Re: [PATCH] drm/panel: chipwealth-ch13726a: Fix washed out colors on 120hz mode

[Neil Armstrong]

On 7/28/26 03:38, Aaron Kling wrote:
> On Mon, Jul 27, 2026 at 7:44 PM Aaron Kling via B4 Relay
> <devnull+webgeek1234.gmail.com@xxxxxxxxxx> wrote:
> > From: Aaron Kling <webgeek1234@xxxxxxxxx>
> >
> > The issue was determined to be two-fold:
> >
> > 1. The non-continuous clock dropping into LPM during blanks causes
> >     voltage drift at 120hz enough to affect the gamma curve.
> >
> > 2. The agressive horizontal blanking width did not allow enough settle
> >     time, further compounding the voltage drift.
> >
> > In addition to switching to continuous clock, burst mode is also enabled
> > to further stabilize the blank intervals.
> >
> > Assisted-by: Gemini:2.5-Pro
> > Signed-off-by: Aaron Kling <webgeek1234@xxxxxxxxx>
> > ---
> > For the initial submission, the washed out colors were expected to need
> > to be fixed by a color profile. However, further investigation showed
> > there was a problem with the timings.
> > ---
> >   drivers/gpu/drm/panel/panel-chipwealth-ch13726a.c | 10 +++++-----
> >   1 file changed, 5 insertions(+), 5 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/panel/panel-chipwealth-ch13726a.c b/drivers/gpu/drm/panel/panel-chipwealth-ch13726a.c
> > index 562dc573528dd..92f0c4f1587cb 100644
> > --- a/drivers/gpu/drm/panel/panel-chipwealth-ch13726a.c
> > +++ b/drivers/gpu/drm/panel/panel-chipwealth-ch13726a.c
> > @@ -128,11 +128,11 @@ static int ch13726a_unprepare(struct drm_panel *panel)
> >   static const struct drm_display_mode thor_bottom_modes[] = {
> >          {
> >                  /* 120Hz */
> > -               .clock = (1080 + 28 + 4 + 36) * (1240 + 16 + 4 + 8) * 120 / 1000,
> > +               .clock = (1080 + 60 + 8 + 60) * (1240 + 16 + 4 + 8) * 120 / 1000,
> >                  .hdisplay = 1080,
> > -               .hsync_start = 1080 + 28,
> > -               .hsync_end = 1080 + 28 + 4,
> > -               .htotal = 1080 + 28 + 4 + 36,
> > +               .hsync_start = 1080 + 60,
> > +               .hsync_end = 1080 + 28 + 8,
> > +               .htotal = 1080 + 28 + 4 + 60,
> >                  .vdisplay = 1240,
> >                  .vsync_start = 1240 + 16,
> >                  .vsync_end = 1240 + 16 + 4,
> > @@ -281,7 +281,7 @@ static int ch13726a_probe(struct mipi_dsi_device *dsi)
> >          dsi->lanes = 4;
> >          dsi->format = MIPI_DSI_FMT_RGB888;
> >          dsi->mode_flags = MIPI_DSI_MODE_VIDEO |
> > -                         MIPI_DSI_CLOCK_NON_CONTINUOUS;
> > +                         MIPI_DSI_MODE_VIDEO_BURST;
> >
> >          ctx->panel.prepare_prev_first = true;
> >
> >
> > ---
> > base-commit: 0d33d21e47d9dc66f91e44da3fc9220c74d93df7
> > change-id: 20260727-ch13726a-washout-54ef8fa65be6
> >
> > Best regards,
> > --
> > Aaron Kling <webgeek1234@xxxxxxxxx>
> The sashiko report that the timings in this change are inconsistent is
> correct. Which caused the 120hz mode to be discarded entirely, so the
> 60hz mode was being used which has correct colors. Fixing the timings
> to be consistent restores the 120hz mode, but still with incorrect
> colors. So this patch should be abandoned while further investigations
> continue. Sorry for the churn.

OK, please add a Fixes tag when you find a proper fix.

Thanks,
Neil

> Aaron