This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

RE: [PATCH] Match x86 family machine constraints section with constarints.md


Gentle ping.

Thanks,
Sebastian

-----Original Message-----
From: Peryt, Sebastian 
Sent: Friday, April 28, 2017 11:31 AM
To: Sandra Loosemore <sandra@codesourcery.com>; gcc-patches@gcc.gnu.org
Cc: ubizjak@gmail.com; Koval, Julia <julia.koval@intel.com>; kirill.yukhin@gmail.com
Subject: RE: [PATCH] Match x86 family machine constraints section with constarints.md

Hi,

Thank you for your comments. I edited my patch accordingly. As for some of your doubts:
- REX is  the opcode prefix to access 64-bit register extensions introduced in IA-32e mode.
- EVEX is the encoding prefix which applies to SIMD operating instructions operating on XMM, YMM and ZMM registers. It was introduced with AVX-512 instructions.
- "number factor of four" that means that sources start in a multiple of 4 boundary. This is used for some of instructions.

Also I'd like to add that this whole patch is strictly based on docstring parts of constraints that are present in config/i386/constraints.md but not in documentation (md.texi file). There is no new (new as in nonexistent in code) content.

I'm also adding Kirill Yukhin to CC, because I believe he is the correct person that can catch any technical errors if any has slipped-in.

Thanks,
Sebastian

-----Original Message-----
From: Sandra Loosemore [mailto:sandra@codesourcery.com]
Sent: Thursday, April 27, 2017 10:17 PM
To: Peryt, Sebastian <sebastian.peryt@intel.com>; gcc-patches@gcc.gnu.org
Cc: ubizjak@gmail.com; Koval, Julia <julia.koval@intel.com>
Subject: Re: [PATCH] Match x86 family machine constraints section with constarints.md

On 04/26/2017 08:29 AM, Peryt, Sebastian wrote:
> Hi,
>
> This patch updates x86 family machine constraints section in '16.8.5 Constraints for Particular Machines' section to match the ones in 'config/i386/constraints.md'.
>
> gcc/
> 	* doc/md.texi (Machine Constraints): Update x86 family machine constraints
> 	   section to match 'config/i386/constraints.md'.
>
> Is it ok for trunk?

I have a few comments on grammar and markup, but I can't comment intelligently on whether the technical content is correct.

> @@ -4013,24 +4015,94 @@ Top of 80387 floating-point stack (@code{%st(0)}).
>  @item u
>  Second from top of 80387 floating-point stack (@code{%st(1)}).
>
> +@ifset INTERNALS
> +@item Yk
> +Any mask register that can be used as predicate, i.e. k1-k7.

s/predicate/a predicate/

Other places in this section use @code markup on literal register names.

> +
> +@item k
> +Any mask register.
> +@end ifset
> +
>  @item y
>  Any MMX register.
>
>  @item x
>  Any SSE register.
>
> +@item v
> +Any EVEX encodable SSE register (@code{%xmm0-%xmm31}).
> +
> +@ifset INTERNALS
> +@item w
> +Any bound register.
> +@end ifset
> +
>  @item Yz
>  First SSE register (@code{%xmm0}).
>
>  @ifset INTERNALS
> -@item Y2
> -Any SSE register, when SSE2 is enabled.
> -
>  @item Yi
>  Any SSE register, when SSE2 and inter-unit moves are enabled.
>
> +@item Yj
> +Any SSE register, when SSE2 and inter-unit moves from vector registers are enabled.
> +
>  @item Ym
>  Any MMX register, when inter-unit moves are enabled.
> +
> +@item Yn
> +Any MMX register, when inter-unit moves from vector registers are enabled.
> +
> +@item Yp
> +Any integer register when TARGET_PARTIAL_REG_STALL is disabled.

@code markup on that.

> +
> +@item Ya
> +Any integer register when zero extensions with AND are disabled.

I'm not sure what "AND" is, but it probably needs @code markup too.
> +
> +@item Yb
> +Any register that can be used as the GOT base when calling ___tls_get_addr:

@code{___tls_get_addr}

> +that is, any general register except @code{a} and @code{sp} 
> +registers, for -fno-plt if linker supports it. Otherwise, @code{b} register.

@option{-fno-plt}

> +
> +@item Yf
> +Any x87 register when 80387 FP arithmetic is enabled.

Is "FP" a literal feature name used in the processor documentation, or do you just mean "floating-point arithmetic" here?

> +
> +@item Yr
> +Lower SSE register when avoiding REX prefix and all SSE registers otherwise.

I don't know what "avoiding REX prefix" means, and don't see the string "REX" in any other GCC documentation.

> +
> +@item Yv
> +For AVX512VL, any EVEX encodable SSE register (@code{%xmm0-%xmm31}), 
> +otherwise any SSE register.

This should probably be "EVEX-encodable", whatever that means.

> +
> +@item Yh
> +Any EVEX encodable SSE register, which has number factor of four.

Same here, but what is "number factor of four"?  Also, if this is supposed to designate a subset of the EVEX-encodable SSE registers rather than describe all of them, you need "that" instead of "which".

> +
> +@item Bf
> +Flags register operand.
> +
> +@item Bg
> +GOT memory operand.
> +
> +@item Bm
> +Vector memory operand.
> +
> +@item Bc
> +Constant memory operand.
> +
> +@item Bn
> +Memory operand without REX prefix.
> +
> +@item Bs
> +Sibcall memory operand.
> +
> +@item Bw
> +Call memory operand.
> +
> +@item Bz
> +Constant call address operand.
> +
> +@item BC
> +SSE constant -1 operand.
>  @end ifset
>
>  @item I
> @@ -4068,11 +4140,37 @@ SSE constant zero operand.
>  to fit that range (for immediate operands in sign-extending x86-64 
> instructions).
>
> +@item We
> +32-bit signed integer constant, or a symbolic reference known to fit 
> +that range (for sign-extending conversion operations that require 
> +non-VOIDmode immediate operands).

@code{VOIDmode}.
> +
> +@item Wz
> +32-bit unsigned integer constant, or a symbolic reference known to 
> +fit that range (for zero-extending conversion operations that require 
> +non-VOIDmode immediate operands).

Ditto.

> +
> +@item Wd
> +128-bit integer constant where both the high and low 64-bit word of 
> +it satisfies the e constraint.

...where both the high and low 64-bit words satisfy the @code{e} constraint.

> +
>  @item Z
>  32-bit unsigned integer constant, or a symbolic reference known  to 
> fit that range (for immediate operands in zero-extending x86-64 
> instructions).
>
> +@item Tv
> +VSIB address operand.
> +
> +@item Ts
> +Address operand without segment register.
> +
> +@item Ti
> +MPX address operand without index.
> +
> +@item Tb
> +MPX address operand without base.
> +
>  @end table
>
>  @item Xstormy16---@file{config/stormy16/stormy16.h}

-Sandra


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]