Xen project Mailing List

[Xen-devel] [PATCH v3 19/34] x86emul: use simd_128 also for legacy vector shift insns

To: "xen-devel" <xen-devel@xxxxxxxxxxxxxxxxxxxx>

From: "Jan Beulich" <JBeulich@xxxxxxxx>

Date: Tue, 18 Sep 2018 06:05:30 -0600

Cc: George Dunlap <George.Dunlap@xxxxxxxxxxxxx>, Andrew Cooper <andrew.cooper3@xxxxxxxxxx>, Wei Liu <wei.liu2@xxxxxxxxxx>

Delivery-date: Tue, 18 Sep 2018 12:05:35 +0000

List-id: Xen developer discussion <xen-devel.lists.xenproject.org>

This eliminates a separate case block here, and allows to get away with fewer new ones when adding AVX512 vector shifts. Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx> --- v3: New. --- a/xen/arch/x86/x86_emulate/x86_emulate.c +++ b/xen/arch/x86/x86_emulate/x86_emulate.c @@ -364,19 +364,19 @@ static const struct twobyte_table { [0xc7] = { ImplicitOps|ModRM }, [0xc8 ... 0xcf] = { ImplicitOps }, [0xd0] = { DstImplicit|SrcMem|ModRM, simd_other }, - [0xd1 ... 0xd3] = { DstImplicit|SrcMem|ModRM, simd_other }, + [0xd1 ... 0xd3] = { DstImplicit|SrcMem|ModRM, simd_128 }, [0xd4 ... 0xd5] = { DstImplicit|SrcMem|ModRM, simd_packed_int, d8s_vl }, [0xd6] = { DstMem|SrcImplicit|ModRM|Mov, simd_other, 3 }, [0xd7] = { DstReg|SrcImplicit|ModRM|Mov }, [0xd8 ... 0xdf] = { DstImplicit|SrcMem|ModRM, simd_packed_int, d8s_vl }, [0xe0] = { DstImplicit|SrcMem|ModRM, simd_packed_int, d8s_vl }, - [0xe1 ... 0xe2] = { DstImplicit|SrcMem|ModRM, simd_other }, + [0xe1 ... 0xe2] = { DstImplicit|SrcMem|ModRM, simd_128 }, [0xe3 ... 0xe5] = { DstImplicit|SrcMem|ModRM, simd_packed_int, d8s_vl }, [0xe6] = { DstImplicit|SrcMem|ModRM|Mov, simd_other }, [0xe7] = { DstMem|SrcImplicit|ModRM|Mov, simd_packed_int, d8s_vl }, [0xe8 ... 0xef] = { DstImplicit|SrcMem|ModRM, simd_packed_int, d8s_vl }, [0xf0] = { DstImplicit|SrcMem|ModRM|Mov, simd_other }, - [0xf1 ... 0xf3] = { DstImplicit|SrcMem|ModRM, simd_other }, + [0xf1 ... 0xf3] = { DstImplicit|SrcMem|ModRM, simd_128 }, [0xf4 ... 0xf6] = { DstImplicit|SrcMem|ModRM, simd_packed_int, d8s_vl }, [0xf7] = { DstMem|SrcMem|ModRM|Mov, simd_packed_int }, [0xf8 ... 0xfe] = { DstImplicit|SrcMem|ModRM, simd_packed_int, d8s_vl }, @@ -3327,7 +3327,8 @@ x86_decode( break; case simd_128: - op_bytes = 16; + /* The special case here are MMX shift insns. */ + op_bytes = vex.opcx || vex.pfx ? 16 : 8; break; case simd_256: @@ -6455,6 +6456,12 @@ x86_emulate( case X86EMUL_OPC_VEX_66(0x0f, 0x75): /* vpcmpeqw {x,y}mm/mem,{x,y}mm,{x,y}mm */ CASE_SIMD_PACKED_INT(0x0f, 0x76): /* pcmpeqd {,x}mm/mem,{,x}mm */ case X86EMUL_OPC_VEX_66(0x0f, 0x76): /* vpcmpeqd {x,y}mm/mem,{x,y}mm,{x,y}mm */ + CASE_SIMD_PACKED_INT(0x0f, 0xd1): /* psrlw {,x}mm/mem,{,x}mm */ + case X86EMUL_OPC_VEX_66(0x0f, 0xd1): /* vpsrlw xmm/m128,{x,y}mm,{x,y}mm */ + CASE_SIMD_PACKED_INT(0x0f, 0xd2): /* psrld {,x}mm/mem,{,x}mm */ + case X86EMUL_OPC_VEX_66(0x0f, 0xd2): /* vpsrld xmm/m128,{x,y}mm,{x,y}mm */ + CASE_SIMD_PACKED_INT(0x0f, 0xd3): /* psrlq {,x}mm/mem,{,x}mm */ + case X86EMUL_OPC_VEX_66(0x0f, 0xd3): /* vpsrlq xmm/m128,{x,y}mm,{x,y}mm */ case X86EMUL_OPC_66(0x0f, 0xd4): /* paddq xmm/m128,xmm */ case X86EMUL_OPC_VEX_66(0x0f, 0xd4): /* vpaddq {x,y}mm/mem,{x,y}mm,{x,y}mm */ CASE_SIMD_PACKED_INT(0x0f, 0xd5): /* pmullw {,x}mm/mem,{,x}mm */ @@ -6477,6 +6484,10 @@ x86_emulate( case X86EMUL_OPC_VEX_66(0x0f, 0xdf): /* vpandn {x,y}mm/mem,{x,y}mm,{x,y}mm */ case X86EMUL_OPC_66(0x0f, 0xe0): /* pavgb xmm/m128,xmm */ case X86EMUL_OPC_VEX_66(0x0f, 0xe0): /* vpavgb {x,y}mm/mem,{x,y}mm,{x,y}mm */ + CASE_SIMD_PACKED_INT(0x0f, 0xe1): /* psraw {,x}mm/mem,{,x}mm */ + case X86EMUL_OPC_VEX_66(0x0f, 0xe1): /* vpsraw xmm/m128,{x,y}mm,{x,y}mm */ + CASE_SIMD_PACKED_INT(0x0f, 0xe2): /* psrad {,x}mm/mem,{,x}mm */ + case X86EMUL_OPC_VEX_66(0x0f, 0xe2): /* vpsrad xmm/m128,{x,y}mm,{x,y}mm */ case X86EMUL_OPC_66(0x0f, 0xe3): /* pavgw xmm/m128,xmm */ case X86EMUL_OPC_VEX_66(0x0f, 0xe3): /* vpavgw {x,y}mm/mem,{x,y}mm,{x,y}mm */ case X86EMUL_OPC_66(0x0f, 0xe4): /* pmulhuw xmm/m128,xmm */ @@ -6499,6 +6510,12 @@ x86_emulate( case X86EMUL_OPC_VEX_66(0x0f, 0xee): /* vpmaxsw {x,y}mm/mem,{x,y}mm,{x,y}mm */ CASE_SIMD_PACKED_INT(0x0f, 0xef): /* pxor {,x}mm/mem,{,x}mm */ case X86EMUL_OPC_VEX_66(0x0f, 0xef): /* vpxor {x,y}mm/mem,{x,y}mm,{x,y}mm */ + CASE_SIMD_PACKED_INT(0x0f, 0xf1): /* psllw {,x}mm/mem,{,x}mm */ + case X86EMUL_OPC_VEX_66(0x0f, 0xf1): /* vpsllw xmm/m128,{x,y}mm,{x,y}mm */ + CASE_SIMD_PACKED_INT(0x0f, 0xf2): /* pslld {,x}mm/mem,{,x}mm */ + case X86EMUL_OPC_VEX_66(0x0f, 0xf2): /* vpslld xmm/m128,{x,y}mm,{x,y}mm */ + CASE_SIMD_PACKED_INT(0x0f, 0xf3): /* psllq {,x}mm/mem,{,x}mm */ + case X86EMUL_OPC_VEX_66(0x0f, 0xf3): /* vpsllq xmm/m128,{x,y}mm,{x,y}mm */ case X86EMUL_OPC_66(0x0f, 0xf4): /* pmuludq xmm/m128,xmm */ case X86EMUL_OPC_VEX_66(0x0f, 0xf4): /* vpmuludq {x,y}mm/mem,{x,y}mm,{x,y}mm */ case X86EMUL_OPC_66(0x0f, 0xf6): /* psadbw xmm/m128,xmm */ @@ -7831,25 +7848,6 @@ x86_emulate( } break; - CASE_SIMD_PACKED_INT(0x0f, 0xd1): /* psrlw {,x}mm/mem,{,x}mm */ - case X86EMUL_OPC_VEX_66(0x0f, 0xd1): /* vpsrlw xmm/m128,{x,y}mm,{x,y}mm */ - CASE_SIMD_PACKED_INT(0x0f, 0xd2): /* psrld {,x}mm/mem,{,x}mm */ - case X86EMUL_OPC_VEX_66(0x0f, 0xd2): /* vpsrld xmm/m128,{x,y}mm,{x,y}mm */ - CASE_SIMD_PACKED_INT(0x0f, 0xd3): /* psrlq {,x}mm/mem,{,x}mm */ - case X86EMUL_OPC_VEX_66(0x0f, 0xd3): /* vpsrlq xmm/m128,{x,y}mm,{x,y}mm */ - CASE_SIMD_PACKED_INT(0x0f, 0xe1): /* psraw {,x}mm/mem,{,x}mm */ - case X86EMUL_OPC_VEX_66(0x0f, 0xe1): /* vpsraw xmm/m128,{x,y}mm,{x,y}mm */ - CASE_SIMD_PACKED_INT(0x0f, 0xe2): /* psrad {,x}mm/mem,{,x}mm */ - case X86EMUL_OPC_VEX_66(0x0f, 0xe2): /* vpsrad xmm/m128,{x,y}mm,{x,y}mm */ - CASE_SIMD_PACKED_INT(0x0f, 0xf1): /* psllw {,x}mm/mem,{,x}mm */ - case X86EMUL_OPC_VEX_66(0x0f, 0xf1): /* vpsllw xmm/m128,{x,y}mm,{x,y}mm */ - CASE_SIMD_PACKED_INT(0x0f, 0xf2): /* pslld {,x}mm/mem,{,x}mm */ - case X86EMUL_OPC_VEX_66(0x0f, 0xf2): /* vpslld xmm/m128,{x,y}mm,{x,y}mm */ - CASE_SIMD_PACKED_INT(0x0f, 0xf3): /* psllq {,x}mm/mem,{,x}mm */ - case X86EMUL_OPC_VEX_66(0x0f, 0xf3): /* vpsllq xmm/m128,{x,y}mm,{x,y}mm */ - op_bytes = vex.pfx ? 16 : 8; - goto simd_0f_int; - case X86EMUL_OPC(0x0f, 0xd4): /* paddq mm/m64,mm */ case X86EMUL_OPC(0x0f, 0xf4): /* pmuludq mm/m64,mm */ case X86EMUL_OPC(0x0f, 0xfb): /* psubq mm/m64,mm */ _______________________________________________ Xen-devel mailing list Xen-devel@xxxxxxxxxxxxxxxxxxxx https://lists.xenproject.org/mailman/listinfo/xen-devel

©2013 Xen Project, A Linux Foundation Collaborative Project. All Rights Reserved.
Linux Foundation is a registered trademark of The Linux Foundation.
Xen Project is a trademark of The Linux Foundation.