[PATCH V3] Support APX CFCMOV

[Jan Beulich]

On 04.07.2024 03:40, Cui, Lili wrote:
>> On 01.07.2024 15:29, Cui, Lili wrote:
>>> Changes in V3
>>> 1. Renamed CFCMOV_Fixup_op to CFCMOV_Fixup and refined the function.
>>> 2. Added all cc formats of cfcmove/cmove in SDM order and optimize test
>> cases.
>>
>> What is "optimize" about here? You don't do any optimization anywhere afaics?
>>
> 
> The "optimize" here refers to the following changes.
> 
> Use 
> 
> 	.irp m, be, l, le, nb, nbe, nl, nle, no, np, ns, nz, o, p, s, z, ae, e, ne, a, e, g
> 	cfcmov\m	%dx,%ax
> 	cmov\m	%dx,%ax,%r31w
> 	.endr
> 
> Instead of
> 
> +	.irp m, cfcmovbe, cfcmovl, cfcmovle, cfcmovnb, cfcmovnbe, cfcmovnl, cfcmovnle, cfcmovno, cfcmovnp, cfcmovns, cfcmovnz, cfcmovo, cfcmovp, cfcmovs, cfcmovz, cfcmovae, cfcmove, cfcmovne, cfcmova, cfcmovge, cfcmovg
> +	\m	%dx,%ax
> +	.endr
> 
> +            .irp m, cmovbe, cmovl, cmovle, cmovnb, cmovnbe, cmovnl, cmovnle, cmovno, cmovnp, cmovns, cmovnz, cmovo, cmovp, cmovs, cmovz, cmovae, cmove, cmovne, cmova, cmovge, cmovg
> +	\m	%dx,%ax,%r31w
> +	.endr

Oh, I see. Please go ahead then.

Jan