RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-23 11:13:28 +01:00
Code Issues Projects Releases Wiki Activity

[X86] Remove an autoupgrade legacy cvtss2sd intrinsics.

Browse Source 
llvm-svn: 332187
This commit is contained in:
Craig Topper 2018-05-13 00:29:40 +00:00
parent 5b8c2736c0
commit 7a07867ed7
7 changed files with 67 additions and 102 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
include/llvm/IR/IntrinsicsX86.td
View File

@ -488,9 +488,6 @@ let TargetPrefix = "x86" in { // All intrinsics start with "llvm.x86.".
def int_x86_sse2_cvtsd2ss : GCCBuiltin<"__builtin_ia32_cvtsd2ss">,
Intrinsic<[llvm_v4f32_ty], [llvm_v4f32_ty,
llvm_v2f64_ty], [IntrNoMem]>;
def int_x86_sse2_cvtss2sd : // TODO: Remove this intrinsic.
Intrinsic<[llvm_v2f64_ty], [llvm_v2f64_ty,
llvm_v4f32_ty], [IntrNoMem]>;
def int_x86_sse_cvtpd2pi : GCCBuiltin<"__builtin_ia32_cvtpd2pi">,
Intrinsic<[llvm_x86mmx_ty], [llvm_v2f64_ty], [IntrNoMem]>;
def int_x86_sse_cvttpd2pi: GCCBuiltin<"__builtin_ia32_cvttpd2pi">,

5
lib/IR/AutoUpgrade.cpp
View File

@ -258,6 +258,7 @@ static bool ShouldUpgradeX86Intrinsic(Function *F, StringRef Name) {
Name == "sse.cvtsi642ss" || // Added in 7.0
Name == "sse2.cvtsi2sd" || // Added in 7.0
Name == "sse2.cvtsi642sd" || // Added in 7.0
Name == "sse2.cvtss2sd" || // Added in 7.0
Name == "sse2.cvtdq2pd" || // Added in 3.9
Name == "sse2.cvtps2pd" || // Added in 3.9
Name == "avx.cvtdq2.pd.256" || // Added in 3.9
@ -1559,6 +1560,10 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) {
Rep = Builder.CreateSIToFP(CI->getArgOperand(1),
CI->getType()->getVectorElementType());
Rep = Builder.CreateInsertElement(CI->getArgOperand(0), Rep, (uint64_t)0);
} else if (IsX86 && Name == "sse2.cvtss2sd") {
Rep = Builder.CreateExtractElement(CI->getArgOperand(1), (uint64_t)0);
Rep = Builder.CreateFPExt(Rep, CI->getType()->getVectorElementType());
Rep = Builder.CreateInsertElement(CI->getArgOperand(0), Rep, (uint64_t)0);
} else if (IsX86 && (Name == "sse2.cvtdq2pd" ||
Name == "sse2.cvtps2pd" ||
Name == "avx.cvtdq2.pd.256" ||

20
lib/Target/X86/X86InstrSSE.td
View File

@ -1360,35 +1360,29 @@ def : Pat<(fpextend (loadf32 addr:$src)),
def : Pat<(extloadf32 addr:$src),
(CVTSS2SDrr (MOVSSrm addr:$src))>, Requires<[UseSSE2, OptForSpeed]>;
let isCodeGenOnly = 1 in {
let isCodeGenOnly = 1, hasSideEffects = 0 in {
def VCVTSS2SDrr_Int: I<0x5A, MRMSrcReg,
(outs VR128:$dst), (ins VR128:$src1, VR128:$src2),
"vcvtss2sd\t{$src2, $src1, $dst|$dst, $src1, $src2}",
[(set VR128:$dst,
(int_x86_sse2_cvtss2sd VR128:$src1, VR128:$src2))]>,
XS, VEX_4V, VEX_WIG,
[]>, XS, VEX_4V, VEX_WIG,
Requires<[HasAVX]>, Sched<[WriteCvtF2F]>;
let mayLoad = 1 in
def VCVTSS2SDrm_Int: I<0x5A, MRMSrcMem,
(outs VR128:$dst), (ins VR128:$src1, ssmem:$src2),
"vcvtss2sd\t{$src2, $src1, $dst|$dst, $src1, $src2}",
[(set VR128:$dst,
(int_x86_sse2_cvtss2sd VR128:$src1, sse_load_f32:$src2))]>,
XS, VEX_4V, VEX_WIG,
[]>, XS, VEX_4V, VEX_WIG,
Requires<[HasAVX]>, Sched<[WriteCvtF2FLd, ReadAfterLd]>;
let Constraints = "$src1 = $dst" in { // SSE2 instructions with XS prefix
def CVTSS2SDrr_Int: I<0x5A, MRMSrcReg,
(outs VR128:$dst), (ins VR128:$src1, VR128:$src2),
"cvtss2sd\t{$src2, $dst|$dst, $src2}",
[(set VR128:$dst,
(int_x86_sse2_cvtss2sd VR128:$src1, VR128:$src2))]>,
XS, Requires<[UseSSE2]>,
[]>, XS, Requires<[UseSSE2]>,
Sched<[WriteCvtF2F]>;
let mayLoad = 1 in
def CVTSS2SDrm_Int: I<0x5A, MRMSrcMem,
(outs VR128:$dst), (ins VR128:$src1, ssmem:$src2),
"cvtss2sd\t{$src2, $dst|$dst, $src2}",
[(set VR128:$dst,
(int_x86_sse2_cvtss2sd VR128:$src1, sse_load_f32:$src2))]>,
XS, Requires<[UseSSE2]>,
[]>, XS, Requires<[UseSSE2]>,
Sched<[WriteCvtF2FLd, ReadAfterLd]>;
}
} // isCodeGenOnly = 1

1
lib/Transforms/Instrumentation/MemorySanitizer.cpp
View File

@ -2563,7 +2563,6 @@ struct MemorySanitizerVisitor : public InstVisitor<MemorySanitizerVisitor> {
case Intrinsic::x86_sse2_cvtsd2si64:
case Intrinsic::x86_sse2_cvtsd2si:
case Intrinsic::x86_sse2_cvtsd2ss:
case Intrinsic::x86_sse2_cvtss2sd:
case Intrinsic::x86_sse2_cvttsd2si64:
case Intrinsic::x86_sse2_cvttsd2si:
case Intrinsic::x86_sse_cvtss2si64:

55
test/CodeGen/X86/sse2-intrinsics-x86-upgrade.ll
View File

@ -259,23 +259,6 @@ declare <2 x i64> @llvm.x86.sse2.pmulu.dq(<4 x i32>, <4 x i32>) nounwind readnon
define <2 x double> @test_x86_sse2_cvtsi2sd(<2 x double> %a0, i32 %a1) {
; SSE-LABEL: test_x86_sse2_cvtsi2sd:
; SSE: ## %bb.0:
; SSE-NEXT: movl {{[0-9]+}}(%esp), %eax ## encoding: [0x8b,0x44,0x24,0x04]
; SSE-NEXT: cvtsi2sdl %eax, %xmm0 ## encoding: [0xf2,0x0f,0x2a,0xc0]
; SSE-NEXT: retl ## encoding: [0xc3]
;
; AVX2-LABEL: test_x86_sse2_cvtsi2sd:
; AVX2: ## %bb.0:
; AVX2-NEXT: movl {{[0-9]+}}(%esp), %eax ## encoding: [0x8b,0x44,0x24,0x04]
; AVX2-NEXT: vcvtsi2sdl %eax, %xmm0, %xmm0 ## encoding: [0xc5,0xfb,0x2a,0xc0]
; AVX2-NEXT: retl ## encoding: [0xc3]
;
; SKX-LABEL: test_x86_sse2_cvtsi2sd:
; SKX: ## %bb.0:
; SKX-NEXT: movl {{[0-9]+}}(%esp), %eax ## encoding: [0x8b,0x44,0x24,0x04]
; SKX-NEXT: vcvtsi2sdl %eax, %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xfb,0x2a,0xc0]
; SKX-NEXT: retl ## encoding: [0xc3]
; CHECK-LABEL: test_x86_sse2_cvtsi2sd:
; CHECK: ## %bb.0:
; CHECK-NEXT: cvtsi2sdl {{[0-9]+}}(%esp), %xmm0
@ -284,3 +267,41 @@ define <2 x double> @test_x86_sse2_cvtsi2sd(<2 x double> %a0, i32 %a1) {
ret <2 x double> %res
}
declare <2 x double> @llvm.x86.sse2.cvtsi2sd(<2 x double>, i32) nounwind readnone
define <2 x double> @test_x86_sse2_cvtss2sd(<2 x double> %a0, <4 x float> %a1) {
; CHECK-LABEL: test_x86_sse2_cvtss2sd:
; CHECK: ## %bb.0:
; CHECK-NEXT: cvtss2sd %xmm1, %xmm0
; CHECK-NEXT: retl
%res = call <2 x double> @llvm.x86.sse2.cvtss2sd(<2 x double> %a0, <4 x float> %a1) ; <<2 x double>> [#uses=1]
ret <2 x double> %res
}
declare <2 x double> @llvm.x86.sse2.cvtss2sd(<2 x double>, <4 x float>) nounwind readnone
define <2 x double> @test_x86_sse2_cvtss2sd_load(<2 x double> %a0, <4 x float>* %p1) {
; CHECK-LABEL: test_x86_sse2_cvtss2sd_load:
; CHECK: ## %bb.0:
; CHECK-NEXT: movl {{[0-9]+}}(%esp), %eax
; CHECK-NEXT: movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
; CHECK-NEXT: cvtss2sd %xmm1, %xmm1
; CHECK-NEXT: movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; CHECK-NEXT: retl
%a1 = load <4 x float>, <4 x float>* %p1
%res = call <2 x double> @llvm.x86.sse2.cvtss2sd(<2 x double> %a0, <4 x float> %a1) ; <<2 x double>> [#uses=1]
ret <2 x double> %res
}
define <2 x double> @test_x86_sse2_cvtss2sd_load_optsize(<2 x double> %a0, <4 x float>* %p1) optsize {
; CHECK-LABEL: test_x86_sse2_cvtss2sd_load_optsize:
; CHECK: ## %bb.0:
; CHECK-NEXT: movl {{[0-9]+}}(%esp), %eax
; CHECK-NEXT: cvtss2sd (%eax), %xmm1
; CHECK-NEXT: movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; CHECK-NEXT: retl
%a1 = load <4 x float>, <4 x float>* %p1
%res = call <2 x double> @llvm.x86.sse2.cvtss2sd(<2 x double> %a0, <4 x float> %a1) ; <<2 x double>> [#uses=1]
ret <2 x double> %res
}

76
test/CodeGen/X86/sse2-intrinsics-x86.ll
View File

@ -457,58 +457,6 @@ define <4 x float> @test_x86_sse2_cvtsd2ss_load_optsize(<4 x float> %a0, <2 x do
}
define <2 x double> @test_x86_sse2_cvtss2sd(<2 x double> %a0, <4 x float> %a1) {
; SSE-LABEL: test_x86_sse2_cvtss2sd:
; SSE: ## %bb.0:
; SSE-NEXT: cvtss2sd %xmm1, %xmm0 ## encoding: [0xf3,0x0f,0x5a,0xc1]
; SSE-NEXT: retl ## encoding: [0xc3]
;
; VCHECK-LABEL: test_x86_sse2_cvtss2sd:
; VCHECK: ## %bb.0:
; VCHECK-NEXT: vcvtss2sd %xmm1, %xmm0, %xmm0 ## encoding: [0xc5,0xfa,0x5a,0xc1]
; VCHECK-NEXT: retl ## encoding: [0xc3]
%res = call <2 x double> @llvm.x86.sse2.cvtss2sd(<2 x double> %a0, <4 x float> %a1) ; <<2 x double>> [#uses=1]
ret <2 x double> %res
}
declare <2 x double> @llvm.x86.sse2.cvtss2sd(<2 x double>, <4 x float>) nounwind readnone
define <2 x double> @test_x86_sse2_cvtss2sd_load(<2 x double> %a0, <4 x float>* %p1) {
; SSE-LABEL: test_x86_sse2_cvtss2sd_load:
; SSE: ## %bb.0:
; SSE-NEXT: movl {{[0-9]+}}(%esp), %eax ## encoding: [0x8b,0x44,0x24,0x04]
; SSE-NEXT: cvtss2sd (%eax), %xmm0 ## encoding: [0xf3,0x0f,0x5a,0x00]
; SSE-NEXT: retl ## encoding: [0xc3]
;
; VCHECK-LABEL: test_x86_sse2_cvtss2sd_load:
; VCHECK: ## %bb.0:
; VCHECK-NEXT: movl {{[0-9]+}}(%esp), %eax ## encoding: [0x8b,0x44,0x24,0x04]
; VCHECK-NEXT: vcvtss2sd (%eax), %xmm0, %xmm0 ## encoding: [0xc5,0xfa,0x5a,0x00]
; VCHECK-NEXT: retl ## encoding: [0xc3]
%a1 = load <4 x float>, <4 x float>* %p1
%res = call <2 x double> @llvm.x86.sse2.cvtss2sd(<2 x double> %a0, <4 x float> %a1) ; <<2 x double>> [#uses=1]
ret <2 x double> %res
}
define <2 x double> @test_x86_sse2_cvtss2sd_load_optsize(<2 x double> %a0, <4 x float>* %p1) optsize {
; SSE-LABEL: test_x86_sse2_cvtss2sd_load_optsize:
; SSE: ## %bb.0:
; SSE-NEXT: movl {{[0-9]+}}(%esp), %eax ## encoding: [0x8b,0x44,0x24,0x04]
; SSE-NEXT: cvtss2sd (%eax), %xmm0 ## encoding: [0xf3,0x0f,0x5a,0x00]
; SSE-NEXT: retl ## encoding: [0xc3]
;
; VCHECK-LABEL: test_x86_sse2_cvtss2sd_load_optsize:
; VCHECK: ## %bb.0:
; VCHECK-NEXT: movl {{[0-9]+}}(%esp), %eax ## encoding: [0x8b,0x44,0x24,0x04]
; VCHECK-NEXT: vcvtss2sd (%eax), %xmm0, %xmm0 ## encoding: [0xc5,0xfa,0x5a,0x00]
; VCHECK-NEXT: retl ## encoding: [0xc3]
%a1 = load <4 x float>, <4 x float>* %p1
%res = call <2 x double> @llvm.x86.sse2.cvtss2sd(<2 x double> %a0, <4 x float> %a1) ; <<2 x double>> [#uses=1]
ret <2 x double> %res
}
define <4 x i32> @test_x86_sse2_cvttpd2dq(<2 x double> %a0) {
; SSE-LABEL: test_x86_sse2_cvttpd2dq:
; SSE: ## %bb.0:
@ -746,21 +694,21 @@ define <8 x i16> @test_x86_sse2_packssdw_128_fold() {
; SSE: ## %bb.0:
; SSE-NEXT: movaps {{.*#+}} xmm0 = [0,0,0,0,32767,32767,65535,32768]
; SSE-NEXT: ## encoding: [0x0f,0x28,0x05,A,A,A,A]
; SSE-NEXT: ## fixup A - offset: 3, value: LCPI34_0, kind: FK_Data_4
; SSE-NEXT: ## fixup A - offset: 3, value: LCPI31_0, kind: FK_Data_4
; SSE-NEXT: retl ## encoding: [0xc3]
;
; AVX2-LABEL: test_x86_sse2_packssdw_128_fold:
; AVX2: ## %bb.0:
; AVX2-NEXT: vmovaps {{.*#+}} xmm0 = [0,0,0,0,32767,32767,65535,32768]
; AVX2-NEXT: ## encoding: [0xc5,0xf8,0x28,0x05,A,A,A,A]
; AVX2-NEXT: ## fixup A - offset: 4, value: LCPI34_0, kind: FK_Data_4
; AVX2-NEXT: ## fixup A - offset: 4, value: LCPI31_0, kind: FK_Data_4
; AVX2-NEXT: retl ## encoding: [0xc3]
;
; SKX-LABEL: test_x86_sse2_packssdw_128_fold:
; SKX: ## %bb.0:
; SKX-NEXT: vmovaps LCPI34_0, %xmm0 ## EVEX TO VEX Compression xmm0 = [0,0,0,0,32767,32767,65535,32768]
; SKX-NEXT: vmovaps LCPI31_0, %xmm0 ## EVEX TO VEX Compression xmm0 = [0,0,0,0,32767,32767,65535,32768]
; SKX-NEXT: ## encoding: [0xc5,0xf8,0x28,0x05,A,A,A,A]
; SKX-NEXT: ## fixup A - offset: 4, value: LCPI34_0, kind: FK_Data_4
; SKX-NEXT: ## fixup A - offset: 4, value: LCPI31_0, kind: FK_Data_4
; SKX-NEXT: retl ## encoding: [0xc3]
%res = call <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32> zeroinitializer, <4 x i32> <i32 65535, i32 65536, i32 -1, i32 -131072>)
ret <8 x i16> %res
@ -793,21 +741,21 @@ define <16 x i8> @test_x86_sse2_packsswb_128_fold() {
; SSE: ## %bb.0:
; SSE-NEXT: movaps {{.*#+}} xmm0 = [0,127,127,255,255,128,128,128,0,0,0,0,0,0,0,0]
; SSE-NEXT: ## encoding: [0x0f,0x28,0x05,A,A,A,A]
; SSE-NEXT: ## fixup A - offset: 3, value: LCPI36_0, kind: FK_Data_4
; SSE-NEXT: ## fixup A - offset: 3, value: LCPI33_0, kind: FK_Data_4
; SSE-NEXT: retl ## encoding: [0xc3]
;
; AVX2-LABEL: test_x86_sse2_packsswb_128_fold:
; AVX2: ## %bb.0:
; AVX2-NEXT: vmovaps {{.*#+}} xmm0 = [0,127,127,255,255,128,128,128,0,0,0,0,0,0,0,0]
; AVX2-NEXT: ## encoding: [0xc5,0xf8,0x28,0x05,A,A,A,A]
; AVX2-NEXT: ## fixup A - offset: 4, value: LCPI36_0, kind: FK_Data_4
; AVX2-NEXT: ## fixup A - offset: 4, value: LCPI33_0, kind: FK_Data_4
; AVX2-NEXT: retl ## encoding: [0xc3]
;
; SKX-LABEL: test_x86_sse2_packsswb_128_fold:
; SKX: ## %bb.0:
; SKX-NEXT: vmovaps LCPI36_0, %xmm0 ## EVEX TO VEX Compression xmm0 = [0,127,127,255,255,128,128,128,0,0,0,0,0,0,0,0]
; SKX-NEXT: vmovaps LCPI33_0, %xmm0 ## EVEX TO VEX Compression xmm0 = [0,127,127,255,255,128,128,128,0,0,0,0,0,0,0,0]
; SKX-NEXT: ## encoding: [0xc5,0xf8,0x28,0x05,A,A,A,A]
; SKX-NEXT: ## fixup A - offset: 4, value: LCPI36_0, kind: FK_Data_4
; SKX-NEXT: ## fixup A - offset: 4, value: LCPI33_0, kind: FK_Data_4
; SKX-NEXT: retl ## encoding: [0xc3]
%res = call <16 x i8> @llvm.x86.sse2.packsswb.128(<8 x i16> <i16 0, i16 255, i16 256, i16 65535, i16 -1, i16 -255, i16 -256, i16 -32678>, <8 x i16> zeroinitializer)
ret <16 x i8> %res
@ -840,21 +788,21 @@ define <16 x i8> @test_x86_sse2_packuswb_128_fold() {
; SSE: ## %bb.0:
; SSE-NEXT: movaps {{.*#+}} xmm0 = [0,255,255,0,0,0,0,0,0,0,0,0,0,0,0,0]
; SSE-NEXT: ## encoding: [0x0f,0x28,0x05,A,A,A,A]
; SSE-NEXT: ## fixup A - offset: 3, value: LCPI38_0, kind: FK_Data_4
; SSE-NEXT: ## fixup A - offset: 3, value: LCPI35_0, kind: FK_Data_4
; SSE-NEXT: retl ## encoding: [0xc3]
;
; AVX2-LABEL: test_x86_sse2_packuswb_128_fold:
; AVX2: ## %bb.0:
; AVX2-NEXT: vmovaps {{.*#+}} xmm0 = [0,255,255,0,0,0,0,0,0,0,0,0,0,0,0,0]
; AVX2-NEXT: ## encoding: [0xc5,0xf8,0x28,0x05,A,A,A,A]
; AVX2-NEXT: ## fixup A - offset: 4, value: LCPI38_0, kind: FK_Data_4
; AVX2-NEXT: ## fixup A - offset: 4, value: LCPI35_0, kind: FK_Data_4
; AVX2-NEXT: retl ## encoding: [0xc3]
;
; SKX-LABEL: test_x86_sse2_packuswb_128_fold:
; SKX: ## %bb.0:
; SKX-NEXT: vmovaps LCPI38_0, %xmm0 ## EVEX TO VEX Compression xmm0 = [0,255,255,0,0,0,0,0,0,0,0,0,0,0,0,0]
; SKX-NEXT: vmovaps LCPI35_0, %xmm0 ## EVEX TO VEX Compression xmm0 = [0,255,255,0,0,0,0,0,0,0,0,0,0,0,0,0]
; SKX-NEXT: ## encoding: [0xc5,0xf8,0x28,0x05,A,A,A,A]
; SKX-NEXT: ## fixup A - offset: 4, value: LCPI38_0, kind: FK_Data_4
; SKX-NEXT: ## fixup A - offset: 4, value: LCPI35_0, kind: FK_Data_4
; SKX-NEXT: retl ## encoding: [0xc3]
%res = call <16 x i8> @llvm.x86.sse2.packuswb.128(<8 x i16> <i16 0, i16 255, i16 256, i16 65535, i16 -1, i16 -255, i16 -256, i16 -32678>, <8 x i16> zeroinitializer)
ret <16 x i8> %res

9
test/CodeGen/X86/sse_partial_update.ll
View File

@ -98,8 +98,9 @@ declare <2 x double> @llvm.x86.sse2.sqrt.sd(<2 x double>) nounwind readnone
define <2 x double> @load_fold_cvtss2sd_int(<4 x float> *%a) {
; CHECK-LABEL: load_fold_cvtss2sd_int:
; CHECK: ## %bb.0:
; CHECK-NEXT: xorps %xmm0, %xmm0
; CHECK-NEXT: cvtss2sd (%rdi), %xmm0
; CHECK-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
; CHECK-NEXT: cvtss2sd %xmm0, %xmm0
; CHECK-NEXT: movq {{.*#+}} xmm0 = xmm0[0],zero
; CHECK-NEXT: retq
%ld = load <4 x float>, <4 x float> *%a
%x = call <2 x double> @llvm.x86.sse2.cvtss2sd(<2 x double> <double 0x0, double 0x0>, <4 x float> %ld)
@ -109,8 +110,8 @@ define <2 x double> @load_fold_cvtss2sd_int(<4 x float> *%a) {
define <2 x double> @load_fold_cvtss2sd_int_optsize(<4 x float> *%a) optsize {
; CHECK-LABEL: load_fold_cvtss2sd_int_optsize:
; CHECK: ## %bb.0:
; CHECK-NEXT: xorps %xmm0, %xmm0
; CHECK-NEXT: cvtss2sd (%rdi), %xmm0
; CHECK-NEXT: movq {{.*#+}} xmm0 = xmm0[0],zero
; CHECK-NEXT: retq
%ld = load <4 x float>, <4 x float> *%a
%x = call <2 x double> @llvm.x86.sse2.cvtss2sd(<2 x double> <double 0x0, double 0x0>, <4 x float> %ld)
@ -120,8 +121,8 @@ define <2 x double> @load_fold_cvtss2sd_int_optsize(<4 x float> *%a) optsize {
define <2 x double> @load_fold_cvtss2sd_int_minsize(<4 x float> *%a) minsize {
; CHECK-LABEL: load_fold_cvtss2sd_int_minsize:
; CHECK: ## %bb.0:
; CHECK-NEXT: xorps %xmm0, %xmm0
; CHECK-NEXT: cvtss2sd (%rdi), %xmm0
; CHECK-NEXT: movq {{.*#+}} xmm0 = xmm0[0],zero
; CHECK-NEXT: retq
%ld = load <4 x float>, <4 x float> *%a
%x = call <2 x double> @llvm.x86.sse2.cvtss2sd(<2 x double> <double 0x0, double 0x0>, <4 x float> %ld)