All of lore.kernel.org
 help / color / mirror / Atom feed
From: Vlad Zahorodnii <vlad.zahorodnii@kde.org>
To: "Kazlauskas, Nicholas" <nicholas.kazlauskas@amd.com>,
	amd-gfx@lists.freedesktop.org
Cc: alexander.deucher@amd.com, contact@emersion.fr,
	Rodrigo.Siqueira@amd.com, xaver.hugl@gmail.com,
	christian.koenig@amd.com
Subject: Re: [PATCH v2] drm/amd/display: Use oriented source size when checking cursor scaling
Date: Thu, 2 Dec 2021 19:47:34 +0200	[thread overview]
Message-ID: <2b8890fe-febe-d527-7da9-ae0a2c23d776@kde.org> (raw)
In-Reply-To: <c2560a82-a4c6-6913-dc67-d11478f46afd@amd.com>

On 12/2/21 16:54, Kazlauskas, Nicholas wrote:
> On 2021-12-02 7:52 a.m., Vlad Zahorodnii wrote:
>> dm_check_crtc_cursor() doesn't take into account plane transforms when
>> calculating plane scaling, this can result in false positives.
>>
>> For example, if there's an output with resolution 3840x2160 and the
>> output is rotated 90 degrees, CRTC_W and CRTC_H will be 3840 and 2160,
>> respectively, but SRC_W and SRC_H will be 2160 and 3840, respectively.
>>
>> Since the cursor plane usually has a square buffer attached to it, the
>> dm_check_crtc_cursor() will think that there's a scale factor mismatch
>> even though there isn't really.
>>
>> This fixes an issue where kwin fails to use hardware plane transforms.
>>
>> Changes since version 1:
>> - s/orientated/oriented/g
>>
>> Signed-off-by: Vlad Zahorodnii <vlad.zahorodnii@kde.org>
> 
> This looks correct to me. I guess it's also not modifying the actual 
> programming position, just the check to ensure that the cursor is going 
> to be unscaled in the correct orientation.
> 
> Would be good to have some IGT tests for these scaled cases to verify 
> atomic check pass/fail assumptions, but for now:

I'd be glad to add the tests, but I would need some assistance. Can we 
continue this conversation in the radeon irc channel (I'm zzag)?

> Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
> 
> Regards,
> Nicholas Kazlauskas
> 
>> ---
>>   .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 35 ++++++++++++++-----
>>   1 file changed, 27 insertions(+), 8 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c 
>> b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>> index a3c0f2e4f4c1..c009c668fbe2 100644
>> --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>> +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>> @@ -10736,6 +10736,24 @@ static int dm_update_plane_state(struct dc *dc,
>>       return ret;
>>   }
>> +static void dm_get_oriented_plane_size(struct drm_plane_state 
>> *plane_state,
>> +                       int *src_w, int *src_h)
>> +{
>> +    switch (plane_state->rotation & DRM_MODE_ROTATE_MASK) {
>> +    case DRM_MODE_ROTATE_90:
>> +    case DRM_MODE_ROTATE_270:
>> +        *src_w = plane_state->src_h >> 16;
>> +        *src_h = plane_state->src_w >> 16;
>> +        break;
>> +    case DRM_MODE_ROTATE_0:
>> +    case DRM_MODE_ROTATE_180:
>> +    default:
>> +        *src_w = plane_state->src_w >> 16;
>> +        *src_h = plane_state->src_h >> 16;
>> +        break;
>> +    }
>> +}
>> +
>>   static int dm_check_crtc_cursor(struct drm_atomic_state *state,
>>                   struct drm_crtc *crtc,
>>                   struct drm_crtc_state *new_crtc_state)
>> @@ -10744,6 +10762,8 @@ static int dm_check_crtc_cursor(struct 
>> drm_atomic_state *state,
>>       struct drm_plane_state *new_cursor_state, *new_underlying_state;
>>       int i;
>>       int cursor_scale_w, cursor_scale_h, underlying_scale_w, 
>> underlying_scale_h;
>> +    int cursor_src_w, cursor_src_h;
>> +    int underlying_src_w, underlying_src_h;
>>       /* On DCE and DCN there is no dedicated hardware cursor plane. 
>> We get a
>>        * cursor per pipe but it's going to inherit the scaling and
>> @@ -10755,10 +10775,9 @@ static int dm_check_crtc_cursor(struct 
>> drm_atomic_state *state,
>>           return 0;
>>       }
>> -    cursor_scale_w = new_cursor_state->crtc_w * 1000 /
>> -             (new_cursor_state->src_w >> 16);
>> -    cursor_scale_h = new_cursor_state->crtc_h * 1000 /
>> -             (new_cursor_state->src_h >> 16);
>> +    dm_get_oriented_plane_size(new_cursor_state, &cursor_src_w, 
>> &cursor_src_h);
>> +    cursor_scale_w = new_cursor_state->crtc_w * 1000 / cursor_src_w;
>> +    cursor_scale_h = new_cursor_state->crtc_h * 1000 / cursor_src_h;
>>       for_each_new_plane_in_state_reverse(state, underlying, 
>> new_underlying_state, i) {
>>           /* Narrow down to non-cursor planes on the same CRTC as the 
>> cursor */
>> @@ -10769,10 +10788,10 @@ static int dm_check_crtc_cursor(struct 
>> drm_atomic_state *state,
>>           if (!new_underlying_state->fb)
>>               continue;
>> -        underlying_scale_w = new_underlying_state->crtc_w * 1000 /
>> -                     (new_underlying_state->src_w >> 16);
>> -        underlying_scale_h = new_underlying_state->crtc_h * 1000 /
>> -                     (new_underlying_state->src_h >> 16);
>> +        dm_get_oriented_plane_size(new_underlying_state,
>> +                       &underlying_src_w, &underlying_src_h);
>> +        underlying_scale_w = new_underlying_state->crtc_w * 1000 / 
>> underlying_src_w;
>> +        underlying_scale_h = new_underlying_state->crtc_h * 1000 / 
>> underlying_src_h;
>>           if (cursor_scale_w != underlying_scale_w ||
>>               cursor_scale_h != underlying_scale_h) {
>>
> 


      parent reply	other threads:[~2021-12-02 17:47 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-12-02 12:16 [PATCH] drm/amd/display: Use orientated source size when checking cursor scaling Vlad Zahorodnii
2021-12-02 12:21 ` Simon Ser
2021-12-02 12:23   ` Vlad Zahorodnii
2021-12-02 12:52 ` [PATCH v2] drm/amd/display: Use oriented " Vlad Zahorodnii
2021-12-02 14:54   ` Kazlauskas, Nicholas
2021-12-02 17:11     ` Alex Deucher
2021-12-02 17:47     ` Vlad Zahorodnii [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=2b8890fe-febe-d527-7da9-ae0a2c23d776@kde.org \
    --to=vlad.zahorodnii@kde.org \
    --cc=Rodrigo.Siqueira@amd.com \
    --cc=alexander.deucher@amd.com \
    --cc=amd-gfx@lists.freedesktop.org \
    --cc=christian.koenig@amd.com \
    --cc=contact@emersion.fr \
    --cc=nicholas.kazlauskas@amd.com \
    --cc=xaver.hugl@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.