Date:   Mon, 28 Sep 2020 13:23:11 +0200
From:   Ulf Hansson <ulf.hansson@...aro.org>
To:     Jonathan Cameron <jic23@...nel.org>,
        Fabrice Gasnier <fabrice.gasnier@...com>
Cc:     "Rafael J. Wysocki" <rafael.j.wysocki@...el.com>,
        "Rafael J. Wysocki" <rjw@...ysocki.net>,
        Linux ARM <linux-arm-kernel@...ts.infradead.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Maxime Coquelin <mcoquelin.stm32@...il.com>,
        Alexandre Torgue <alexandre.torgue@...com>,
        olivier.moysan@...com, linux-iio@...r.kernel.org,
        linux-stm32@...md-mailman.stormreply.com
Subject: Re: [RESEND PATCH v2] iio: adc: stm32-adc: fix runtime autosuspend
 delay when slow polling

Jonathan, Fabrice,

On Sat, 26 Sep 2020 at 17:17, Jonathan Cameron <jic23@...nel.org> wrote:
>
> On Wed, 16 Sep 2020 12:28:00 +0200
> Fabrice Gasnier <fabrice.gasnier@...com> wrote:
>
> > On 7/1/20 4:55 PM, Fabrice Gasnier wrote:
> > > When the ADC is runtime suspended and starting a conversion, the stm32-adc
> > > driver calls pm_runtime_get_sync() that gets cascaded to the parent
> > > (e.g. runtime resume of stm32-adc-core driver). This also kicks the
> > > autosuspend delay (e.g. 2s) of the parent.
> > > Once the ADC is active, calling pm_runtime_get_sync() again (upon a new
> > > capture) won't kick the autosuspend delay for the parent (stm32-adc-core
> > > driver) as already active.
> > >
> > > Currently, this makes the stm32-adc-core driver go in suspend state
> > > every 2s when doing slow polling. As an example, doing a capture, e.g.
> > > cat in_voltageY_raw at a 0.2s rate, the auto suspend delay for the parent
> > > isn't refreshed. Once it expires, the parent immediately falls into
> > > runtime suspended state, in between two captures, as soon as the child
> > > driver falls into runtime suspend state:
> > > - e.g. after 2s, + child calls pm_runtime_put_autosuspend() + 100ms
> > >   autosuspend delay of the child.
> > > - stm32-adc-core switches off regulators, clocks and so on.
> > > - They get switched on back again 100ms later in this example (at 2.2s).
> > >
> > > So, use runtime_idle() callback in stm32-adc-core driver to call
> > > pm_runtime_mark_last_busy() for the parent driver (stm32-adc-core),
> > > to avoid this.
> > >
> > > Fixes: 9bdbb1139ca1 ("iio: adc: stm32-adc: add power management support")
> > >
> > > Signed-off-by: Fabrice Gasnier <fabrice.gasnier@...com>
> > > ---
> > > Changes in v2:
> > > - Use runtime_idle callback in stm32-adc-core driver, instead of refreshing
> > >   last_busy from the child (for the parent) at many place. Initial patch v1
> > >   looked like "somewhat adhoc solution" as commented by Jonathan.
> >
> > Hi all,
> >
> > Gentle reminder for this patch. Earlier discussions on it were as per
> > [1] and [2].
> >
> > Ideally, Jonathan was looking for an ack from Rafael on this patch.
> > This is a long pending issue. I'd like to progress on this.
> >
> > [1] https://patchwork.kernel.org/patch/11349841/
> > [2] https://lkml.org/lkml/2020/6/11/279
>
> Fabrice, I think this one has sat waiting for inputs for
> too long. Hence I'm going to take a slight gamble that you are correct
> on doing the fix this way (I'm reasonably convinced)

My apologies for the huge and unacceptable delay. I have re-started
looking at this several times, but just never got the point of writing
a proper reply. Let me do this now, better late than never I guess.

In general, I think this problem (nicely described by Fabrice), should
be solved in the runtime PM core, without having to involve drivers
for parents/childs. I have looked into that, but I don't have a patch
to propose, at least not yet.

FYI, I have also stumbled over the same problem, for a card controller
(parent), serving both sd and memstick cards. For that case, we simply
decided to skip using autosuspend for the child devices (represented
by an sd host and a memstick host), not optimal, but there were other
reasons why we decided for this approach as well.

That said, I also think the solution proposed in $subject patch, which
uses the ->runtime_idle() callback for the parent is perfectly fine,
at least until we have figured out something that can replace it.

>
> Applied to the fixes-togreg branch of iio.git.
> It won't go in for 5.9 now, so we have a bit of time for any last
> minute comments.

Feel free to add:

Reviewed-by: Ulf Hansson <ulf.hansson@...aro.org>

Kind regards
Uffe

>
> Thanks,
>
> Jonathan
>
> >
> > Please advise,
> > Thanks in advance,
> > Fabrice
> >
> > > ---
> > >  drivers/iio/adc/stm32-adc-core.c | 9 ++++++++-
> > >  1 file changed, 8 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/drivers/iio/adc/stm32-adc-core.c b/drivers/iio/adc/stm32-adc-core.c
> > > index 0e2068e..3586369 100644
> > > --- a/drivers/iio/adc/stm32-adc-core.c
> > > +++ b/drivers/iio/adc/stm32-adc-core.c
> > > @@ -794,6 +794,13 @@ static int stm32_adc_core_runtime_resume(struct device *dev)
> > >  {
> > >     return stm32_adc_core_hw_start(dev);
> > >  }
> > > +
> > > +static int stm32_adc_core_runtime_idle(struct device *dev)
> > > +{
> > > +   pm_runtime_mark_last_busy(dev);
> > > +
> > > +   return 0;
> > > +}
> > >  #endif
> > >
> > >  static const struct dev_pm_ops stm32_adc_core_pm_ops = {
> > > @@ -801,7 +808,7 @@ static const struct dev_pm_ops stm32_adc_core_pm_ops = {
> > >                             pm_runtime_force_resume)
> > >     SET_RUNTIME_PM_OPS(stm32_adc_core_runtime_suspend,
> > >                        stm32_adc_core_runtime_resume,
> > > -                      NULL)
> > > +                      stm32_adc_core_runtime_idle)
> > >  };
> > >
> > >  static const struct stm32_adc_priv_cfg stm32f4_adc_priv_cfg = {
> > >
>

Powered by blists - more mailing lists