Hi Niklas, Kieran,
On Wed, Nov 25, 2020 at 2:40 PM Niklas Söderlund
<niklas.soderlund+renesas@xxxxxxxxxxxx> wrote:
> On 2020-11-25 13:09:39 +0000, Kieran Bingham wrote:
> > On 22/11/2020 16:36, Niklas Söderlund wrote:
> > > This series enables usage of the ADV748x after the system have been
> > > suspended to ram. During s2ram the ADV748x may be powered down and thus
> > > lose its configuration from probe time. The configuration contains
> > > among other things the i2c slave address mappings for the different
> > > blocks inside the ADV748x. If this is lost the hardware listens to the
> > > "wrong" i2c addresses and becomes inaccessible.
> > >
> > > Example trying to read the analog standard before and after s2ram with
> > > and without this this series.
> >
> > Should we be considering runtime_pm for this instead?
>
> I don't think so, why do you think we should?
>
> I opted for this solution because we need fine grain control of when the
> registers are restored when resuming from s2ram. If they are not
> restored before (in my case) the VIN driver is resumed and it was
> streaming at suspend time it will fail as the i2c address map is wrong
> at this time. For this reason the registers are restored in the early
> resume callback.
>
> Second I'm unsure how we could properly test such a solution as I don't
> think we can powerdown the ADV7482 without also s2ram the whole SoC on
> our test platforms as it's power is not controllable by the SoC. For
> example it's not powered down in s2idel.
Also, there is no need to reset the ADV7482 every time it is
runtime-resumed. In fact such needless reset may impact performance, as
runtime-resume happens much more frequently than system resume during
the system's lifetime.
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
