK-LABEL: log_v2f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    subq $40, %rsp
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    callq logf
 ; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    callq logf
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
 ; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
 ; CHECK-NEXT:    callq logf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
 ; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq logf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilps $231, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,1,2,3]
-; CHECK-NEXT:    callq logf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
 ; CHECK-NEXT:    addq $40, %rsp
 ; CHECK-NEXT:    retq
   %r = call <2 x float> @llvm.log.v2f32(<2 x float> %x)
@@ -435,26 +333,14 @@
 ; CHECK-LABEL: log10_v2f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    subq $40, %rsp
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    callq log10f
 ; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    callq log10f
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
 ; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
 ; CHECK-NEXT:    callq log10f
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
 ; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq log10f
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilps $231, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,1,2,3]
-; CHECK-NEXT:    callq log10f
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
 ; CHECK-NEXT:    addq $40, %rsp
 ; CHECK-NEXT:    retq
   %r = call <2 x float> @llvm.log10.v2f32(<2 x float> %x)
@@ -465,26 +351,14 @@
 ; CHECK-LABEL: log2_v2f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    subq $40, %rsp
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    callq log2f
 ; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    callq log2f
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
 ; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
 ; CHECK-NEXT:    callq log2f
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
 ; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq log2f
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilps $231, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,1,2,3]
-; CHECK-NEXT:    callq log2f
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
 ; CHECK-NEXT:    addq $40, %rsp
 ; CHECK-NEXT:    retq
   %r = call <2 x float> @llvm.log2.v2f32(<2 x float> %x)
@@ -513,26 +387,14 @@
 ; CHECK-LABEL: round_v2f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    subq $40, %rsp
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    callq roundf
 ; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    callq roundf
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
 ; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
 ; CHECK-NEXT:    callq roundf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
 ; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq roundf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vpermilps $231, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,1,2,3]
-; CHECK-NEXT:    callq roundf
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
 ; CHECK-NEXT:    addq $40, %rsp
 ; CHECK-NEXT:    retq
   %r = call <2 x float> @llvm.round.v2f32(<2 x float> %x)