Re: [PATCH v6 12/13] media: i2c: imx214: Fix clock handling on probe error or remove

From: Sakari Ailus
Date: Thu Jan 09 2025 - 04:00:27 EST


Hi André,

On Wed, Jan 08, 2025 at 10:46:51PM +0100, André Apitzsch wrote:
> Hi Sakari,
>
> Am Mittwoch, dem 08.01.2025 um 11:51 +0000 schrieb Sakari Ailus:
> > Hi André,
> >
> > Thanks for the update.
> >
> > On Fri, Dec 20, 2024 at 02:26:11PM +0100, André Apitzsch via B4 Relay
> > wrote:
> > > From: André Apitzsch <git@xxxxxxxxxxx>
> > >
> > > The driver should disable or unprepare the clock on probe error and
> > > on
> > > removing the module.
> > >
> > > Remove regulator_bulk_disable(), because it is called by
> > > imx214_power_off().
> > >
> > > Acked-by: Ricardo Ribalda <ribalda@xxxxxxxxxxxx>
> > > Signed-off-by: André Apitzsch <git@xxxxxxxxxxx>
> >
> > This and the next one appear to be bugfixes. I think it'd be
> > reasonable to backport them. Any idea where this got broken, for the
> > Fixes: tag?
> > The bug fixed by the last patch was probably introduced with the
> > driver?
> >
>
> I think both, the missing imx214_power_off() and the wrong link
> frequency where introduced with
> > 436190596241 ("media: imx214: Add imx214 camera sensor driver")
>
> The first one could also be considered as a fix for the incomplete
> > 9bc92332cc3f ("media: imx214: Fix the error handling in imx214_probe()")

This one seems to have missed review. :-(

>
> Shall I submit a new version with the Fixes: tag?

I checked the fix again and it seems the problem isn't as simple:
pm_runtime_idle() call leads to powering the device off. It needs to either
be resumed first or the call to imx214_power_off() be omitted.

I tried to find a good example but the few I found are buggy. :-( Getting
runtime PM right is just hard. I'll drop the last patch and post another to
address the problem (plus the intended example drivers) if you're fine with
that.

--
Kind regards,

Sakari Ailus