Re: [PATCH 3/4] PM / OPP: take RCU lock in dev_pm_opp_get_opp_count

[Dmitry Torokhov]

On Wednesday, December 17, 2014 10:06:17 AM Viresh Kumar wrote:
> On 17 December 2014 at 04:39, Dmitry Torokhov <dtor@xxxxxxxxxxxx> wrote:
> > A lot of callers are missing the fact that dev_pm_opp_get_opp_count
> > needs to be called under RCU lock. Given that RCU locks can safely be
> > nested, instead of providing *_locked() API, let's take RCU lock inside
> 
> Hmm, I asked for a *_locked() API because many users of
> dev_pm_opp_get_opp_count() are already calling it from rcu read side
> critical sections.
> 
> Now, there are two questions:
> - Can rcu-read side critical sections be nested ?
> 
> Yes, this is what the comment over rcu_read_lock() says
> 
>  * RCU read-side critical sections may be nested.  Any deferred actions
>  * will be deferred until the outermost RCU read-side critical section
>  * completes.
> 
> - Would it be better to drop these double rcu_read_locks() ? i.e. either
> get a *_locked() API or fix the callers of dev_pm_opp_get_opp_count().
> 
> @Paul: What do you say ?
>

FWIW the change is a stop-gap; I hope we'll get away from using 
dev_pm_opp_get_opp_count() in cpufreq drivers and then we can revert the 
change. I just did not want to touch cpufreq drivers unless necessary.

Thanks,
Dmitry
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/