Re: [PATCH 2/2] x86: extend insn decoder to understand xop and evex prefixes

From: Masami Hiramatsu
Date: Fri Sep 05 2014 - 08:42:03 EST

Next message: Johannes Weiner: "[patch] mm: memcontrol: revert use of root_mem_cgroup res_counter"
Previous message: Andy Shevchenko: "Re: [PATCH v4 01/12] hexdump: introduce test suite"
In reply to: Denys Vlasenko: "[PATCH 2/2] x86: extend insn decoder to understand xop and evex prefixes"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Hi Denys,

(2014/09/04 21:54), Denys Vlasenko wrote:
> Since xop and evex prefixes are extensions of vex mechanism,
> they have similar bit layouts, and they can never be combined
> (an instruction can have only one of them),
> (ab)use insn->vex_prefix to store data of xop and evex too.
>
> Users will need to conditionalize on insn->vex_prefix.bytes[0]
> instead of insn->vex_prefix.nbytes if they want to determine
> which of vex(-like) prefixes are there.

Hmm, this is just a basic support of AVX512, since this doesn't
recognize the instructions which do not support evex prefixes.
We'd better to check Intel's new SDM and clarify which instructions
do really support it or not.

Thank you,

>
> v2: on Masami's request, retained use of inattr bits
> for determining prefixes.
>
> Signed-off-by: Denys Vlasenko <dvlasenk@xxxxxxxxxx>
> Cc: Masami Hiramatsu <masami.hiramatsu.pt@xxxxxxxxxxx>
> Cc: Frank Ch. Eigler <fche@xxxxxxxxxx>
> Cc: Srikar Dronamraju <srikar@xxxxxxxxxxxxxxxxxx>
> Cc: Ananth N Mavinakayanahalli <ananth@xxxxxxxxxx>
> Cc: Jim Keniston <jkenisto@xxxxxxxxxxxxxxxxxx>
> Cc: Oleg Nesterov <oleg@xxxxxxxxxx>
> Cc: Andi Kleen <andi@xxxxxxxxxxxxxx>
> Cc: Ingo Molnar <mingo@xxxxxxxxxx>
> ---
> arch/x86/include/asm/inat.h | 26 +++++++++++++----
> arch/x86/lib/insn.c | 54 ++++++++++++++++++++++++------------
> arch/x86/lib/x86-opcode-map.txt | 4 +--
> arch/x86/tools/gen-insn-attr-x86.awk | 2 ++
> 4 files changed, 62 insertions(+), 24 deletions(-)
>
> diff --git a/arch/x86/include/asm/inat.h b/arch/x86/include/asm/inat.h
> index 74a2e31..2c7b4cc 100644
> --- a/arch/x86/include/asm/inat.h
> +++ b/arch/x86/include/asm/inat.h
> @@ -48,6 +48,8 @@
> /* AVX VEX prefixes */
> #define INAT_PFX_VEX2 13 /* 2-bytes VEX prefix */
> #define INAT_PFX_VEX3 14 /* 3-bytes VEX prefix */
> +#define INAT_PFX_XOP 15 /* 3-bytes XOP prefix */
> +#define INAT_PFX_EVEX 16 /* 4-bytes EVEX prefix */
>
> #define INAT_LSTPFX_MAX 3
> #define INAT_LGCPFX_MAX 11
> @@ -63,7 +65,7 @@
>
> /* Legacy prefix */
> #define INAT_PFX_OFFS 0
> -#define INAT_PFX_BITS 4
> +#define INAT_PFX_BITS 5
> #define INAT_PFX_MAX ((1 << INAT_PFX_BITS) - 1)
> #define INAT_PFX_MASK (INAT_PFX_MAX << INAT_PFX_OFFS)
> /* Escape opcodes */
> @@ -138,15 +140,29 @@ static inline int inat_last_prefix_id(insn_attr_t attr)
> return attr & INAT_PFX_MASK;
> }
>
> +static inline int inat_is_vex3_prefix(insn_attr_t attr)
> +{
> + return attr & INAT_PFX_VEX3;
> +}
> +
> +static inline int inat_is_vex2_prefix(insn_attr_t attr)
> +{
> + return attr & INAT_PFX_VEX2;
> +}
> +
> +static inline int inat_is_xop_prefix(insn_attr_t attr)
> +{
> + return attr & INAT_PFX_XOP;
> +}
> +
> static inline int inat_is_vex_prefix(insn_attr_t attr)
> {
> - attr &= INAT_PFX_MASK;
> - return attr == INAT_PFX_VEX2 || attr == INAT_PFX_VEX3;
> + return attr & (INAT_PFX_VEX2 | INAT_PFX_VEX3);
> }
>
> -static inline int inat_is_vex3_prefix(insn_attr_t attr)
> +static inline int inat_is_vex_like_prefix(insn_attr_t attr)
> {
> - return (attr & INAT_PFX_MASK) == INAT_PFX_VEX3;
> + return attr & (INAT_PFX_VEX2 | INAT_PFX_VEX3 | INAT_PFX_XOP | INAT_PFX_EVEX);
> }
>
> static inline int inat_is_escape(insn_attr_t attr)
> diff --git a/arch/x86/lib/insn.c b/arch/x86/lib/insn.c
> index 829ca4c..def44c8 100644
> --- a/arch/x86/lib/insn.c
> +++ b/arch/x86/lib/insn.c
> @@ -138,40 +138,60 @@ found:
> }
> insn->rex_prefix.got = 1;
>
> - /* Decode VEX prefix */
> + /* Decode VEX prefixes et al. Layouts are:
> + * vex2: c5 rvvvvLpp
> + * vex3/xop: c4/8f rxbmmmmm wvvvvLpp
> + * evex: 62 rxbR00mm wvvvv1pp zllBVaaa
> + */
> b = peek_next(insn_byte_t, insn);
> attr = inat_get_opcode_attribute(b);
> - if (inat_is_vex_prefix(attr)) {
> - insn_byte_t b2 = peek_nbyte_next(insn_byte_t, insn, 1);
> - if (!insn->x86_64) {
> + if (inat_is_vex_like_prefix(attr)) {
> + insn_byte_t b2;
> +
> + b2 = peek_nbyte_next(insn_byte_t, insn, 1);
> + if (inat_is_xop_prefix(attr)) {
> /*
> - * In 32-bits mode, if the [7:6] bits (mod bits of
> - * ModRM) on the second byte are not 11b, it is
> - * LDS or LES.
> + * XOP: If modrm.reg bits are 000, it's POP reg/mem.
> + */
> + if (X86_MODRM_REG(b2) == 0)
> + goto vex_end;
> + }
> + else if (!insn->x86_64) {
> + /*
> + * [E]VEX: In 32-bits mode, if modrm.mod bits
> + * are not 11b, it is LDS, LES or BOUND.
> */
> if (X86_MODRM_MOD(b2) != 3)
> goto vex_end;
> }
> insn->vex_prefix.bytes[0] = b;
> insn->vex_prefix.bytes[1] = b2;
> - if (inat_is_vex3_prefix(attr)) {
> - b2 = peek_nbyte_next(insn_byte_t, insn, 2);
> - insn->vex_prefix.bytes[2] = b2;
> - insn->vex_prefix.nbytes = 3;
> - insn->next_byte += 3;
> - if (insn->x86_64 && X86_VEX_W(b2))
> - /* VEX.W overrides opnd_size */
> - insn->opnd_bytes = 8;
> - } else {
> + if (inat_is_vex2_prefix(attr)) {
> /*
> * For VEX2, fake VEX3-like byte#2.
> * Makes it easier to decode vex.W, vex.vvvv,
> - * vex.L and vex.pp. Masking with 0x7f sets vex.W == 0.
> + * vex.L and vex.pp. Masking with 0x7f sets vex.W = 0.
> */
> insn->vex_prefix.bytes[2] = b2 & 0x7f;
> insn->vex_prefix.nbytes = 2;
> insn->next_byte += 2;
> + goto vex_end;
> + }
> + b2 = peek_nbyte_next(insn_byte_t, insn, 2);
> + insn->vex_prefix.bytes[2] = b2;
> + if (insn->x86_64 && X86_VEX_W(b2)) {
> + /* VEX.W overrides opnd_size */
> + insn->opnd_bytes = 8;
> + }
> + if (inat_is_vex3_prefix(attr) || inat_is_xop_prefix(attr)) {
> + insn->vex_prefix.nbytes = 3;
> + insn->next_byte += 3;
> + goto vex_end;
> }
> + b2 = peek_nbyte_next(insn_byte_t, insn, 3);
> + insn->vex_prefix.bytes[3] = b2;
> + insn->vex_prefix.nbytes = 4;
> + insn->next_byte += 4;
> }
> vex_end:
> insn->vex_prefix.got = 1;
> diff --git a/arch/x86/lib/x86-opcode-map.txt b/arch/x86/lib/x86-opcode-map.txt
> index 1a2be7c..9d8a964 100644
> --- a/arch/x86/lib/x86-opcode-map.txt
> +++ b/arch/x86/lib/x86-opcode-map.txt
> @@ -137,7 +137,7 @@ AVXcode:
> # 0x60 - 0x6f
> 60: PUSHA/PUSHAD (i64)
> 61: POPA/POPAD (i64)
> -62: BOUND Gv,Ma (i64)
> +62: BOUND Gv,Ma (i64) | EVEX+3byte (Prefix)
> 63: ARPL Ew,Gw (i64) | MOVSXD Gv,Ev (o64)
> 64: SEG=FS (Prefix)
> 65: SEG=GS (Prefix)
> @@ -184,7 +184,7 @@ AVXcode:
> 8c: MOV Ev,Sw
> 8d: LEA Gv,M
> 8e: MOV Sw,Ew
> -8f: Grp1A (1A) | POP Ev (d64)
> +8f: Grp1A (1A) | POP Ev (d64) | XOP+2byte (Prefix)
> # 0x90 - 0x9f
> 90: NOP | PAUSE (F3) | XCHG r8,rAX
> 91: XCHG rCX/r9,rAX
> diff --git a/arch/x86/tools/gen-insn-attr-x86.awk b/arch/x86/tools/gen-insn-attr-x86.awk
> index 093a892..d288754 100644
> --- a/arch/x86/tools/gen-insn-attr-x86.awk
> +++ b/arch/x86/tools/gen-insn-attr-x86.awk
> @@ -95,6 +95,8 @@ BEGIN {
> prefix_num["Address-Size"] = "INAT_PFX_ADDRSZ"
> prefix_num["VEX+1byte"] = "INAT_PFX_VEX2"
> prefix_num["VEX+2byte"] = "INAT_PFX_VEX3"
> + prefix_num["XOP+2byte"] = "INAT_PFX_XOP"
> + prefix_num["EVEX+3byte"] = "INAT_PFX_EVEX"
>
> clear_vars()
> }
>

--
Masami HIRAMATSU
Software Platform Research Dept. Linux Technology Research Center
Hitachi, Ltd., Yokohama Research Laboratory
E-mail: masami.hiramatsu.pt@xxxxxxxxxxx

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Next message: Johannes Weiner: "[patch] mm: memcontrol: revert use of root_mem_cgroup res_counter"
Previous message: Andy Shevchenko: "Re: [PATCH v4 01/12] hexdump: introduce test suite"
In reply to: Denys Vlasenko: "[PATCH 2/2] x86: extend insn decoder to understand xop and evex prefixes"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]