lkml.org 
[lkml]   [2015]   [Jun]   [29]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH v3 2/4] cpufreq: introduce cpufreq_driver_might_sleep
Hi,

On Mon, Jun 29, 2015 at 09:56:55AM -0700, Michael Turquette wrote:
> > > > > @@ -112,6 +112,12 @@ bool have_governor_per_policy(void)
> > > > > }
> > > > > EXPORT_SYMBOL_GPL(have_governor_per_policy);
> > > > >
> > > > > +bool cpufreq_driver_might_sleep(void)
> > > > > +{
> > > > > + return !(cpufreq_driver->flags & CPUFREQ_DRIVER_WILL_NOT_SLEEP);
> > > > > +}
> > > > > +EXPORT_SYMBOL_GPL(cpufreq_driver_might_sleep);
> > > > > +
> > > > > struct kobject *get_governor_parent_kobj(struct cpufreq_policy *policy)
> > > > > {
> > > > > if (have_governor_per_policy())
> > > > > diff --git a/include/linux/cpufreq.h b/include/linux/cpufreq.h
> > > > > index 2ee4888..1f2c9a1 100644
> > > > > --- a/include/linux/cpufreq.h
> > > > > +++ b/include/linux/cpufreq.h
> > > > > @@ -157,6 +157,7 @@ u64 get_cpu_idle_time(unsigned int cpu, u64 *wall, int io_busy);
> > > > > int cpufreq_get_policy(struct cpufreq_policy *policy, unsigned int cpu);
> > > > > int cpufreq_update_policy(unsigned int cpu);
> > > > > bool have_governor_per_policy(void);
> > > > > +bool cpufreq_driver_might_sleep(void);
> > > > > struct kobject *get_governor_parent_kobj(struct cpufreq_policy *policy);
> > > > > #else
> > > > > static inline unsigned int cpufreq_get(unsigned int cpu)
> > > > > @@ -314,6 +315,14 @@ struct cpufreq_driver {
> > > > > */
> > > > > #define CPUFREQ_NEED_INITIAL_FREQ_CHECK (1 << 5)
> > > > >
> > > > > +/*
> > > > > + * Set by drivers that will never block or sleep during their frequency
> > > > > + * transition. Used to indicate when it is safe to call cpufreq_driver_target
> > > > > + * from non-interruptable context. Drivers must opt-in to this flag, as the
> > > > > + * safe default is that they might sleep.
> > > > > + */
> > > > > +#define CPUFREQ_DRIVER_WILL_NOT_SLEEP (1 << 6)
> > > >
> > > > don't you need to update current drivers and pass this flag where
> > > > necessary ?
> > >
> > > Felipe,
> > >
> > > Thanks for the review.
> > >
> > > Setting the flag can be done, but it is an opt-in feature. First, none
> > > of the legacy cpufreq governors would actually make use of this flag.
> > > Everything they do is in process context. The first potential user of it
> > > is in patch #3.
> > >
> > > Secondly, the governor in patch #3 will work without this flag set for a
> > > cpufreq driver. It will just defer the dvfs transition to a kthread
> > > instead of performing it in the hot path of the scheduler.
> > >
> > > Finally, the only hardware I am aware of that can make use of this flag
> > > is Intel hardware. I know nothing about it and am happy for someone more
> > > knowledgeable than myself submit a patch enabling this flag for that
> > > architecture.
> >
> > the follow-up question would be: then why introduce the flag at all ?
> > :-p
>
> I included it at Rafael's request:
>
> http://lkml.kernel.org/r/<49407954.UBSF2FlX46@vostro.rjw.lan>

Fair enough, just think it might be an unused code path for a while,
since the flag isn't enabled anywhere :-s

--
balbi
[unhandled content-type:application/pgp-signature]
\
 
 \ /
  Last update: 2015-06-29 19:21    [W:0.048 / U:34.548 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site