[PATCH 02/21] drm/msm/dsi: fix DSC width for the bonded DSI case

[Jun Nie]

Marijn Suijten <marijn.suijten at somainline.org> 于2024年9月3日周二 19:51写道：
>
> On 2024-09-03 11:50:36, Marijn Suijten wrote:
> > On 2024-08-29 18:17:31, Jun Nie wrote:
> > > From: Jonathan Marek <jonathan at marek.ca>
> > >
> > > For the bonded DSI case, DSC pic_width and timing calculations should use
> > > the width of a single panel instead of the total combined width.
> >
> > When this patch was originally proposed we already discussed [1] that this is
> > **not** universally true.  On my hardware a single bonded panel always receives
> > the full width, at least on downstream kernels, and it works [2].
> >
> > [1]: https://lore.kernel.org/linux-arm-msm/eanx45rnasj7lu3r2tfhtg4qkqkcidd6zctsz6ci6jlklu4fgi@3nf73w2ka4li/T/#u
> > [2]: https://gitlab.freedesktop.org/drm/msm/-/issues/41
> >
> > Can we please figure this out before landing this patch?
>
> For completeness I've picked this patch, together with the following
> mis-squashed change from patch 03/21:
>
>         diff --git a/drivers/gpu/drm/msm/dsi/dsi_host.c b/drivers/gpu/drm/msm/dsi/dsi_host.c
>         index 7a4d9c071be5a..5abade8f26b88 100644
>         --- a/drivers/gpu/drm/msm/dsi/dsi_host.c
>         +++ b/drivers/gpu/drm/msm/dsi/dsi_host.c
>         @@ -953,7 +953,7 @@ static void dsi_timing_setup(struct msm_dsi_host *msm_host, bool is_bonded_dsi)
>                                 return;
>                         }
>
>         -               dsc->pic_width = mode->hdisplay;
>         +               dsc->pic_width = hdisplay;
>                         dsc->pic_height = mode->vdisplay;
>                         DBG("Mode %dx%d\n", dsc->pic_width, dsc->pic_height);
>
> And this is what it looks like on a bonded DSI CMD-mode display:
> https://gitlab.freedesktop.org/drm/msm/-/issues/41#note_2553207
> https://gitlab.freedesktop.org/-/project/2206/uploads/dc5c53d09ecb635fdc9f190fbc9b37ac/1000027079.jpg
>
> That's a clear regression :)
>
> - Marijn

Surely we should figure out what's going on here before we land this
patch. Please help me to understand
your setup first. Your panel works well on mainline kernel with
2:2:2(?) topology, and you see issue with applying
the change here, right?
There are several parameter that impact the issue, video/command DSI
mode, single panel with 2 DSI vs
2 panels with single DSI on each panel, 4:4:2 topology on my device vs
your topology. Let's list all of them
before further discussion.

-Jun