Re: [PATCH v4 2/2] thermal: core: Add a back up thermal shutdown mechanism

From: Eduardo Valentin
Date: Fri Apr 14 2017 - 14:18:26 EST


Hey,

On Fri, Apr 14, 2017 at 08:42:20AM -0700, Eduardo Valentin wrote:
> Hello again,
>
> On Fri, Apr 14, 2017 at 08:38:40AM -0700, Eduardo Valentin wrote:
> > Hey,
> >
> > On Fri, Apr 14, 2017 at 02:22:13PM +0530, Keerthy wrote:
> > > orderly_poweroff is triggered when a graceful shutdown
> > > of system is desired. This may be used in many critical states of the
> > > kernel such as when subsystems detects conditions such as critical
> > > temperature conditions. However, in certain conditions in system
> > > boot up sequences like those in the middle of driver probes being
> > > initiated, userspace will be unable to power off the system in a clean
> > > manner and leaves the system in a critical state. In cases like these,
> > > the /sbin/poweroff will return success (having forked off to attempt
> > > powering off the system. However, the system overall will fail to
> > > completely poweroff (since other modules will be probed) and the system
> > > is still functional with no userspace (since that would have shut itself
> > > off).
> > >
> > > However, there is no clean way of detecting such failure of userspace
> > > powering off the system. In such scenarios, it is necessary for a backup
> > > workqueue to be able to force a shutdown of the system when orderly
> > > shutdown is not successful after a configurable time period.
> > >
> > > Reported-by: Nishanth Menon <nm@xxxxxx>
> > > Signed-off-by: Keerthy <j-keerthy@xxxxxx>
> > > ---
> > >
> > > Changes in v4:
> > >
> > > * Updated documentation
> > > * changed emergency_poweroff_func to thermal_emergency_poweroff_func
> > >
> > > Changes in v3:
> > >
> > > * Removed unnecessary mutex init.
> > > * Added WARN messages instead of a simple warning message.
> > > * Added Documentation.
> > >
> > > Documentation/thermal/sysfs-api.txt | 19 +++++++++++++++
> > > drivers/thermal/Kconfig | 13 +++++++++++
> > > drivers/thermal/thermal_core.c | 46 +++++++++++++++++++++++++++++++++++++
> > > 3 files changed, 78 insertions(+)
> > >
> > > diff --git a/Documentation/thermal/sysfs-api.txt b/Documentation/thermal/sysfs-api.txt
> > > index ef473dc..e73cc12 100644
> > > --- a/Documentation/thermal/sysfs-api.txt
> > > +++ b/Documentation/thermal/sysfs-api.txt
> > > @@ -582,3 +582,22 @@ platform data is provided, this uses the step_wise throttling policy.
> > > This function serves as an arbitrator to set the state of a cooling
> > > device. It sets the cooling device to the deepest cooling state if
> > > possible.
> > > +
> > > +6. thermal_emergency_poweroff:
> > > +
> > > +On an event of critical trip temperature crossing. Thermal framework
> > > +allows the system to shutdown gracefully by calling orderly_poweroff().
> > > +In the event of a failure of orderly_poweroff() to shut down the system
> > > +we are in danger of keeping the system alive at undesirably high
> > > +temperatures. To mitigate this high risk scenario we program a work
> > > +queue to fire after a pre-determined number of seconds to start
> > > +an emergency shutdown of the device using the kernel_power_off()
> > > +function. In case kernel_power_off() fails then finally
> > > +emergency_restart() is called in the worst case.
> > > +
> > > +The delay should be carefully profiled so as to give adequate time for
> > > +orderly_poweroff(). In case of failure of an orderly_poweroff() the
> > > +emergency poweroff kicks in after the delay has elapsed and shuts down
> > > +the system.
> > > +
> > > +If set to 0 emergency poweroff will happen immediately.
> > > diff --git a/drivers/thermal/Kconfig b/drivers/thermal/Kconfig
> > > index 9347401..0dd5b85 100644
> > > --- a/drivers/thermal/Kconfig
> > > +++ b/drivers/thermal/Kconfig
> > > @@ -15,6 +15,19 @@ menuconfig THERMAL
> > >
> > > if THERMAL
> > >
> > > +config THERMAL_EMERGENCY_POWEROFF_DELAY_MS
> > > + int "Emergency poweroff delay in milli-seconds"
> > > + depends on THERMAL
> > > + default 0
> >
> > Only now I realized that merging this may break the working
> > orderly_poweroff() out there, because you are defaulting this to 0, no
> > delay, therefore giving no time for orderly_poweroff() to finish. This
> > is not good.
> >
> > I think using 0 delay as immediate power off is not good as we give no
> > time for graceful shutdown, and by default. My suggestion here
> > is to use 0 delay as no forced shutdown. Meaning, by default, this
> > feature is disabled, and all other systems out there, despite DRA7 with
> > arago over NFS, work as before.

A better solution could be to have bool Kconfig, say
THERMAL_EMERGENCY_POWEROFF, which would default to false. If one selects
that option, you get the DELAY_MS configurable, and then you could get
the 0 ms still as a valid entry, with the same semantics of immediate
power off, no orderly_poweroff.

I just want to avoid breaking everybody (or changing userland
expectation) in honor of this change.

> >
> > > + help
> > > + The number of milliseconds to delay before emergency
> > > + poweroff kicks in. The delay should be carefully profiled
> > > + so as to give adequate time for orderly_poweroff(). In case
> > > + of failure of an orderly_poweroff() the emergency poweroff
> > > + kicks in after the delay has elapsed and shuts down the system.
> > > +
> > > + If set to 0 poweroff will happen immediately.
> > > +
> > > config THERMAL_HWMON
> > > bool
> > > prompt "Expose thermal sensors as hwmon device"
> > > diff --git a/drivers/thermal/thermal_core.c b/drivers/thermal/thermal_core.c
> > > index 8337c27..aed614d 100644
> > > --- a/drivers/thermal/thermal_core.c
> > > +++ b/drivers/thermal/thermal_core.c
> > > @@ -324,6 +324,47 @@ static void handle_non_critical_trips(struct thermal_zone_device *tz,
> > > def_governor->throttle(tz, trip);
> > > }
> > >
> > > +/**
> > > + * thermal_emergency_poweroff_func - emergency poweroff work after a known delay
> > > + * @work: work_struct associated with the emergency poweroff function
> > > + *
> > > + * This function is called in very critical situations to force
> > > + * a kernel poweroff after a configurable timeout value.
> > > + */
> > > +static void thermal_emergency_poweroff_func(struct work_struct *work)
> > > +{
> > > + /*
> > > + * We have reached here after the emergency thermal shutdown
> > > + * Waiting period has expired. This means orderly_poweroff has
> > > + * not been able to shut off the system for some reason.
> > > + * Try to shut down the system immediately using kernel_power_off
> > > + * if populated
> > > + */
> > > + WARN(1, "Attempting kernel_power_off: Temperature too high\n");
> > > + kernel_power_off();
> > > +
> > > + /*
> > > + * Worst of the worst case trigger emergency restart
> > > + */
> > > + WARN(1, "Attempting emergency_restart: Temperature too high\n");
> > > + emergency_restart();
> > > +}
> > > +
> > > +static DECLARE_DELAYED_WORK(thermal_emergency_poweroff_work,
> > > + thermal_emergency_poweroff_func);
> > > +
> > > +/**
> > > + * thermal_emergency_poweroff - Trigger an emergency system poweroff
> > > + *
> > > + * This may be called from any critical situation to trigger a system shutdown
> > > + * after a known period of time. By default the delay is 0 millisecond
> > > + */
> > > +void thermal_emergency_poweroff(void)
> > > +{
> > > + schedule_delayed_work(&thermal_emergency_poweroff_work,
> > > + msecs_to_jiffies(CONFIG_THERMAL_EMERGENCY_POWEROFF_DELAY_MS));
> >
> > So, please, only schedule if delay is greater than 0.
>
> Please update documentation accordingly..
>


Attachment: signature.asc
Description: Digital signature