Re: [PATCH] drivers: core: Detach device from power domain on shutdown

[Rafael J. Wysocki]

On Tue, Jan 12, 2021 at 10:55 AM Dmitry Osipenko <digetx@xxxxxxxxx> wrote:
>
> 02.12.2020 00:30, Furquan Shaikh пишет:
> > When the system is powered off or rebooted, devices are not detached
> > from their PM domain. This results in ACPI PM not being invoked and
> > hence PowerResouce _OFF method not being invoked for any of the
> > devices. Because the ACPI power resources are not turned off in case
> > of poweroff and reboot, it violates the power sequencing requirements
> > which impacts the reliability of the devices over the lifetime of the
> > platform. This is currently observed on all Chromebooks using ACPI.
> >
> > In order to solve the above problem, this change detaches a device
> > from its PM domain whenever it is shutdown. This action is basically
> > analogous to ->remove() from driver model perspective. Detaching the
> > device from its PM domain ensures that the ACPI PM gets a chance to
> > turn off the power resources for the device thus complying with its
> > power sequencing requirements.
> >
> > Signed-off-by: Furquan Shaikh <furquan@xxxxxxxxxx>
> > ---
> >  drivers/base/core.c | 3 +++
> >  1 file changed, 3 insertions(+)
> >
> > diff --git a/drivers/base/core.c b/drivers/base/core.c
> > index d661ada1518f..5823f1d719e1 100644
> > --- a/drivers/base/core.c
> > +++ b/drivers/base/core.c
> > @@ -23,6 +23,7 @@
> >  #include <linux/of_device.h>
> >  #include <linux/genhd.h>
> >  #include <linux/mutex.h>
> > +#include <linux/pm_domain.h>
> >  #include <linux/pm_runtime.h>
> >  #include <linux/netdevice.h>
> >  #include <linux/sched/signal.h>
> > @@ -4057,6 +4058,8 @@ void device_shutdown(void)
> >                       dev->driver->shutdown(dev);
> >               }
> >
> > +             dev_pm_domain_detach(dev, true);
> > +
> >               device_unlock(dev);
> >               if (parent)
> >                       device_unlock(parent);
> >
>
> This patch broke system shutdown on NVIDIA Tegra using today's
> linux-next because power domain can't be turned off until device drivers
> handed control over device resets to the power domain of Power
> Management controller on Tegra. This patch introduced the wrong
> behaviour, apparently it should be made specific to ACPI only.
>
> Please fix, thanks in advance.

OK, so Greg please drop it.