Message-ID: <CAPDyKFqBB7gDXSDV-GbMqWDBbT8RMVdc-9+iDOvEdpVV5TACyw@mail.gmail.com>
Date: Tue, 18 Nov 2025 13:49:16 +0100
From: Ulf Hansson <ulf.hansson@...aro.org>
To: "Rafael J. Wysocki" <rafael@...nel.org>
Cc: linux-pm@...r.kernel.org, Rose Wu <ya-jou.wu@...iatek.com>, 
	rafael.j.wysocki@...el.com, regressions@...ts.linux.dev, 
	linux-kernel@...r.kernel.org, wsd_upstream <wsd_upstream@...iatek.com>, 
	linux-mediatek@...ts.infradead.org, "artis. chiu" <artis.chiu@...iatek.com>, 
	"Johnny-cc. Kao" <Johnny-cc.Kao@...iatek.com>
Subject: Re: [PATCH v2] PM: sleep: core: Fix runtime PM enabling in device_resume_early()

On Tue, 18 Nov 2025 at 13:26, Rafael J. Wysocki <rafael@...nel.org> wrote:
>
> On Tue, Nov 18, 2025 at 1:18 PM Ulf Hansson <ulf.hansson@...aro.org> wrote:
> >
> > On Tue, 18 Nov 2025 at 12:48, Rafael J. Wysocki <rafael@...nel.org> wrote:
> > >
> > > On Tuesday, November 18, 2025 9:31:08 AM CET Rose Wu wrote:
> > > > Hi,
> > > >
> > > > On Mon, 2025-11-17 at 19:57 +0100, Rafael J. Wysocki wrote:
> > > > >
> > > > > Make two changes to address this problem.
> > > > >
> > > > > First, reorder device_suspend_late() to only disable runtime PM for a
> > > > > device if the power.is_late_suspended flag is going to be set for it.
> > > > > In all of the other cases, disabling runtime PM for the device is not
> > > > > in fact necessary.
> > > > >
> > > > > Second, make device_resume_early() only enable runtime PM for the
> > > > > devices with the power.is_late_suspended flag set.
> > > > >
> > > >
> > > > My concern is with the error path in device_suspend_late().
> > > > If a device fails its dpm_run_callback(), it appears that its
> > > > power.is_late_suspended flag is not set, potentially leaving its runtime
> > > > PM disabled during the resume sequence.
> > >
> > > Right, pm_runtime_enable() is missing in the error path after calling
> > > dpm_run_callback().
> > >
> > > ---
> > > From: Rafael J. Wysocki <rafael.j.wysocki@...el.com>
> > >
> > > Runtime PM should only be enabled in device_resume_early() if it has
> > > been disabled for the given device by device_suspend_late().  Otherwise,
> > > it may cause runtime PM callbacks to run prematurely in some cases
> > > which leads to further functional issues.
> > >
> > > Make two changes to address this problem.
> > >
> > > First, reorder device_suspend_late() to only disable runtime PM for a
> > > device when it is going to look for the device's callback.  In all of
> > > the other cases, disabling runtime PM for the device is not in fact
> > > necessary.  However, if the device's callback returns an error and the
> > > power.is_late_suspended flag is not going to be set, enable runtime
> > > PM so it only remains disabled when power.is_late_suspended is set.
> > >
> > > Second, make device_resume_early() only enable runtime PM for the
> > > devices with the power.is_late_suspended flag set.
> > >
> > > Fixes: 443046d1ad66 ("PM: sleep: Make suspend of devices more asynchronous")
> > > Reported-by: Rose Wu <ya-jou.wu@...iatek.com>
> > > Closes: https://lore.kernel.org/linux-pm/70b25dca6f8c2756d78f076f4a7dee7edaaffc33.camel@mediatek.com/
> > > Cc: 6.16+ <stable@...r.kernel.org> # 6.16+
> > > Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@...el.com>
> > > ---
> > >
> > > v1 -> v2: Add pm_runtime_enable() to device_suspend_late() error path (Rose).
> > >
> > > ---
> > >  drivers/base/power/main.c |   15 ++++++++-------
> > >  1 file changed, 8 insertions(+), 7 deletions(-)
> > >
> > > --- a/drivers/base/power/main.c
> > > +++ b/drivers/base/power/main.c
> > > @@ -941,11 +941,11 @@ Run:
> > >
> > >  Skip:
> > >         dev->power.is_late_suspended = false;
> > > +       pm_runtime_enable(dev);
> > >
> > >  Out:
> > >         TRACE_RESUME(error);
> > >
> > > -       pm_runtime_enable(dev);
> > >         complete_all(&dev->power.completion);
> > >
> > >         if (error) {
> > > @@ -1630,12 +1630,6 @@ static void device_suspend_late(struct d
> > >         TRACE_DEVICE(dev);
> > >         TRACE_SUSPEND(0);
> > >
> > > -       /*
> > > -        * Disable runtime PM for the device without checking if there is a
> > > -        * pending resume request for it.
> > > -        */
> > > -       __pm_runtime_disable(dev, false);
> > > -
> > >         dpm_wait_for_subordinate(dev, async);
> > >
> > >         if (READ_ONCE(async_error))
> > > @@ -1649,6 +1643,12 @@ static void device_suspend_late(struct d
> > >         if (dev->power.syscore || dev->power.direct_complete)
> > >                 goto Complete;
> > >
> > > +       /*
> > > +        * Disable runtime PM for the device without checking if there is a
> > > +        * pending resume request for it.
> > > +        */
> > > +       __pm_runtime_disable(dev, false);
> > > +
> >
> > Moving this here means we are going to keep runtime pm enabled for
> > syscore devices during system wide suspend/resume. That's quite a
> > change in behaviour for a fix for a regression, I think. Not saying
> > that it can't work though.
>
> syscore devices can be a special case, but I thought it wouldn't be
> necessary to special-case them.
>
> Do you actually know about any of them needing special casing?

There are a couple of clocksource drivers, cpuidle-psci,
cpuidle-riscv-sbi and a usb driver that marks their devices as syscore
devices.

It *probably* works to keep runtime pm enabled for all of them, but I
am not sure.

>
> > Although, perhaps better to call __pm_runtime_disable() a few lines
> > earlier and use a separate flag to track that we need to call
> > pm_runtime_enable() in device_resume_early()?
>
> I don't think it is necessary or even useful to introduce new flags for this.

Kind regards
Uffe

Powered by blists - more mailing lists