Re: [PATCH 04/10] Support Intel CMPccXADD

[Jan Beulich <jbeulich@suse.com>]

On 14.10.2022 23:51, H.J. Lu wrote:
> On Fri, Oct 14, 2022 at 11:27 AM H.J. Lu <hjl.tools@gmail.com> wrote:
>>  On Fri, Oct 14, 2022 at 6:46 AM Jan Beulich <jbeulich@suse.com> wrote:
>>> On 14.10.2022 11:12, Haochen Jiang wrote:
>>>> @@ -8480,6 +8609,70 @@ static const struct dis386 mod_table[][2] = {
>>>>      /* MOD_VEX_0F38B1*/
>>>>      { VEX_W_TABLE (VEX_W_0F38B1) },
>>>>    },
>>>> +  {
>>>> +    /* MOD_VEX_0F38E0_X86_64 */
>>>> +    { "cmpoxadd", { Mdq, Gdq, VexGdq }, PREFIX_DATA },
>>>> +  },
>>>> +  {
>>>> +    /* MOD_VEX_0F38E1_X86_64 */
>>>> +    { "cmpnoxadd", { Mdq, Gdq, VexGdq }, PREFIX_DATA },
>>>> +  },
>>>> +  {
>>>> +    /* MOD_VEX_0F38E2_X86_64 */
>>>> +    { "cmpbxadd", { Mdq, Gdq, VexGdq }, PREFIX_DATA },
>>>> +  },
>>>> +  {
>>>> +    /* MOD_VEX_0F38E3_X86_64 */
>>>> +    { "cmpnbxadd", { Mdq, Gdq, VexGdq }, PREFIX_DATA },
>>>
>>> I understand the ISA extensions document names the insn this way and doesn't
>>> list cmpaexadd (same for other aliases), but I think this is a mistake in
>>
>> Lack of aliases is a bad thing.  In any case, assembler should follow
> 
> Oops.  I meant "Lack of aliases isn't a bad thing."   Aliases make me
> wonder if 2 different jcc are really different.

Have you taken the time to read through my question raised in the forum?
There is a reason for there being multiple names for a single condition
code: What is meant depends on context. As said there, JZ makes sense to
use with e.g. TEST or SUB, but it doesn't make sense to use with CMP
(where JE is the appropriate mnemonic). Here we're talking of only CMP
(an advanced form of it), so not being able to use CMPEXADD (for there
only being CMPZXADD) is an issue.

>> the spec.

First of all the (early) spec should be sanitized. I'd be surprised if
there were actually rational reasons for the choice of mnemonics which
the doc currently lists; the set looks rather arbitrary to me. If the
spec cannot be relied on, I think it's better to defer implementation of
something that's questionable.

Plus: What harm would there be if gas supported the full set of mnemonics
even without the spec listing them all? It's not like there's any non-
negligible risk of these mnemonics later gaining some different meaning.

Jan

>>> the doc. I've raised a respective question in the ISA extensions forum: I
>>> think representation of conditions to check for should be uniform among
>>> insns, and hence it should be "ae" here. (That would also be the effect if
>>> you used %C<whatever> here.)