On 20/12/2019 16:01, Andrew Cooper wrote: >> Suggested-by: Andrew Cooper <[email protected]> >> Signed-off-by: Jan Beulich <[email protected]> >> --- >> I'll be happy to take suggestions allowing to avoid -Wno-unused-label. > I think I'm going to need a little while to figure out how this works.
So, after having had an evening playing with this, things get massively simpler when NO_MMX is folded with NO_SIMD. MMX is a SIMD technology, and I can't see a compelling reason to control their inclusion separately. We're either going to want everything or nothing. The attached incremental works for me without a single out-of-place label. There is some further cleanup which can be done such as not making the CASE_ macros conditional. (OTOH, the compile error from might be helpful to keep in some form). Thoughts? ~Andrew
>From aa8ea9998c643bc518ec8b4208e20da352a54129 Mon Sep 17 00:00:00 2001 From: Andrew Cooper <[email protected]> Date: Thu, 2 Apr 2020 23:11:58 +0100 Subject: incremental diff --git a/xen/arch/x86/Makefile b/xen/arch/x86/Makefile index 71f1d7784e..e954edbc2e 100644 --- a/xen/arch/x86/Makefile +++ b/xen/arch/x86/Makefile @@ -72,9 +72,6 @@ obj-y += hpet.o obj-y += vm_event.o obj-y += xstate.o -ifneq ($(CONFIG_HVM),y) -x86_emulate.o: CFLAGS += -Wno-unused-label -endif x86_emulate.o: x86_emulate/x86_emulate.c x86_emulate/x86_emulate.h efi-y := $(shell if [ ! -r $(BASEDIR)/include/xen/compile.h -o \ diff --git a/xen/arch/x86/x86_emulate/x86_emulate.c b/xen/arch/x86/x86_emulate/x86_emulate.c index 6d5e5b8b2a..21c57941ef 100644 --- a/xen/arch/x86/x86_emulate/x86_emulate.c +++ b/xen/arch/x86/x86_emulate/x86_emulate.c @@ -6172,6 +6172,8 @@ x86_emulate( #endif +#ifndef X86EMUL_NO_SIMD + CASE_SIMD_SCALAR_FP(, 0x0f, 0x2b): /* movnts{s,d} xmm,mem */ host_and_vcpu_must_have(sse4a); /* fall through */ @@ -6309,8 +6311,6 @@ x86_emulate( insn_bytes = EVEX_PFX_BYTES + 2; break; -#ifndef X86EMUL_NO_SIMD - case X86EMUL_OPC_66(0x0f, 0x12): /* movlpd m64,xmm */ case X86EMUL_OPC_VEX_66(0x0f, 0x12): /* vmovlpd m64,xmm,xmm */ CASE_SIMD_PACKED_FP_VEX(0x0f, 0x13): /* movlp{s,d} xmm,m64 */ @@ -6445,7 +6445,7 @@ x86_emulate( goto done; break; -#if !defined(X86EMUL_NO_MMX) && !defined(X86EMUL_NO_SIMD) +#ifndef X86EMUL_NO_SIMD case X86EMUL_OPC_66(0x0f, 0x2a): /* cvtpi2pd mm/m64,xmm */ if ( ea.type == OP_REG ) @@ -6458,8 +6458,6 @@ x86_emulate( op_bytes = (b & 4) && (vex.pfx & VEX_PREFIX_DOUBLE_MASK) ? 16 : 8; goto simd_0f_fp; -#endif /* !X86EMUL_NO_MMX && !X86EMUL_NO_SIMD */ - CASE_SIMD_SCALAR_FP_VEX(0x0f, 0x2a): /* {,v}cvtsi2s{s,d} r/m,xmm */ if ( vex.opcx == vex_none ) { @@ -6681,6 +6679,8 @@ x86_emulate( op_bytes = 4 << evex.w; goto vcomi; +#endif /* !X86EMUL_NO_SIMD */ + case X86EMUL_OPC(0x0f, 0x30): /* wrmsr */ generate_exception_if(!mode_ring0(), EXC_GP, 0); fail_if(ops->write_msr == NULL); @@ -6862,8 +6862,6 @@ x86_emulate( generate_exception_if(!vex.l || vex.w, EXC_UD); goto opmask_common; -#endif /* X86EMUL_NO_SIMD */ - CASE_SIMD_PACKED_FP_VEX(0x0f, 0x50): /* movmskp{s,d} xmm,reg */ /* vmovmskp{s,d} {x,y}mm,reg */ CASE_SIMD_PACKED_INT_VEX(0x0f, 0xd7): /* pmovmskb {,x}mm,reg */ @@ -6947,8 +6945,6 @@ x86_emulate( evex.w); goto avx512f_all_fp; -#ifndef X86EMUL_NO_SIMD - CASE_SIMD_PACKED_FP_VEX(0x0f, 0x5b): /* cvt{ps,dq}2{dq,ps} xmm/mem,xmm */ /* vcvt{ps,dq}2{dq,ps} {x,y}mm/mem,{x,y}mm */ case X86EMUL_OPC_F3(0x0f, 0x5b): /* cvttps2dq xmm/mem,xmm */ @@ -6979,8 +6975,6 @@ x86_emulate( op_bytes = 16 << evex.lr; goto simd_zmm; -#endif /* !X86EMUL_NO_SIMD */ - CASE_SIMD_PACKED_INT_VEX(0x0f, 0x60): /* punpcklbw {,x}mm/mem,{,x}mm */ /* vpunpcklbw {x,y}mm/mem,{x,y}mm,{x,y}mm */ CASE_SIMD_PACKED_INT_VEX(0x0f, 0x61): /* punpcklwd {,x}mm/mem,{,x}mm */ @@ -7007,7 +7001,6 @@ x86_emulate( /* vpackusbw {x,y}mm/mem,{x,y}mm,{x,y}mm */ CASE_SIMD_PACKED_INT_VEX(0x0f, 0x6b): /* packsswd {,x}mm/mem,{,x}mm */ /* vpacksswd {x,y}mm/mem,{x,y}mm,{x,y}mm */ -#ifndef X86EMUL_NO_SIMD case X86EMUL_OPC_66(0x0f, 0x6c): /* punpcklqdq xmm/m128,xmm */ case X86EMUL_OPC_VEX_66(0x0f, 0x6c): /* vpunpcklqdq {x,y}mm/mem,{x,y}mm,{x,y}mm */ case X86EMUL_OPC_66(0x0f, 0x6d): /* punpckhqdq xmm/m128,xmm */ @@ -7092,7 +7085,6 @@ x86_emulate( /* vpsubd {x,y}mm/mem,{x,y}mm,{x,y}mm */ case X86EMUL_OPC_66(0x0f, 0xfb): /* psubq xmm/m128,xmm */ case X86EMUL_OPC_VEX_66(0x0f, 0xfb): /* vpsubq {x,y}mm/mem,{x,y}mm,{x,y}mm */ -#endif /* !X86EMUL_NO_SIMD */ CASE_SIMD_PACKED_INT_VEX(0x0f, 0xfc): /* paddb {,x}mm/mem,{,x}mm */ /* vpaddb {x,y}mm/mem,{x,y}mm,{x,y}mm */ CASE_SIMD_PACKED_INT_VEX(0x0f, 0xfd): /* paddw {,x}mm/mem,{,x}mm */ @@ -7100,7 +7092,6 @@ x86_emulate( CASE_SIMD_PACKED_INT_VEX(0x0f, 0xfe): /* paddd {,x}mm/mem,{,x}mm */ /* vpaddd {x,y}mm/mem,{x,y}mm,{x,y}mm */ simd_0f_int: -#ifndef X86EMUL_NO_SIMD if ( vex.opcx != vex_none ) { case X86EMUL_OPC_VEX_66(0x0f38, 0x00): /* vpshufb {x,y}mm/mem,{x,y}mm,{x,y}mm */ @@ -7142,14 +7133,11 @@ x86_emulate( } if ( vex.pfx ) goto simd_0f_sse2; -#endif /* !X86EMUL_NO_SIMD */ simd_0f_mmx: host_and_vcpu_must_have(mmx); get_fpu(X86EMUL_FPU_mmx); goto simd_0f_common; -#ifndef X86EMUL_NO_SIMD - case X86EMUL_OPC_EVEX_66(0x0f, 0xf6): /* vpsadbw [xyz]mm/mem,[xyz]mm,[xyz]mm */ generate_exception_if(evex.opmsk, EXC_UD); /* fall through */ @@ -7243,8 +7231,6 @@ x86_emulate( generate_exception_if(!evex.w, EXC_UD); goto avx512f_no_sae; -#endif /* X86EMUL_NO_SIMD */ - CASE_SIMD_PACKED_INT_VEX(0x0f, 0x6e): /* mov{d,q} r/m,{,x}mm */ /* vmov{d,q} r/m,xmm */ CASE_SIMD_PACKED_INT_VEX(0x0f, 0x7e): /* mov{d,q} {,x}mm,r/m */ @@ -7286,8 +7272,6 @@ x86_emulate( ASSERT(!state->simd_size); break; -#ifndef X86EMUL_NO_SIMD - case X86EMUL_OPC_EVEX_66(0x0f, 0x6e): /* vmov{d,q} r/m,xmm */ case X86EMUL_OPC_EVEX_66(0x0f, 0x7e): /* vmov{d,q} xmm,r/m */ generate_exception_if((evex.lr || evex.opmsk || evex.brs || @@ -7360,15 +7344,11 @@ x86_emulate( d |= TwoOp; /* fall through */ case X86EMUL_OPC_66(0x0f, 0xd6): /* movq xmm,xmm/m64 */ -#endif /* !X86EMUL_NO_SIMD */ -#ifndef X86EMUL_NO_MMX case X86EMUL_OPC(0x0f, 0x6f): /* movq mm/m64,mm */ case X86EMUL_OPC(0x0f, 0x7f): /* movq mm,mm/m64 */ -#endif op_bytes = 8; goto simd_0f_int; -#ifndef X86EMUL_NO_SIMD CASE_SIMD_PACKED_INT_VEX(0x0f, 0x70):/* pshuf{w,d} $imm8,{,x}mm/mem,{,x}mm */ /* vpshufd $imm8,{x,y}mm/mem,{x,y}mm */ case X86EMUL_OPC_F3(0x0f, 0x70): /* pshufhw $imm8,xmm/m128,xmm */ @@ -7377,15 +7357,12 @@ x86_emulate( case X86EMUL_OPC_VEX_F2(0x0f, 0x70): /* vpshuflw $imm8,{x,y}mm/mem,{x,y}mm */ d = (d & ~SrcMask) | SrcMem | TwoOp; op_bytes = vex.pfx ? 16 << vex.l : 8; -#endif simd_0f_int_imm8: if ( vex.opcx != vex_none ) { -#ifndef X86EMUL_NO_SIMD case X86EMUL_OPC_VEX_66(0x0f3a, 0x0e): /* vpblendw $imm8,{x,y}mm/mem,{x,y}mm,{x,y}mm */ case X86EMUL_OPC_VEX_66(0x0f3a, 0x0f): /* vpalignr $imm8,{x,y}mm/mem,{x,y}mm,{x,y}mm */ case X86EMUL_OPC_VEX_66(0x0f3a, 0x42): /* vmpsadbw $imm8,{x,y}mm/mem,{x,y}mm,{x,y}mm */ -#endif if ( vex.l ) { simd_0f_imm8_avx2: @@ -7393,7 +7370,6 @@ x86_emulate( } else { -#ifndef X86EMUL_NO_SIMD case X86EMUL_OPC_VEX_66(0x0f3a, 0x08): /* vroundps $imm8,{x,y}mm/mem,{x,y}mm */ case X86EMUL_OPC_VEX_66(0x0f3a, 0x09): /* vroundpd $imm8,{x,y}mm/mem,{x,y}mm */ case X86EMUL_OPC_VEX_66(0x0f3a, 0x0a): /* vroundss $imm8,{x,y}mm/mem,{x,y}mm,{x,y}mm */ @@ -7401,7 +7377,6 @@ x86_emulate( case X86EMUL_OPC_VEX_66(0x0f3a, 0x0c): /* vblendps $imm8,{x,y}mm/mem,{x,y}mm,{x,y}mm */ case X86EMUL_OPC_VEX_66(0x0f3a, 0x0d): /* vblendpd $imm8,{x,y}mm/mem,{x,y}mm,{x,y}mm */ case X86EMUL_OPC_VEX_66(0x0f3a, 0x40): /* vdpps $imm8,{x,y}mm/mem,{x,y}mm,{x,y}mm */ -#endif simd_0f_imm8_avx: host_and_vcpu_must_have(avx); } @@ -7435,8 +7410,6 @@ x86_emulate( insn_bytes = PFX_BYTES + 3; break; -#ifndef X86EMUL_NO_SIMD - case X86EMUL_OPC_EVEX_66(0x0f, 0x70): /* vpshufd $imm8,[xyz]mm/mem,[xyz]mm{k} */ case X86EMUL_OPC_EVEX_F3(0x0f, 0x70): /* vpshufhw $imm8,[xyz]mm/mem,[xyz]mm{k} */ case X86EMUL_OPC_EVEX_F2(0x0f, 0x70): /* vpshuflw $imm8,[xyz]mm/mem,[xyz]mm{k} */ @@ -8320,6 +8293,8 @@ x86_emulate( } goto add; +#ifndef X86EMUL_NO_SIMD + CASE_SIMD_ALL_FP_VEX(0x0f, 0xc2): /* cmp{p,s}{s,d} $imm8,xmm/mem,xmm */ /* vcmp{p,s}{s,d} $imm8,{x,y}mm/mem,{x,y}mm,{x,y}mm */ CASE_SIMD_PACKED_FP_VEX(0x0f, 0xc6): /* shufp{s,d} $imm8,xmm/mem,xmm */ @@ -8335,8 +8310,6 @@ x86_emulate( } goto simd_0f_imm8_avx; -#ifndef X86EMUL_NO_SIMD - CASE_SIMD_ALL_FP(_EVEX, 0x0f, 0xc2): /* vcmp{p,s}{s,d} $imm8,[xyz]mm/mem,[xyz]mm,k{k} */ generate_exception_if((evex.w != (evex.pfx & VEX_PREFIX_DOUBLE_MASK) || (ea.type != OP_REG && evex.brs && @@ -8372,6 +8345,8 @@ x86_emulate( sfence = true; break; +#ifndef X86EMUL_NO_SIMD + CASE_SIMD_PACKED_INT_VEX(0x0f, 0xc4): /* pinsrw $imm8,r32/m16,{,x}mm */ /* vpinsrw $imm8,r32/m16,xmm,xmm */ generate_exception_if(vex.l, EXC_UD); @@ -8379,8 +8354,6 @@ x86_emulate( ea.type = OP_MEM; goto simd_0f_int_imm8; -#ifndef X86EMUL_NO_SIMD - case X86EMUL_OPC_EVEX_66(0x0f, 0xc4): /* vpinsrw $imm8,r32/m16,xmm,xmm */ case X86EMUL_OPC_EVEX_66(0x0f3a, 0x20): /* vpinsrb $imm8,r32/m8,xmm,xmm */ case X86EMUL_OPC_EVEX_66(0x0f3a, 0x22): /* vpinsr{d,q} $imm8,r/m,xmm,xmm */ @@ -8398,8 +8371,6 @@ x86_emulate( state->simd_size = simd_other; goto avx512f_imm8_no_sae; -#endif /* !X86EMUL_NO_SIMD */ - CASE_SIMD_PACKED_INT_VEX(0x0f, 0xc5): /* pextrw $imm8,{,x}mm,reg */ /* vpextrw $imm8,xmm,reg */ generate_exception_if(vex.l, EXC_UD); @@ -8415,8 +8386,6 @@ x86_emulate( insn_bytes = PFX_BYTES + 3; goto simd_0f_to_gpr; -#ifndef X86EMUL_NO_SIMD - CASE_SIMD_PACKED_FP(_EVEX, 0x0f, 0xc6): /* vshufp{s,d} $imm8,[xyz]mm/mem,[xyz]mm,[xyz]mm{k} */ generate_exception_if(evex.w != (evex.pfx & VEX_PREFIX_DOUBLE_MASK), EXC_UD); @@ -8706,8 +8675,6 @@ x86_emulate( op_bytes = 8 << (!!(vex.pfx & VEX_PREFIX_DOUBLE_MASK) + vex.l); goto simd_0f_cvt; -#endif /* !X86EMUL_NO_SIMD */ - CASE_SIMD_PACKED_INT_VEX(0x0f, 0xf7): /* {,v}maskmov{q,dqu} {,x}mm,{,x}mm */ generate_exception_if(ea.type != OP_REG, EXC_UD); if ( vex.opcx != vex_none ) @@ -8811,8 +8778,6 @@ x86_emulate( insn_bytes = PFX_BYTES + 3; break; -#ifndef X86EMUL_NO_SIMD - case X86EMUL_OPC_VEX_66(0x0f38, 0x19): /* vbroadcastsd xmm/m64,ymm */ case X86EMUL_OPC_VEX_66(0x0f38, 0x1a): /* vbroadcastf128 m128,ymm */ generate_exception_if(!vex.l, EXC_UD); @@ -10233,8 +10198,6 @@ x86_emulate( avx512_vlen_check(b & 2); goto simd_imm8_zmm; -#endif /* X86EMUL_NO_SIMD */ - CASE_SIMD_PACKED_INT(0x0f3a, 0x0f): /* palignr $imm8,{,x}mm/mem,{,x}mm */ host_and_vcpu_must_have(ssse3); if ( vex.pfx ) @@ -10262,8 +10225,6 @@ x86_emulate( insn_bytes = PFX_BYTES + 4; break; -#ifndef X86EMUL_NO_SIMD - case X86EMUL_OPC_EVEX_66(0x0f3a, 0x42): /* vdbpsadbw $imm8,[xyz]mm/mem,[xyz]mm,[xyz]mm{k} */ generate_exception_if(evex.w, EXC_UD); /* fall through */ @@ -11613,9 +11574,11 @@ x86_insn_is_mem_access(const struct x86_emulate_state *state, case 0xa4 ... 0xa7: /* MOVS / CMPS */ case 0xaa ... 0xaf: /* STOS / LODS / SCAS */ case 0xd7: /* XLAT */ +#ifndef X86EMUL_NO_SIMD CASE_SIMD_PACKED_INT_VEX(0x0f, 0xf7): /* MASKMOV{Q,DQU} */ /* VMASKMOVDQU */ return true; +#endif case X86EMUL_OPC(0x0f, 0x01): /* Cover CLZERO. */
