Re: [igt-dev] [PATCH i-g-t 2/3] lib/igt_fb: include lib/intel_blt functions to blitter path

[Karolina Stolarek <karolina.stolarek@intel.com>]

On 29.06.2023 18:04, Juha-Pekka Heikkila wrote:
<big snip>

>>>>> +            blt_set_copy_object(&blt.src, src);
>>>>> +            blt_set_copy_object(&blt.dst, dst);
>>>>> +
>>>>> +            if (HAS_FLATCCS(intel_get_drm_devid(src_fb->fd))) {
>>>>
>>>> This check might not be correct on some platforms. To check if we 
>>>> need to fill in ext part, you could do a similar check as in commit 
>>>> 45da871dd268 ("tests/gem_ccs: Check for extended block-copy and 
>>>> compression support").
>>>>
>>>> +Zbigniew, what is the status of HAS_FLATCCS? How is it useful in 
>>>> the context of blitter copy? I believe we want to switch to using 
>>>> command flags, like BLT_CMD_EXTENDED? Or am I misremembering something?
>>>>
>>>
>>> This is bit double edged situation. Even if for MTL there would be 
>>> added support for compression into intel_blt, all the related 
>>> handling for aux ccs would still not be here on igt_fb.c. For flat 
>>> ccs everything that is needed is now in place and for time being it 
>>> seems correct, everything matches. Do note this in not only 
>>> controlled by HAS_FLATCCS() but also by modifiers which actually 
>>> arrive onto this part of code.
>>
>> Hmm, I see, let's leave it for now and we can revisit it in the future.
>>
>> Thanks for answering my questions. Like I said, there are two nits 
>> that would be good to tidy up, but everything else looks good to me:
>>
>> Reviewed-by: Karolina Stolarek <karolina.stolarek@intel.com>
> 
> Thanks Karolina. In the mean time this patch was separated into two 
> parts, one where there were couple of more Xe related changes and other 
> where these blitting related changes. May I use this R-b here
> 
> https://patchwork.freedesktop.org/patch/544642/?series=119886&rev=4

Yes, please do

> ?
> 
> Those few alignment fixes and small adjustments I'll apply before 
> pushing if I will not need another ci round.

I think you're good to go, the reported regressions seem unrelated to 
your patches.

Many thanks,
Karolina

> 
> /Juha-Pekka