[igt-dev] [PATCH 1/2] [i-g-t] tests/kms_cursor_crc: Fix tests be skipped on AMD gpu

[Wu, Hersen]

[AMD Official Use Only - General]

Hi Kamil,

I have made commit title change according to your suggestion.
There are two issues for cursor test from AMD GPU sides.

1. AMD GPU could not do cursor crop.  Patch 1/2 handle this case.
2. By AMD GPU/Driver cursor update implementation, pipe CRC could not be generated reliably within one vblank for both HW and SW cursor. This is why set data.vblank_wait_count = 2.
    For SW cursor, we need adding wait for two vblanks which are not needed for other GPU vendors.

    We still need patch 2/2 for AMD GPU.

Thanks!
Hersen


-----Original Message-----
From: Kamil Konieczny <kamil.konieczny at linux.intel.com>
Sent: Thursday, October 12, 2023 12:51 PM
To: igt-dev at lists.freedesktop.org
Cc: Wu, Hersen <hersenxs.wu at amd.com>; Siqueira, Rodrigo <Rodrigo.Siqueira at amd.com>; Pillai, Aurabindo <Aurabindo.Pillai at amd.com>; Hung, Alex <Alex.Hung at amd.com>; Mahfooz, Hamza <Hamza.Mahfooz at amd.com>; Li, Sun peng (Leo) <Sunpeng.Li at amd.com>; markyacoub at google.com; Juha-Pekka Heikkila <juhapekka.heikkila at gmail.com>
Subject: Re: [igt-dev] [PATCH 1/2] [i-g-t] tests/kms_cursor_crc: Fix tests be skipped on AMD gpu

Hi Hersen,

On 2023-10-01 at 12:28:27 -0400, Hersen Wu wrote:

please improve your patch description, instead of:

tests/kms_cursor_crc: Fix tests be skipped on AMD gpu

it is better to write why or what you fixed, for example:

tests/kms_cursor_crc: Fix cursor crop on AMD gpu

Please resend with this fix and added r-b from Juha-Pekka.

Also no need to retest your test on our CI, just reply to igt mailing list that failures are not related to AMD gpu (sometimes they could be when you change library function or you touch common logic). After that our bug filing team will mask and/or retest it.

I assume that you dropped 2/2 from this series?

Regards,
Kamil

> Remove extra line when create cursor fb for AMD gpu.
>
> AMD gpu does not support cursor buffer crop. AMD driver fails
> validation with fb size not equal to cursor size.
>
> Signed-off-by: Hersen Wu <hersenxs.wu at amd.com>
> ---
>  tests/kms_cursor_crc.c | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/tests/kms_cursor_crc.c b/tests/kms_cursor_crc.c index
> 76c01f557..ba29ff65d 100644
> --- a/tests/kms_cursor_crc.c
> +++ b/tests/kms_cursor_crc.c
> @@ -560,14 +560,18 @@ static void create_cursor_fb(data_t *data, int
> cur_w, int cur_h)  {
>       cairo_t *cr;
>       uint32_t fb_id;
> +     int cur_h_extra_line = 1;
>
> +     /* Cropping is not supported for cursor plane by AMD */
> +     if (is_amdgpu_device(data->drm_fd))
> +             cur_h_extra_line = 0;
>       /*
>        * Make the FB slightly taller and leave the extra
>        * line opaque white, so that we can see that the
>        * hardware won't scan beyond what it should (esp.
>        * with non-square cursors).
>        */
> -     fb_id = igt_create_color_fb(data->drm_fd, cur_w, cur_h + 1,
> +     fb_id = igt_create_color_fb(data->drm_fd, cur_w, cur_h +
> +cur_h_extra_line,
>                                   DRM_FORMAT_ARGB8888,
>                                   DRM_FORMAT_MOD_LINEAR,
>                                   1.0, 1.0, 1.0,
> --
> 2.25.1
>