Index: llvm/trunk/lib/Target/X86/X86InstrInfo.cpp =================================================================== --- llvm/trunk/lib/Target/X86/X86InstrInfo.cpp +++ llvm/trunk/lib/Target/X86/X86InstrInfo.cpp @@ -333,6 +333,8 @@ { X86::MUL32r, X86::MUL32m, TB_FOLDED_LOAD }, { X86::MUL64r, X86::MUL64m, TB_FOLDED_LOAD }, { X86::MUL8r, X86::MUL8m, TB_FOLDED_LOAD }, + { X86::PEXTRDrr, X86::PEXTRDmr, TB_FOLDED_STORE }, + { X86::PEXTRQrr, X86::PEXTRQmr, TB_FOLDED_STORE }, { X86::SETAEr, X86::SETAEm, TB_FOLDED_STORE }, { X86::SETAr, X86::SETAm, TB_FOLDED_STORE }, { X86::SETBEr, X86::SETBEm, TB_FOLDED_STORE }, @@ -367,6 +369,8 @@ { X86::VMOVSS2DIrr, X86::VMOVSS2DImr, TB_FOLDED_STORE }, { X86::VMOVUPDrr, X86::VMOVUPDmr, TB_FOLDED_STORE }, { X86::VMOVUPSrr, X86::VMOVUPSmr, TB_FOLDED_STORE }, + { X86::VPEXTRDrr, X86::VPEXTRDmr, TB_FOLDED_STORE }, + { X86::VPEXTRQrr, X86::VPEXTRQmr, TB_FOLDED_STORE }, // AVX 256-bit foldable instructions { X86::VEXTRACTI128rr, X86::VEXTRACTI128mr, TB_FOLDED_STORE | TB_ALIGN_16 }, { X86::VMOVAPDYrr, X86::VMOVAPDYmr, TB_FOLDED_STORE | TB_ALIGN_32 }, @@ -493,9 +497,27 @@ { X86::PABSBrr128, X86::PABSBrm128, TB_ALIGN_16 }, { X86::PABSDrr128, X86::PABSDrm128, TB_ALIGN_16 }, { X86::PABSWrr128, X86::PABSWrm128, TB_ALIGN_16 }, + { X86::PCMPESTRIrr, X86::PCMPESTRIrm, TB_ALIGN_16 }, + { X86::PCMPESTRM128rr, X86::PCMPESTRM128rm, TB_ALIGN_16 }, + { X86::PCMPISTRIrr, X86::PCMPISTRIrm, TB_ALIGN_16 }, + { X86::PCMPISTRM128rr, X86::PCMPISTRM128rm, TB_ALIGN_16 }, + { X86::PHMINPOSUWrr128, X86::PHMINPOSUWrm128, TB_ALIGN_16 }, + { X86::PMOVSXBDrr, X86::PMOVSXBDrm, TB_ALIGN_16 }, + { X86::PMOVSXBQrr, X86::PMOVSXBQrm, TB_ALIGN_16 }, + { X86::PMOVSXBWrr, X86::PMOVSXBWrm, TB_ALIGN_16 }, + { X86::PMOVSXDQrr, X86::PMOVSXDQrm, TB_ALIGN_16 }, + { X86::PMOVSXWDrr, X86::PMOVSXWDrm, TB_ALIGN_16 }, + { X86::PMOVSXWQrr, X86::PMOVSXWQrm, TB_ALIGN_16 }, + { X86::PMOVZXBDrr, X86::PMOVZXBDrm, TB_ALIGN_16 }, + { X86::PMOVZXBQrr, X86::PMOVZXBQrm, TB_ALIGN_16 }, + { X86::PMOVZXBWrr, X86::PMOVZXBWrm, TB_ALIGN_16 }, + { X86::PMOVZXDQrr, X86::PMOVZXDQrm, TB_ALIGN_16 }, + { X86::PMOVZXWDrr, X86::PMOVZXWDrm, TB_ALIGN_16 }, + { X86::PMOVZXWQrr, X86::PMOVZXWQrm, TB_ALIGN_16 }, { X86::PSHUFDri, X86::PSHUFDmi, TB_ALIGN_16 }, { X86::PSHUFHWri, X86::PSHUFHWmi, TB_ALIGN_16 }, { X86::PSHUFLWri, X86::PSHUFLWmi, TB_ALIGN_16 }, + { X86::PTESTrr, X86::PTESTrm, TB_ALIGN_16 }, { X86::RCPPSr, X86::RCPPSm, TB_ALIGN_16 }, { X86::RCPPSr_Int, X86::RCPPSm_Int, TB_ALIGN_16 }, { X86::ROUNDPDr, X86::ROUNDPDm, TB_ALIGN_16 }, @@ -559,11 +581,29 @@ { X86::VPABSBrr128, X86::VPABSBrm128, 0 }, { X86::VPABSDrr128, X86::VPABSDrm128, 0 }, { X86::VPABSWrr128, X86::VPABSWrm128, 0 }, + { X86::VPCMPESTRIrr, X86::VPCMPESTRIrm, 0 }, + { X86::VPCMPESTRM128rr, X86::VPCMPESTRM128rm, 0 }, + { X86::VPCMPISTRIrr, X86::VPCMPISTRIrm, 0 }, + { X86::VPCMPISTRM128rr, X86::VPCMPISTRM128rm, 0 }, + { X86::VPHMINPOSUWrr128, X86::VPHMINPOSUWrm128, 0 }, { X86::VPERMILPDri, X86::VPERMILPDmi, 0 }, { X86::VPERMILPSri, X86::VPERMILPSmi, 0 }, + { X86::VPMOVSXBDrr, X86::VPMOVSXBDrm, 0 }, + { X86::VPMOVSXBQrr, X86::VPMOVSXBQrm, 0 }, + { X86::VPMOVSXBWrr, X86::VPMOVSXBWrm, 0 }, + { X86::VPMOVSXDQrr, X86::VPMOVSXDQrm, 0 }, + { X86::VPMOVSXWDrr, X86::VPMOVSXWDrm, 0 }, + { X86::VPMOVSXWQrr, X86::VPMOVSXWQrm, 0 }, + { X86::VPMOVZXBDrr, X86::VPMOVZXBDrm, 0 }, + { X86::VPMOVZXBQrr, X86::VPMOVZXBQrm, 0 }, + { X86::VPMOVZXBWrr, X86::VPMOVZXBWrm, 0 }, + { X86::VPMOVZXDQrr, X86::VPMOVZXDQrm, 0 }, + { X86::VPMOVZXWDrr, X86::VPMOVZXWDrm, 0 }, + { X86::VPMOVZXWQrr, X86::VPMOVZXWQrm, 0 }, { X86::VPSHUFDri, X86::VPSHUFDmi, 0 }, { X86::VPSHUFHWri, X86::VPSHUFHWmi, 0 }, { X86::VPSHUFLWri, X86::VPSHUFLWmi, 0 }, + { X86::VPTESTrr, X86::VPTESTrm, 0 }, { X86::VRCPPSr, X86::VRCPPSm, 0 }, { X86::VRCPPSr_Int, X86::VRCPPSm_Int, 0 }, { X86::VROUNDPDr, X86::VROUNDPDm, 0 }, @@ -881,7 +921,9 @@ { X86::PANDrr, X86::PANDrm, TB_ALIGN_16 }, { X86::PAVGBrr, X86::PAVGBrm, TB_ALIGN_16 }, { X86::PAVGWrr, X86::PAVGWrm, TB_ALIGN_16 }, + { X86::PBLENDVBrr0, X86::PBLENDVBrm0, TB_ALIGN_16 }, { X86::PBLENDWrri, X86::PBLENDWrmi, TB_ALIGN_16 }, + { X86::PCLMULQDQrr, X86::PCLMULQDQrm, TB_ALIGN_16 }, { X86::PCMPEQBrr, X86::PCMPEQBrm, TB_ALIGN_16 }, { X86::PCMPEQDrr, X86::PCMPEQDrm, TB_ALIGN_16 }, { X86::PCMPEQQrr, X86::PCMPEQQrm, TB_ALIGN_16 }, @@ -896,7 +938,10 @@ { X86::PHSUBDrr, X86::PHSUBDrm, TB_ALIGN_16 }, { X86::PHSUBSWrr128, X86::PHSUBSWrm128, TB_ALIGN_16 }, { X86::PHSUBWrr, X86::PHSUBWrm, TB_ALIGN_16 }, - { X86::PINSRWrri, X86::PINSRWrmi, TB_ALIGN_16 }, + { X86::PINSRBrr, X86::PINSRBrm, 0 }, + { X86::PINSRDrr, X86::PINSRDrm, 0 }, + { X86::PINSRQrr, X86::PINSRQrm, 0 }, + { X86::PINSRWrri, X86::PINSRWrmi, 0 }, { X86::PMADDUBSWrr128, X86::PMADDUBSWrm128, TB_ALIGN_16 }, { X86::PMADDWDrr, X86::PMADDWDrm, TB_ALIGN_16 }, { X86::PMAXSWrr, X86::PMAXSWrm, TB_ALIGN_16 }, @@ -934,8 +979,11 @@ { X86::PSRLWrr, X86::PSRLWrm, TB_ALIGN_16 }, { X86::PSUBBrr, X86::PSUBBrm, TB_ALIGN_16 }, { X86::PSUBDrr, X86::PSUBDrm, TB_ALIGN_16 }, + { X86::PSUBQrr, X86::PSUBQrm, TB_ALIGN_16 }, { X86::PSUBSBrr, X86::PSUBSBrm, TB_ALIGN_16 }, { X86::PSUBSWrr, X86::PSUBSWrm, TB_ALIGN_16 }, + { X86::PSUBUSBrr, X86::PSUBUSBrm, TB_ALIGN_16 }, + { X86::PSUBUSWrr, X86::PSUBUSWrm, TB_ALIGN_16 }, { X86::PSUBWrr, X86::PSUBWrm, TB_ALIGN_16 }, { X86::PUNPCKHBWrr, X86::PUNPCKHBWrm, TB_ALIGN_16 }, { X86::PUNPCKHDQrr, X86::PUNPCKHDQrm, TB_ALIGN_16 }, @@ -1068,7 +1116,9 @@ { X86::VPANDrr, X86::VPANDrm, 0 }, { X86::VPAVGBrr, X86::VPAVGBrm, 0 }, { X86::VPAVGWrr, X86::VPAVGWrm, 0 }, + { X86::VPBLENDVBrr, X86::VPBLENDVBrm, 0 }, { X86::VPBLENDWrri, X86::VPBLENDWrmi, 0 }, + { X86::VPCLMULQDQrr, X86::VPCLMULQDQrm, 0 }, { X86::VPCMPEQBrr, X86::VPCMPEQBrm, 0 }, { X86::VPCMPEQDrr, X86::VPCMPEQDrm, 0 }, { X86::VPCMPEQQrr, X86::VPCMPEQQrm, 0 }, @@ -1085,6 +1135,9 @@ { X86::VPHSUBWrr, X86::VPHSUBWrm, 0 }, { X86::VPERMILPDrr, X86::VPERMILPDrm, 0 }, { X86::VPERMILPSrr, X86::VPERMILPSrm, 0 }, + { X86::VPINSRBrr, X86::VPINSRBrm, 0 }, + { X86::VPINSRDrr, X86::VPINSRDrm, 0 }, + { X86::VPINSRQrr, X86::VPINSRQrm, 0 }, { X86::VPINSRWrri, X86::VPINSRWrmi, 0 }, { X86::VPMADDUBSWrr128, X86::VPMADDUBSWrm128, 0 }, { X86::VPMADDWDrr, X86::VPMADDWDrm, 0 }, @@ -1123,8 +1176,11 @@ { X86::VPSRLWrr, X86::VPSRLWrm, 0 }, { X86::VPSUBBrr, X86::VPSUBBrm, 0 }, { X86::VPSUBDrr, X86::VPSUBDrm, 0 }, + { X86::VPSUBQrr, X86::VPSUBQrm, 0 }, { X86::VPSUBSBrr, X86::VPSUBSBrm, 0 }, { X86::VPSUBSWrr, X86::VPSUBSWrm, 0 }, + { X86::VPSUBUSBrr, X86::VPSUBUSBrm, 0 }, + { X86::VPSUBUSWrr, X86::VPSUBUSWrm, 0 }, { X86::VPSUBWrr, X86::VPSUBWrm, 0 }, { X86::VPUNPCKHBWrr, X86::VPUNPCKHBWrm, 0 }, { X86::VPUNPCKHDQrr, X86::VPUNPCKHDQrm, 0 }, Index: llvm/trunk/test/CodeGen/X86/2011-11-30-or.ll =================================================================== --- llvm/trunk/test/CodeGen/X86/2011-11-30-or.ll +++ llvm/trunk/test/CodeGen/X86/2011-11-30-or.ll @@ -2,13 +2,13 @@ target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128-n8:16:32" target triple = "x86_64-apple-macosx10.6.6" - -; Test that the order of operands is correct -; CHECK: select_func -; CHECK: pblendvb %xmm1, %xmm2 -; CHECK: ret - -define void @select_func(<8 x i16> %in) { + +; Test that the order of operands is correct +; CHECK: select_func +; CHECK: pblendvb {{LCPI0_[0-9]*}}(%rip), %xmm1 +; CHECK: ret + +define void @select_func(<8 x i16> %in) { entry: %c.lobit.i.i.i = ashr <8 x i16> %in, %and.i56.i.i.i = and <8 x i16> %c.lobit.i.i.i, Index: llvm/trunk/test/CodeGen/X86/psubus.ll =================================================================== --- llvm/trunk/test/CodeGen/X86/psubus.ll +++ llvm/trunk/test/CodeGen/X86/psubus.ll @@ -23,19 +23,19 @@ br i1 %6, label %for.end, label %vector.body for.end: ; preds = %vector.body - ret void - -; SSE2: @test1 -; SSE2: psubusw LCPI0_0(%rip), %xmm0 - -; AVX1: @test1 -; AVX1: vpsubusw LCPI0_0(%rip), %xmm0, %xmm0 - -; AVX2: @test1 -; AVX2: vpsubusw LCPI0_0(%rip), %xmm0, %xmm0 -} - -define void @test2(i16* nocapture %head) nounwind { + ret void + +; SSE2: @test1 +; SSE2: psubusw %xmm0, %xmm1 + +; AVX1: @test1 +; AVX1: vpsubusw %xmm0, %xmm1, %xmm1 + +; AVX2: @test1 +; AVX2: vpsubusw %xmm0, %xmm1, %xmm1 +} + +define void @test2(i16* nocapture %head) nounwind { vector.ph: br label %vector.body @@ -53,19 +53,19 @@ br i1 %6, label %for.end, label %vector.body for.end: ; preds = %vector.body - ret void - -; SSE2: @test2 -; SSE2: psubusw LCPI1_0(%rip), %xmm0 - -; AVX1: @test2 -; AVX1: vpsubusw LCPI1_0(%rip), %xmm0, %xmm0 - -; AVX2: @test2 -; AVX2: vpsubusw LCPI1_0(%rip), %xmm0, %xmm0 -} - -define void @test3(i16* nocapture %head, i16 zeroext %w) nounwind { + ret void + +; SSE2: @test2 +; SSE2: psubusw %xmm0, %xmm1 + +; AVX1: @test2 +; AVX1: vpsubusw %xmm0, %xmm1, %xmm1 + +; AVX2: @test2 +; AVX2: vpsubusw %xmm0, %xmm1, %xmm1 +} + +define void @test3(i16* nocapture %head, i16 zeroext %w) nounwind { vector.ph: %0 = insertelement <8 x i16> undef, i16 %w, i32 0 %broadcast15 = shufflevector <8 x i16> %0, <8 x i16> undef, <8 x i32> zeroinitializer @@ -115,19 +115,19 @@ br i1 %6, label %for.end, label %vector.body for.end: ; preds = %vector.body - ret void - -; SSE2: @test4 -; SSE2: psubusb LCPI3_0(%rip), %xmm0 - -; AVX1: @test4 -; AVX1: vpsubusb LCPI3_0(%rip), %xmm0, %xmm0 - -; AVX2: @test4 -; AVX2: vpsubusb LCPI3_0(%rip), %xmm0, %xmm0 -} - -define void @test5(i8* nocapture %head) nounwind { + ret void + +; SSE2: @test4 +; SSE2: psubusb %xmm0, %xmm1 + +; AVX1: @test4 +; AVX1: vpsubusb %xmm0, %xmm1, %xmm1 + +; AVX2: @test4 +; AVX2: vpsubusb %xmm0, %xmm1, %xmm1 +} + +define void @test5(i8* nocapture %head) nounwind { vector.ph: br label %vector.body @@ -145,19 +145,19 @@ br i1 %6, label %for.end, label %vector.body for.end: ; preds = %vector.body - ret void - -; SSE2: @test5 -; SSE2: psubusb LCPI4_0(%rip), %xmm0 - -; AVX1: @test5 -; AVX1: vpsubusb LCPI4_0(%rip), %xmm0, %xmm0 - -; AVX2: @test5 -; AVX2: vpsubusb LCPI4_0(%rip), %xmm0, %xmm0 -} - -define void @test6(i8* nocapture %head, i8 zeroext %w) nounwind { + ret void + +; SSE2: @test5 +; SSE2: psubusb %xmm0, %xmm1 + +; AVX1: @test5 +; AVX1: vpsubusb %xmm0, %xmm1, %xmm1 + +; AVX2: @test5 +; AVX2: vpsubusb %xmm0, %xmm1, %xmm1 +} + +define void @test6(i8* nocapture %head, i8 zeroext %w) nounwind { vector.ph: %0 = insertelement <16 x i8> undef, i8 %w, i32 0 %broadcast15 = shufflevector <16 x i8> %0, <16 x i8> undef, <16 x i32> zeroinitializer Index: llvm/trunk/test/CodeGen/X86/stack-folding-int-avx1.ll =================================================================== --- llvm/trunk/test/CodeGen/X86/stack-folding-int-avx1.ll +++ llvm/trunk/test/CodeGen/X86/stack-folding-int-avx1.ll @@ -1,4 +1,4 @@ -; RUN: llc -O3 -disable-peephole -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=+avx,+aes < %s | FileCheck %s +; RUN: llc -O3 -disable-peephole -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=+avx,+aes,+pclmul < %s | FileCheck %s target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" target triple = "x86_64-unknown-unknown" @@ -62,6 +62,38 @@ } declare <2 x i64> @llvm.x86.aesni.aeskeygenassist(<2 x i64>, i8) nounwind readnone +define <4 x i32> @stack_fold_movd_load(i32 %a0) { + ;CHECK-LABEL: stack_fold_movd_load + ;CHECK: movd {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 4-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + %2 = insertelement <4 x i32> zeroinitializer, i32 %a0, i32 0 + ret <4 x i32> %2 +} + +define i32 @stack_fold_movd_store(<4 x i32> %a0) { + ;CHECK-LABEL: stack_fold_movd_store + ;CHECK: movd {{%xmm[0-9][0-9]*}}, {{-?[0-9]*}}(%rsp) {{.*#+}} 4-byte Folded Spill + %1 = extractelement <4 x i32> %a0, i32 0 + %2 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + ret i32 %1 +} + +define <2 x i64> @stack_fold_movq_load(<2 x i64> %a0) { + ;CHECK-LABEL: stack_fold_movq_load + ;CHECK: movq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = shufflevector <2 x i64> %a0, <2 x i64> zeroinitializer, <2 x i32> + ret <2 x i64> %2 +} + +define i64 @stack_fold_movq_store(<2 x i64> %a0) { + ;CHECK-LABEL: stack_fold_movq_store + ;CHECK: movq {{%xmm[0-9][0-9]*}}, {{-?[0-9]*}}(%rsp) {{.*#+}} 8-byte Folded Spill + %1 = extractelement <2 x i64> %a0, i32 0 + %2 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + ret i64 %1 +} + define <16 x i8> @stack_fold_pabsb(<16 x i8> %a0) { ;CHECK-LABEL: stack_fold_pabsb ;CHECK: vpabsb {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload @@ -240,7 +272,13 @@ } declare <8 x i16> @llvm.x86.sse2.pavg.w(<8 x i16>, <8 x i16>) nounwind readnone -; TODO stack_fold_pblendvb +define <16 x i8> @stack_fold_pblendvb(<16 x i8> %a0, <16 x i8> %a1, <16 x i8> %c) { + ;CHECK-LABEL: stack_fold_pblendvb + ;CHECK: vpblendvb {{%xmm[0-9][0-9]*}}, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a1, <16 x i8> %c, <16 x i8> %a0) + ret <16 x i8> %2 +} declare <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8>, <16 x i8>, <16 x i8>) nounwind readnone define <8 x i16> @stack_fold_pblendw(<8 x i16> %a0, <8 x i16> %a1) { @@ -252,7 +290,13 @@ } declare <8 x i16> @llvm.x86.sse41.pblendw(<8 x i16>, <8 x i16>, i8) nounwind readnone -; TODO stack_fold_pclmulqdq +define <2 x i64> @stack_fold_pclmulqdq(<2 x i64> %a0, <2 x i64> %a1) { + ;CHECK-LABEL: stack_fold_pclmulqdq + ;CHECK: vpclmulqdq $0, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <2 x i64> @llvm.x86.pclmulqdq(<2 x i64> %a0, <2 x i64> %a1, i8 0) + ret <2 x i64> %2 +} declare <2 x i64> @llvm.x86.pclmulqdq(<2 x i64>, <2 x i64>, i8) nounwind readnone define <16 x i8> @stack_fold_pcmpeqb(<16 x i8> %a0, <16 x i8> %a1) { @@ -291,10 +335,22 @@ ret <8 x i16> %3 } -; TODO stack_fold_pcmpestri +define i32 @stack_fold_pcmpestri(<16 x i8> %a0, <16 x i8> %a1) { + ;CHECK-LABEL: stack_fold_pcmpestri + ;CHECK: vpcmpestri $7, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{rax},~{flags}"() + %2 = call i32 @llvm.x86.sse42.pcmpestri128(<16 x i8> %a0, i32 7, <16 x i8> %a1, i32 7, i8 7) + ret i32 %2 +} declare i32 @llvm.x86.sse42.pcmpestri128(<16 x i8>, i32, <16 x i8>, i32, i8) nounwind readnone -; TODO stack_fold_pcmpestrm +define <16 x i8> @stack_fold_pcmpestrm(<16 x i8> %a0, <16 x i8> %a1) { + ;CHECK-LABEL: stack_fold_pcmpestrm + ;CHECK: vpcmpestrm $7, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{rax},~{flags}"() + %2 = call <16 x i8> @llvm.x86.sse42.pcmpestrm128(<16 x i8> %a0, i32 7, <16 x i8> %a1, i32 7, i8 7) + ret <16 x i8> %2 +} declare <16 x i8> @llvm.x86.sse42.pcmpestrm128(<16 x i8>, i32, <16 x i8>, i32, i8) nounwind readnone define <16 x i8> @stack_fold_pcmpgtb(<16 x i8> %a0, <16 x i8> %a1) { @@ -333,15 +389,44 @@ ret <8 x i16> %3 } -; TODO stack_fold_pcmpistri +define i32 @stack_fold_pcmpistri(<16 x i8> %a0, <16 x i8> %a1) { + ;CHECK-LABEL: stack_fold_pcmpistri + ;CHECK: vpcmpistri $7, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call i32 @llvm.x86.sse42.pcmpistri128(<16 x i8> %a0, <16 x i8> %a1, i8 7) + ret i32 %2 +} declare i32 @llvm.x86.sse42.pcmpistri128(<16 x i8>, <16 x i8>, i8) nounwind readnone -; TODO stack_fold_pcmpistrm +define <16 x i8> @stack_fold_pcmpistrm(<16 x i8> %a0, <16 x i8> %a1) { + ;CHECK-LABEL: stack_fold_pcmpistrm + ;CHECK: vpcmpistrm $7, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <16 x i8> @llvm.x86.sse42.pcmpistrm128(<16 x i8> %a0, <16 x i8> %a1, i8 7) + ret <16 x i8> %2 +} declare <16 x i8> @llvm.x86.sse42.pcmpistrm128(<16 x i8>, <16 x i8>, i8) nounwind readnone ; TODO stack_fold_pextrb -; TODO stack_fold_pextrd -; TODO stack_fold_pextrq + +define i32 @stack_fold_pextrd(<4 x i32> %a0) { + ;CHECK-LABEL: stack_fold_pextrd + ;CHECK: pextrd $1, {{%xmm[0-9][0-9]*}}, {{-?[0-9]*}}(%rsp) {{.*#+}} 4-byte Folded Spill + ;CHECK: movl {{-?[0-9]*}}(%rsp), %eax {{.*#+}} 4-byte Reload + %1 = extractelement <4 x i32> %a0, i32 1 + %2 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + ret i32 %1 +} + +define i64 @stack_fold_pextrq(<2 x i64> %a0) { + ;CHECK-LABEL: stack_fold_pextrq + ;CHECK: pextrq $1, {{%xmm[0-9][0-9]*}}, {{-?[0-9]*}}(%rsp) {{.*#+}} 8-byte Folded Spill + ;CHECK: movq {{-?[0-9]*}}(%rsp), %rax {{.*#+}} 8-byte Reload + %1 = extractelement <2 x i64> %a0, i32 1 + %2 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + ret i64 %1 +} + ; TODO stack_fold_pextrw define <4 x i32> @stack_fold_phaddd(<4 x i32> %a0, <4 x i32> %a1) { @@ -371,7 +456,13 @@ } declare <8 x i16> @llvm.x86.ssse3.phadd.w.128(<8 x i16>, <8 x i16>) nounwind readnone -; TODO stack_fold_phminposuw +define <8 x i16> @stack_fold_phminposuw(<8 x i16> %a0) { + ;CHECK-LABEL: stack_fold_phminposuw + ;CHECK: vphminposuw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <8 x i16> @llvm.x86.sse41.phminposuw(<8 x i16> %a0) + ret <8 x i16> %2 +} declare <8 x i16> @llvm.x86.sse41.phminposuw(<8 x i16>) nounwind readnone define <4 x i32> @stack_fold_phsubd(<4 x i32> %a0, <4 x i32> %a1) { @@ -401,10 +492,37 @@ } declare <8 x i16> @llvm.x86.ssse3.phsub.w.128(<8 x i16>, <8 x i16>) nounwind readnone -; TODO stack_fold_pinsrb -; TODO stack_fold_pinsrd -; TODO stack_fold_pinsrq -; TODO stack_fold_pinsrw +define <16 x i8> @stack_fold_pinsrb(<16 x i8> %a0, i8 %a1) { + ;CHECK-LABEL: stack_fold_pinsrb + ;CHECK: vpinsrb $1, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 4-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + %2 = insertelement <16 x i8> %a0, i8 %a1, i32 1 + ret <16 x i8> %2 +} + +define <4 x i32> @stack_fold_pinsrd(<4 x i32> %a0, i32 %a1) { + ;CHECK-LABEL: stack_fold_pinsrd + ;CHECK: vpinsrd $1, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 4-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + %2 = insertelement <4 x i32> %a0, i32 %a1, i32 1 + ret <4 x i32> %2 +} + +define <2 x i64> @stack_fold_pinsrq(<2 x i64> %a0, i64 %a1) { + ;CHECK-LABEL: stack_fold_pinsrq + ;CHECK: vpinsrq $1, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 8-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + %2 = insertelement <2 x i64> %a0, i64 %a1, i32 1 + ret <2 x i64> %2 +} + +define <8 x i16> @stack_fold_pinsrw(<8 x i16> %a0, i16 %a1) { + ;CHECK-LABEL: stack_fold_pinsrw + ;CHECK: vpinsrw $1, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 4-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + %2 = insertelement <8 x i16> %a0, i16 %a1, i32 1 + ret <8 x i16> %2 +} define <8 x i16> @stack_fold_pmaddubsw(<16 x i8> %a0, <16 x i8> %a1) { ;CHECK-LABEL: stack_fold_pmaddubsw @@ -532,40 +650,112 @@ } declare <8 x i16> @llvm.x86.sse41.pminuw(<8 x i16>, <8 x i16>) nounwind readnone -; TODO stack_fold_pmovsxbd +define <4 x i32> @stack_fold_pmovsxbd(<16 x i8> %a0) { + ;CHECK-LABEL: stack_fold_pmovsxbd + ;CHECK: vpmovsxbd {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <4 x i32> @llvm.x86.sse41.pmovsxbd(<16 x i8> %a0) + ret <4 x i32> %2 +} declare <4 x i32> @llvm.x86.sse41.pmovsxbd(<16 x i8>) nounwind readnone -; TODO stack_fold_pmovsxbq +define <2 x i64> @stack_fold_pmovsxbq(<16 x i8> %a0) { + ;CHECK-LABEL: stack_fold_pmovsxbq + ;CHECK: pmovsxbq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <2 x i64> @llvm.x86.sse41.pmovsxbq(<16 x i8> %a0) + ret <2 x i64> %2 +} declare <2 x i64> @llvm.x86.sse41.pmovsxbq(<16 x i8>) nounwind readnone -; TODO stack_fold_pmovsxbw +define <8 x i16> @stack_fold_pmovsxbw(<16 x i8> %a0) { + ;CHECK-LABEL: stack_fold_pmovsxbw + ;CHECK: vpmovsxbw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <8 x i16> @llvm.x86.sse41.pmovsxbw(<16 x i8> %a0) + ret <8 x i16> %2 +} declare <8 x i16> @llvm.x86.sse41.pmovsxbw(<16 x i8>) nounwind readnone -; TODO stack_fold_pmovsxdq +define <2 x i64> @stack_fold_pmovsxdq(<4 x i32> %a0) { + ;CHECK-LABEL: stack_fold_pmovsxdq + ;CHECK: vpmovsxdq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <2 x i64> @llvm.x86.sse41.pmovsxdq(<4 x i32> %a0) + ret <2 x i64> %2 +} declare <2 x i64> @llvm.x86.sse41.pmovsxdq(<4 x i32>) nounwind readnone -; TODO stack_fold_pmovsxwd +define <4 x i32> @stack_fold_pmovsxwd(<8 x i16> %a0) { + ;CHECK-LABEL: stack_fold_pmovsxwd + ;CHECK: vpmovsxwd {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <4 x i32> @llvm.x86.sse41.pmovsxwd(<8 x i16> %a0) + ret <4 x i32> %2 +} declare <4 x i32> @llvm.x86.sse41.pmovsxwd(<8 x i16>) nounwind readnone -; TODO stack_fold_pmovsxwq +define <2 x i64> @stack_fold_pmovsxwq(<8 x i16> %a0) { + ;CHECK-LABEL: stack_fold_pmovsxwq + ;CHECK: vpmovsxwq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <2 x i64> @llvm.x86.sse41.pmovsxwq(<8 x i16> %a0) + ret <2 x i64> %2 +} declare <2 x i64> @llvm.x86.sse41.pmovsxwq(<8 x i16>) nounwind readnone -; TODO stack_fold_pmovzxbd +define <4 x i32> @stack_fold_pmovzxbd(<16 x i8> %a0) { + ;CHECK-LABEL: stack_fold_pmovzxbd + ;CHECK: vpmovzxbd {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <4 x i32> @llvm.x86.sse41.pmovzxbd(<16 x i8> %a0) + ret <4 x i32> %2 +} declare <4 x i32> @llvm.x86.sse41.pmovzxbd(<16 x i8>) nounwind readnone -; TODO stack_fold_pmovzxbq +define <2 x i64> @stack_fold_pmovzxbq(<16 x i8> %a0) { + ;CHECK-LABEL: stack_fold_pmovzxbq + ;CHECK: vpmovzxbq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <2 x i64> @llvm.x86.sse41.pmovzxbq(<16 x i8> %a0) + ret <2 x i64> %2 +} declare <2 x i64> @llvm.x86.sse41.pmovzxbq(<16 x i8>) nounwind readnone -; TODO stack_fold_pmovzxbw +define <8 x i16> @stack_fold_pmovzxbw(<16 x i8> %a0) { + ;CHECK-LABEL: stack_fold_pmovzxbw + ;CHECK: vpmovzxbw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <8 x i16> @llvm.x86.sse41.pmovzxbw(<16 x i8> %a0) + ret <8 x i16> %2 +} declare <8 x i16> @llvm.x86.sse41.pmovzxbw(<16 x i8>) nounwind readnone -; TODO stack_fold_pmovzxdq +define <2 x i64> @stack_fold_pmovzxdq(<4 x i32> %a0) { + ;CHECK-LABEL: stack_fold_pmovzxdq + ;CHECK: vpmovzxdq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <2 x i64> @llvm.x86.sse41.pmovzxdq(<4 x i32> %a0) + ret <2 x i64> %2 +} declare <2 x i64> @llvm.x86.sse41.pmovzxdq(<4 x i32>) nounwind readnone -; TODO stack_fold_pmovzxwd +define <4 x i32> @stack_fold_pmovzxwd(<8 x i16> %a0) { + ;CHECK-LABEL: stack_fold_pmovzxwd + ;CHECK: vpmovzxwd {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <4 x i32> @llvm.x86.sse41.pmovzxwd(<8 x i16> %a0) + ret <4 x i32> %2 +} declare <4 x i32> @llvm.x86.sse41.pmovzxwd(<8 x i16>) nounwind readnone -; TODO stack_fold_pmovzxwq +define <2 x i64> @stack_fold_pmovzxwq(<8 x i16> %a0) { + ;CHECK-LABEL: stack_fold_pmovzxwq + ;CHECK: vpmovzxwq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <2 x i64> @llvm.x86.sse41.pmovzxwq(<8 x i16> %a0) + ret <2 x i64> %2 +} declare <2 x i64> @llvm.x86.sse41.pmovzxwq(<8 x i16>) nounwind readnone define <2 x i64> @stack_fold_pmuldq(<4 x i32> %a0, <4 x i32> %a1) { @@ -796,7 +986,13 @@ ret <4 x i32> %2 } -; TODO stack_fold_psubq +define <2 x i64> @stack_fold_psubq(<2 x i64> %a0, <2 x i64> %a1) { + ;CHECK-LABEL: stack_fold_psubq + ;CHECK: vpsubq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = sub <2 x i64> %a0, %a1 + ret <2 x i64> %2 +} define <16 x i8> @stack_fold_psubsb(<16 x i8> %a0, <16 x i8> %a1) { ;CHECK-LABEL: stack_fold_psubsb @@ -816,10 +1012,22 @@ } declare <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16>, <8 x i16>) nounwind readnone -; TODO stack_fold_psubusb +define <16 x i8> @stack_fold_psubusb(<16 x i8> %a0, <16 x i8> %a1) { + ;CHECK-LABEL: stack_fold_psubusb + ;CHECK: vpsubusb {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <16 x i8> @llvm.x86.sse2.psubus.b(<16 x i8> %a0, <16 x i8> %a1) + ret <16 x i8> %2 +} declare <16 x i8> @llvm.x86.sse2.psubus.b(<16 x i8>, <16 x i8>) nounwind readnone -; TODO stack_fold_psubusw +define <8 x i16> @stack_fold_psubusw(<8 x i16> %a0, <8 x i16> %a1) { + ;CHECK-LABEL: stack_fold_psubusw + ;CHECK: vpsubusw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <8 x i16> @llvm.x86.sse2.psubus.w(<8 x i16> %a0, <8 x i16> %a1) + ret <8 x i16> %2 +} declare <8 x i16> @llvm.x86.sse2.psubus.w(<8 x i16>, <8 x i16>) nounwind readnone define <8 x i16> @stack_fold_psubw(<8 x i16> %a0, <8 x i16> %a1) { @@ -830,7 +1038,13 @@ ret <8 x i16> %2 } -; TODO stack_fold_ptest +define i32 @stack_fold_ptest(<2 x i64> %a0, <2 x i64> %a1) { + ;CHECK-LABEL: stack_fold_ptest + ;CHECK: vptest {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call i32 @llvm.x86.sse41.ptestc(<2 x i64> %a0, <2 x i64> %a1) + ret i32 %2 +} declare i32 @llvm.x86.sse41.ptestc(<2 x i64>, <2 x i64>) nounwind readnone define <16 x i8> @stack_fold_punpckhbw(<16 x i8> %a0, <16 x i8> %a1) { Index: llvm/trunk/test/CodeGen/X86/stack-folding-int-sse42.ll =================================================================== --- llvm/trunk/test/CodeGen/X86/stack-folding-int-sse42.ll +++ llvm/trunk/test/CodeGen/X86/stack-folding-int-sse42.ll @@ -1,7 +1,7 @@ -; RUN: llc -O3 -disable-peephole -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=+sse4.2,+aes < %s | FileCheck %s - -target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" -target triple = "x86_64-unknown-unknown" +; RUN: llc -O3 -disable-peephole -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=+sse4.2,+aes,+pclmul < %s | FileCheck %s + +target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" +target triple = "x86_64-unknown-unknown" ; Stack reload folding tests. ; @@ -59,12 +59,44 @@ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() %2 = call <2 x i64> @llvm.x86.aesni.aeskeygenassist(<2 x i64> %a0, i8 7) ret <2 x i64> %2 -} -declare <2 x i64> @llvm.x86.aesni.aeskeygenassist(<2 x i64>, i8) nounwind readnone - -define <16 x i8> @stack_fold_pabsb(<16 x i8> %a0) { - ;CHECK-LABEL: stack_fold_pabsb - ;CHECK: pabsb {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload +} +declare <2 x i64> @llvm.x86.aesni.aeskeygenassist(<2 x i64>, i8) nounwind readnone + +define <4 x i32> @stack_fold_movd_load(i32 %a0) { + ;CHECK-LABEL: stack_fold_movd_load + ;CHECK: movd {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 4-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + %2 = insertelement <4 x i32> zeroinitializer, i32 %a0, i32 0 + ret <4 x i32> %2 +} + +define i32 @stack_fold_movd_store(<4 x i32> %a0) { + ;CHECK-LABEL: stack_fold_movd_store + ;CHECK: movd {{%xmm[0-9][0-9]*}}, {{-?[0-9]*}}(%rsp) {{.*#+}} 4-byte Folded Spill + %1 = extractelement <4 x i32> %a0, i32 0 + %2 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + ret i32 %1 +} + +define <2 x i64> @stack_fold_movq_load(<2 x i64> %a0) { + ;CHECK-LABEL: stack_fold_movq_load + ;CHECK: movq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = shufflevector <2 x i64> %a0, <2 x i64> zeroinitializer, <2 x i32> + ret <2 x i64> %2 +} + +define i64 @stack_fold_movq_store(<2 x i64> %a0) { + ;CHECK-LABEL: stack_fold_movq_store + ;CHECK: movq {{%xmm[0-9][0-9]*}}, {{-?[0-9]*}}(%rsp) {{.*#+}} 8-byte Folded Spill + %1 = extractelement <2 x i64> %a0, i32 0 + %2 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + ret i64 %1 +} + +define <16 x i8> @stack_fold_pabsb(<16 x i8> %a0) { + ;CHECK-LABEL: stack_fold_pabsb + ;CHECK: pabsb {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() %2 = call <16 x i8> @llvm.x86.ssse3.pabs.b.128(<16 x i8> %a0) ret <16 x i8> %2 @@ -237,25 +269,37 @@ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() %2 = call <8 x i16> @llvm.x86.sse2.pavg.w(<8 x i16> %a0, <8 x i16> %a1) ret <8 x i16> %2 -} -declare <8 x i16> @llvm.x86.sse2.pavg.w(<8 x i16>, <8 x i16>) nounwind readnone - -; TODO stack_fold_pblendvb -declare <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8>, <16 x i8>, <16 x i8>) nounwind readnone - -define <8 x i16> @stack_fold_pblendw(<8 x i16> %a0, <8 x i16> %a1) { +} +declare <8 x i16> @llvm.x86.sse2.pavg.w(<8 x i16>, <8 x i16>) nounwind readnone + +define <16 x i8> @stack_fold_pblendvb(<16 x i8> %a0, <16 x i8> %a1, <16 x i8> %c) { + ;CHECK-LABEL: stack_fold_pblendvb + ;CHECK: pblendvb {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a1, <16 x i8> %c, <16 x i8> %a0) + ret <16 x i8> %2 +} +declare <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8>, <16 x i8>, <16 x i8>) nounwind readnone + +define <8 x i16> @stack_fold_pblendw(<8 x i16> %a0, <8 x i16> %a1) { ;CHECK-LABEL: stack_fold_pblendw ;CHECK: pblendw $7, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() %2 = call <8 x i16> @llvm.x86.sse41.pblendw(<8 x i16> %a0, <8 x i16> %a1, i8 7) ret <8 x i16> %2 -} -declare <8 x i16> @llvm.x86.sse41.pblendw(<8 x i16>, <8 x i16>, i8) nounwind readnone - -; TODO stack_fold_pclmulqdq -declare <2 x i64> @llvm.x86.pclmulqdq(<2 x i64>, <2 x i64>, i8) nounwind readnone - -define <16 x i8> @stack_fold_pcmpeqb(<16 x i8> %a0, <16 x i8> %a1) { +} +declare <8 x i16> @llvm.x86.sse41.pblendw(<8 x i16>, <8 x i16>, i8) nounwind readnone + +define <2 x i64> @stack_fold_pclmulqdq(<2 x i64> %a0, <2 x i64> %a1) { + ;CHECK-LABEL: stack_fold_pclmulqdq + ;CHECK: pclmulqdq $0, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <2 x i64> @llvm.x86.pclmulqdq(<2 x i64> %a0, <2 x i64> %a1, i8 0) + ret <2 x i64> %2 +} +declare <2 x i64> @llvm.x86.pclmulqdq(<2 x i64>, <2 x i64>, i8) nounwind readnone + +define <16 x i8> @stack_fold_pcmpeqb(<16 x i8> %a0, <16 x i8> %a1) { ;CHECK-LABEL: stack_fold_pcmpeqb ;CHECK: pcmpeqb {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() @@ -288,16 +332,28 @@ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() %2 = icmp eq <8 x i16> %a0, %a1 %3 = sext <8 x i1> %2 to <8 x i16> - ret <8 x i16> %3 -} - -; TODO stack_fold_pcmpestri -declare i32 @llvm.x86.sse42.pcmpestri128(<16 x i8>, i32, <16 x i8>, i32, i8) nounwind readnone - -; TODO stack_fold_pcmpestrm -declare <16 x i8> @llvm.x86.sse42.pcmpestrm128(<16 x i8>, i32, <16 x i8>, i32, i8) nounwind readnone - -define <16 x i8> @stack_fold_pcmpgtb(<16 x i8> %a0, <16 x i8> %a1) { + ret <8 x i16> %3 +} + +define i32 @stack_fold_pcmpestri(<16 x i8> %a0, <16 x i8> %a1) { + ;CHECK-LABEL: stack_fold_pcmpestri + ;CHECK: pcmpestri $7, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{rax},~{flags}"() + %2 = call i32 @llvm.x86.sse42.pcmpestri128(<16 x i8> %a0, i32 7, <16 x i8> %a1, i32 7, i8 7) + ret i32 %2 +} +declare i32 @llvm.x86.sse42.pcmpestri128(<16 x i8>, i32, <16 x i8>, i32, i8) nounwind readnone + +define <16 x i8> @stack_fold_pcmpestrm(<16 x i8> %a0, <16 x i8> %a1) { + ;CHECK-LABEL: stack_fold_pcmpestrm + ;CHECK: pcmpestrm $7, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{rax},~{flags}"() + %2 = call <16 x i8> @llvm.x86.sse42.pcmpestrm128(<16 x i8> %a0, i32 7, <16 x i8> %a1, i32 7, i8 7) + ret <16 x i8> %2 +} +declare <16 x i8> @llvm.x86.sse42.pcmpestrm128(<16 x i8>, i32, <16 x i8>, i32, i8) nounwind readnone + +define <16 x i8> @stack_fold_pcmpgtb(<16 x i8> %a0, <16 x i8> %a1) { ;CHECK-LABEL: stack_fold_pcmpgtb ;CHECK: pcmpgtb {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() @@ -330,21 +386,50 @@ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() %2 = icmp sgt <8 x i16> %a0, %a1 %3 = sext <8 x i1> %2 to <8 x i16> - ret <8 x i16> %3 -} - -; TODO stack_fold_pcmpistri -declare i32 @llvm.x86.sse42.pcmpistri128(<16 x i8>, <16 x i8>, i8) nounwind readnone - -; TODO stack_fold_pcmpistrm -declare <16 x i8> @llvm.x86.sse42.pcmpistrm128(<16 x i8>, <16 x i8>, i8) nounwind readnone - -; TODO stack_fold_pextrb -; TODO stack_fold_pextrd -; TODO stack_fold_pextrq -; TODO stack_fold_pextrw - -define <4 x i32> @stack_fold_phaddd(<4 x i32> %a0, <4 x i32> %a1) { + ret <8 x i16> %3 +} + +define i32 @stack_fold_pcmpistri(<16 x i8> %a0, <16 x i8> %a1) { + ;CHECK-LABEL: stack_fold_pcmpistri + ;CHECK: pcmpistri $7, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call i32 @llvm.x86.sse42.pcmpistri128(<16 x i8> %a0, <16 x i8> %a1, i8 7) + ret i32 %2 +} +declare i32 @llvm.x86.sse42.pcmpistri128(<16 x i8>, <16 x i8>, i8) nounwind readnone + +define <16 x i8> @stack_fold_pcmpistrm(<16 x i8> %a0, <16 x i8> %a1) { + ;CHECK-LABEL: stack_fold_pcmpistrm + ;CHECK: pcmpistrm $7, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <16 x i8> @llvm.x86.sse42.pcmpistrm128(<16 x i8> %a0, <16 x i8> %a1, i8 7) + ret <16 x i8> %2 +} +declare <16 x i8> @llvm.x86.sse42.pcmpistrm128(<16 x i8>, <16 x i8>, i8) nounwind readnone + +; TODO stack_fold_pextrb + +define i32 @stack_fold_pextrd(<4 x i32> %a0) { + ;CHECK-LABEL: stack_fold_pextrd + ;CHECK: pextrd $1, {{%xmm[0-9][0-9]*}}, {{-?[0-9]*}}(%rsp) {{.*#+}} 4-byte Folded Spill + ;CHECK: movl {{-?[0-9]*}}(%rsp), %eax {{.*#+}} 4-byte Reload + %1 = extractelement <4 x i32> %a0, i32 1 + %2 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + ret i32 %1 +} + +define i64 @stack_fold_pextrq(<2 x i64> %a0) { + ;CHECK-LABEL: stack_fold_pextrq + ;CHECK: pextrq $1, {{%xmm[0-9][0-9]*}}, {{-?[0-9]*}}(%rsp) {{.*#+}} 8-byte Folded Spill + ;CHECK: movq {{-?[0-9]*}}(%rsp), %rax {{.*#+}} 8-byte Reload + %1 = extractelement <2 x i64> %a0, i32 1 + %2 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + ret i64 %1 +} + +; TODO stack_fold_pextrw + +define <4 x i32> @stack_fold_phaddd(<4 x i32> %a0, <4 x i32> %a1) { ;CHECK-LABEL: stack_fold_phaddd ;CHECK: phaddd {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() @@ -368,13 +453,19 @@ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() %2 = call <8 x i16> @llvm.x86.ssse3.phadd.w.128(<8 x i16> %a0, <8 x i16> %a1) ret <8 x i16> %2 -} -declare <8 x i16> @llvm.x86.ssse3.phadd.w.128(<8 x i16>, <8 x i16>) nounwind readnone - -; TODO stack_fold_phminposuw -declare <8 x i16> @llvm.x86.sse41.phminposuw(<8 x i16>) nounwind readnone - -define <4 x i32> @stack_fold_phsubd(<4 x i32> %a0, <4 x i32> %a1) { +} +declare <8 x i16> @llvm.x86.ssse3.phadd.w.128(<8 x i16>, <8 x i16>) nounwind readnone + +define <8 x i16> @stack_fold_phminposuw(<8 x i16> %a0) { + ;CHECK-LABEL: stack_fold_phminposuw + ;CHECK: phminposuw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <8 x i16> @llvm.x86.sse41.phminposuw(<8 x i16> %a0) + ret <8 x i16> %2 +} +declare <8 x i16> @llvm.x86.sse41.phminposuw(<8 x i16>) nounwind readnone + +define <4 x i32> @stack_fold_phsubd(<4 x i32> %a0, <4 x i32> %a1) { ;CHECK-LABEL: stack_fold_phsubd ;CHECK: phsubd {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() @@ -398,16 +489,43 @@ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() %2 = call <8 x i16> @llvm.x86.ssse3.phsub.w.128(<8 x i16> %a0, <8 x i16> %a1) ret <8 x i16> %2 -} -declare <8 x i16> @llvm.x86.ssse3.phsub.w.128(<8 x i16>, <8 x i16>) nounwind readnone - -; TODO stack_fold_pinsrb -; TODO stack_fold_pinsrd -; TODO stack_fold_pinsrq -; TODO stack_fold_pinsrw - -define <8 x i16> @stack_fold_pmaddubsw(<16 x i8> %a0, <16 x i8> %a1) { - ;CHECK-LABEL: stack_fold_pmaddubsw +} +declare <8 x i16> @llvm.x86.ssse3.phsub.w.128(<8 x i16>, <8 x i16>) nounwind readnone + +define <16 x i8> @stack_fold_pinsrb(<16 x i8> %a0, i8 %a1) { + ;CHECK-LABEL: stack_fold_pinsrb + ;CHECK: pinsrb $1, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 4-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + %2 = insertelement <16 x i8> %a0, i8 %a1, i32 1 + ret <16 x i8> %2 +} + +define <4 x i32> @stack_fold_pinsrd(<4 x i32> %a0, i32 %a1) { + ;CHECK-LABEL: stack_fold_pinsrd + ;CHECK: pinsrd $1, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 4-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + %2 = insertelement <4 x i32> %a0, i32 %a1, i32 1 + ret <4 x i32> %2 +} + +define <2 x i64> @stack_fold_pinsrq(<2 x i64> %a0, i64 %a1) { + ;CHECK-LABEL: stack_fold_pinsrq + ;CHECK: pinsrq $1, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 8-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + %2 = insertelement <2 x i64> %a0, i64 %a1, i32 1 + ret <2 x i64> %2 +} + +define <8 x i16> @stack_fold_pinsrw(<8 x i16> %a0, i16 %a1) { + ;CHECK-LABEL: stack_fold_pinsrw + ;CHECK: pinsrw $1, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 4-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{rbp},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15}"() + %2 = insertelement <8 x i16> %a0, i16 %a1, i32 1 + ret <8 x i16> %2 +} + +define <8 x i16> @stack_fold_pmaddubsw(<16 x i8> %a0, <16 x i8> %a1) { + ;CHECK-LABEL: stack_fold_pmaddubsw ;CHECK: pmaddubsw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() %2 = call <8 x i16> @llvm.x86.ssse3.pmadd.ub.sw.128(<16 x i8> %a0, <16 x i8> %a1) @@ -529,46 +647,118 @@ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() %2 = call <8 x i16> @llvm.x86.sse41.pminuw(<8 x i16> %a0, <8 x i16> %a1) ret <8 x i16> %2 -} -declare <8 x i16> @llvm.x86.sse41.pminuw(<8 x i16>, <8 x i16>) nounwind readnone - -; TODO stack_fold_pmovsxbd -declare <4 x i32> @llvm.x86.sse41.pmovsxbd(<16 x i8>) nounwind readnone - -; TODO stack_fold_pmovsxbq -declare <2 x i64> @llvm.x86.sse41.pmovsxbq(<16 x i8>) nounwind readnone - -; TODO stack_fold_pmovsxbw -declare <8 x i16> @llvm.x86.sse41.pmovsxbw(<16 x i8>) nounwind readnone - -; TODO stack_fold_pmovsxdq -declare <2 x i64> @llvm.x86.sse41.pmovsxdq(<4 x i32>) nounwind readnone - -; TODO stack_fold_pmovsxwd -declare <4 x i32> @llvm.x86.sse41.pmovsxwd(<8 x i16>) nounwind readnone - -; TODO stack_fold_pmovsxwq -declare <2 x i64> @llvm.x86.sse41.pmovsxwq(<8 x i16>) nounwind readnone - -; TODO stack_fold_pmovzxbd -declare <4 x i32> @llvm.x86.sse41.pmovzxbd(<16 x i8>) nounwind readnone - -; TODO stack_fold_pmovzxbq -declare <2 x i64> @llvm.x86.sse41.pmovzxbq(<16 x i8>) nounwind readnone - -; TODO stack_fold_pmovzxbw -declare <8 x i16> @llvm.x86.sse41.pmovzxbw(<16 x i8>) nounwind readnone - -; TODO stack_fold_pmovzxdq -declare <2 x i64> @llvm.x86.sse41.pmovzxdq(<4 x i32>) nounwind readnone - -; TODO stack_fold_pmovzxwd -declare <4 x i32> @llvm.x86.sse41.pmovzxwd(<8 x i16>) nounwind readnone - -; TODO stack_fold_pmovzxwq -declare <2 x i64> @llvm.x86.sse41.pmovzxwq(<8 x i16>) nounwind readnone - -define <2 x i64> @stack_fold_pmuldq(<4 x i32> %a0, <4 x i32> %a1) { +} +declare <8 x i16> @llvm.x86.sse41.pminuw(<8 x i16>, <8 x i16>) nounwind readnone + +define <4 x i32> @stack_fold_pmovsxbd(<16 x i8> %a0) { + ;CHECK-LABEL: stack_fold_pmovsxbd + ;CHECK: pmovsxbd {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <4 x i32> @llvm.x86.sse41.pmovsxbd(<16 x i8> %a0) + ret <4 x i32> %2 +} +declare <4 x i32> @llvm.x86.sse41.pmovsxbd(<16 x i8>) nounwind readnone + +define <2 x i64> @stack_fold_pmovsxbq(<16 x i8> %a0) { + ;CHECK-LABEL: stack_fold_pmovsxbq + ;CHECK: pmovsxbq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <2 x i64> @llvm.x86.sse41.pmovsxbq(<16 x i8> %a0) + ret <2 x i64> %2 +} +declare <2 x i64> @llvm.x86.sse41.pmovsxbq(<16 x i8>) nounwind readnone + +define <8 x i16> @stack_fold_pmovsxbw(<16 x i8> %a0) { + ;CHECK-LABEL: stack_fold_pmovsxbw + ;CHECK: pmovsxbw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <8 x i16> @llvm.x86.sse41.pmovsxbw(<16 x i8> %a0) + ret <8 x i16> %2 +} +declare <8 x i16> @llvm.x86.sse41.pmovsxbw(<16 x i8>) nounwind readnone + +define <2 x i64> @stack_fold_pmovsxdq(<4 x i32> %a0) { + ;CHECK-LABEL: stack_fold_pmovsxdq + ;CHECK: pmovsxdq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <2 x i64> @llvm.x86.sse41.pmovsxdq(<4 x i32> %a0) + ret <2 x i64> %2 +} +declare <2 x i64> @llvm.x86.sse41.pmovsxdq(<4 x i32>) nounwind readnone + +define <4 x i32> @stack_fold_pmovsxwd(<8 x i16> %a0) { + ;CHECK-LABEL: stack_fold_pmovsxwd + ;CHECK: pmovsxwd {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <4 x i32> @llvm.x86.sse41.pmovsxwd(<8 x i16> %a0) + ret <4 x i32> %2 +} +declare <4 x i32> @llvm.x86.sse41.pmovsxwd(<8 x i16>) nounwind readnone + +define <2 x i64> @stack_fold_pmovsxwq(<8 x i16> %a0) { + ;CHECK-LABEL: stack_fold_pmovsxwq + ;CHECK: pmovsxwq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <2 x i64> @llvm.x86.sse41.pmovsxwq(<8 x i16> %a0) + ret <2 x i64> %2 +} +declare <2 x i64> @llvm.x86.sse41.pmovsxwq(<8 x i16>) nounwind readnone + +define <4 x i32> @stack_fold_pmovzxbd(<16 x i8> %a0) { + ;CHECK-LABEL: stack_fold_pmovzxbd + ;CHECK: pmovzxbd {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <4 x i32> @llvm.x86.sse41.pmovzxbd(<16 x i8> %a0) + ret <4 x i32> %2 +} +declare <4 x i32> @llvm.x86.sse41.pmovzxbd(<16 x i8>) nounwind readnone + +define <2 x i64> @stack_fold_pmovzxbq(<16 x i8> %a0) { + ;CHECK-LABEL: stack_fold_pmovzxbq + ;CHECK: pmovzxbq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <2 x i64> @llvm.x86.sse41.pmovzxbq(<16 x i8> %a0) + ret <2 x i64> %2 +} +declare <2 x i64> @llvm.x86.sse41.pmovzxbq(<16 x i8>) nounwind readnone + +define <8 x i16> @stack_fold_pmovzxbw(<16 x i8> %a0) { + ;CHECK-LABEL: stack_fold_pmovzxbw + ;CHECK: pmovzxbw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <8 x i16> @llvm.x86.sse41.pmovzxbw(<16 x i8> %a0) + ret <8 x i16> %2 +} +declare <8 x i16> @llvm.x86.sse41.pmovzxbw(<16 x i8>) nounwind readnone + +define <2 x i64> @stack_fold_pmovzxdq(<4 x i32> %a0) { + ;CHECK-LABEL: stack_fold_pmovzxdq + ;CHECK: pmovzxdq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <2 x i64> @llvm.x86.sse41.pmovzxdq(<4 x i32> %a0) + ret <2 x i64> %2 +} +declare <2 x i64> @llvm.x86.sse41.pmovzxdq(<4 x i32>) nounwind readnone + +define <4 x i32> @stack_fold_pmovzxwd(<8 x i16> %a0) { + ;CHECK-LABEL: stack_fold_pmovzxwd + ;CHECK: pmovzxwd {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <4 x i32> @llvm.x86.sse41.pmovzxwd(<8 x i16> %a0) + ret <4 x i32> %2 +} +declare <4 x i32> @llvm.x86.sse41.pmovzxwd(<8 x i16>) nounwind readnone + +define <2 x i64> @stack_fold_pmovzxwq(<8 x i16> %a0) { + ;CHECK-LABEL: stack_fold_pmovzxwq + ;CHECK: pmovzxwq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <2 x i64> @llvm.x86.sse41.pmovzxwq(<8 x i16> %a0) + ret <2 x i64> %2 +} +declare <2 x i64> @llvm.x86.sse41.pmovzxwq(<8 x i16>) nounwind readnone + +define <2 x i64> @stack_fold_pmuldq(<4 x i32> %a0, <4 x i32> %a1) { ;CHECK-LABEL: stack_fold_pmuldq ;CHECK: pmuldq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() @@ -793,13 +983,19 @@ ;CHECK: psubd {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() %2 = sub <4 x i32> %a0, %a1 - ret <4 x i32> %2 -} - -; TODO stack_fold_psubq - -define <16 x i8> @stack_fold_psubsb(<16 x i8> %a0, <16 x i8> %a1) { - ;CHECK-LABEL: stack_fold_psubsb + ret <4 x i32> %2 +} + +define <2 x i64> @stack_fold_psubq(<2 x i64> %a0, <2 x i64> %a1) { + ;CHECK-LABEL: stack_fold_psubq + ;CHECK: psubq {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = sub <2 x i64> %a0, %a1 + ret <2 x i64> %2 +} + +define <16 x i8> @stack_fold_psubsb(<16 x i8> %a0, <16 x i8> %a1) { + ;CHECK-LABEL: stack_fold_psubsb ;CHECK: psubsb {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() %2 = call <16 x i8> @llvm.x86.sse2.psubs.b(<16 x i8> %a0, <16 x i8> %a1) @@ -813,27 +1009,45 @@ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() %2 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> %a0, <8 x i16> %a1) ret <8 x i16> %2 -} -declare <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16>, <8 x i16>) nounwind readnone - -; TODO stack_fold_psubusb -declare <16 x i8> @llvm.x86.sse2.psubus.b(<16 x i8>, <16 x i8>) nounwind readnone - -; TODO stack_fold_psubusw -declare <8 x i16> @llvm.x86.sse2.psubus.w(<8 x i16>, <8 x i16>) nounwind readnone - -define <8 x i16> @stack_fold_psubw(<8 x i16> %a0, <8 x i16> %a1) { +} +declare <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16>, <8 x i16>) nounwind readnone + +define <16 x i8> @stack_fold_psubusb(<16 x i8> %a0, <16 x i8> %a1) { + ;CHECK-LABEL: stack_fold_psubusb + ;CHECK: psubusb {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <16 x i8> @llvm.x86.sse2.psubus.b(<16 x i8> %a0, <16 x i8> %a1) + ret <16 x i8> %2 +} +declare <16 x i8> @llvm.x86.sse2.psubus.b(<16 x i8>, <16 x i8>) nounwind readnone + +define <8 x i16> @stack_fold_psubusw(<8 x i16> %a0, <8 x i16> %a1) { + ;CHECK-LABEL: stack_fold_psubusw + ;CHECK: psubusw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call <8 x i16> @llvm.x86.sse2.psubus.w(<8 x i16> %a0, <8 x i16> %a1) + ret <8 x i16> %2 +} +declare <8 x i16> @llvm.x86.sse2.psubus.w(<8 x i16>, <8 x i16>) nounwind readnone + +define <8 x i16> @stack_fold_psubw(<8 x i16> %a0, <8 x i16> %a1) { ;CHECK-LABEL: stack_fold_psubw ;CHECK: psubw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() %2 = sub <8 x i16> %a0, %a1 - ret <8 x i16> %2 -} - -; TODO stack_fold_ptest -declare i32 @llvm.x86.sse41.ptestc(<2 x i64>, <2 x i64>) nounwind readnone - -define <16 x i8> @stack_fold_punpckhbw(<16 x i8> %a0, <16 x i8> %a1) { + ret <8 x i16> %2 +} + +define i32 @stack_fold_ptest(<2 x i64> %a0, <2 x i64> %a1) { + ;CHECK-LABEL: stack_fold_ptest + ;CHECK: ptest {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload + %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"() + %2 = call i32 @llvm.x86.sse41.ptestc(<2 x i64> %a0, <2 x i64> %a1) + ret i32 %2 +} +declare i32 @llvm.x86.sse41.ptestc(<2 x i64>, <2 x i64>) nounwind readnone + +define <16 x i8> @stack_fold_punpckhbw(<16 x i8> %a0, <16 x i8> %a1) { ;CHECK-LABEL: stack_fold_punpckhbw ;CHECK: punpckhbw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()