Re: [PATCH 11/13] media: i2c: os05b10: Update active format before adjusting framing controls

[Tarang Raval]

Hi Sakari,

> > On Fri, Mar 06, 2026 at 06:03:01PM +0530, Tarang Raval wrote:
> > > os05b10_set_pad_format() calls os05b10_set_framing_limits() before updating
> > > the ACTIVE format. As a result, the VBLANK control handler uses the old
> > > height when recalculating exposure limits, causing -ERANGE when switching
> > > to a larger resolution.
> > >
> > > Update the ACTIVE format before adjusting framing controls so control
> > > callbacks use the correct dimensions.
> > >
> > > Signed-off-by: Tarang Raval <tarang.raval@xxxxxxxxxxxxxxxxx>
> > > ---
> > >  drivers/media/i2c/os05b10.c | 4 ++--
> > >  1 file changed, 2 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/drivers/media/i2c/os05b10.c b/drivers/media/i2c/os05b10.c
> > > index 4601e33b7e8f..476dbcb49351 100644
> > > --- a/drivers/media/i2c/os05b10.c
> > > +++ b/drivers/media/i2c/os05b10.c
> > > @@ -902,14 +902,14 @@ static int os05b10_set_pad_format(struct v4l2_subdev *sd,
> > >
> > >       format = v4l2_subdev_state_get_format(sd_state, 0);
> > >
> > > +     *format = fmt->format;
> > > +
> > >       if (fmt->which == V4L2_SUBDEV_FORMAT_ACTIVE) {
> > >               ret = os05b10_set_framing_limits(os05b10, mode);
> >
> > Does it take a driver bug for this to happen? Presumably? I guess the
> > result would be somewhat inconsistent state in any case.
>
> In the current driver only a single mode is present, so the issue is not
> observed because no resolution change occurs.
>
> This issue became visible while adding a new mode. During a mode switch
> (from a smaller resolution to a larger one) the limits are calculated
> using the previous format, which results in -ERANGE and the new mode is
> not applied.
>
> Updating *format = fmt->format before adjusting the framing controls
> ensures the control handlers see the correct dimensions.

During testing I messed up.

Sorry for the confusion. With patch 10 the -ERANGE error is already resolved.

During a mode change, os05b10_set_framing_limits() receives the selected
mode as an argument, so it already uses the new resolution when calculating
the limits.

Also, the control handlers are only triggered when streaming starts, so
there is no practical difference whether *format = fmt->format is updated
before or after adjusting the framing controls.

Given this, the change proposed in this patch is not necessary.

I will drop this patch.

Best Regards,
Tarang