On Fri, 26 Jul 2013 14:36:08 +0200 Rafael J. Wysocki rjw@xxxxxxx wrote,
> On Friday, July 26, 2013 10:33:21 AM Lukasz Majewski wrote:
> > On Fri, 26 Jul 2013 12:47:15 +0530 Viresh Kumar wrote,
> > > On 25 July 2013 22:03, Lukasz Majewski <l.majewski@xxxxxxxxxxx>
> > > wrote:
> > > > diff --git a/drivers/cpufreq/cpufreq.c
> > > > b/drivers/cpufreq/cpufreq.c
> > >
> > > > /*********************************************************************
> > > > + *
> > > > BOOST *
> > > > +
> > > > *********************************************************************/
> > > > +static int cpufreq_boost_set_sw(int state) +{
> > > > + struct cpufreq_frequency_table *freq_table;
> > > > + struct cpufreq_policy *policy;
> > > > + int ret = -EINVAL;
> > > > +
> > > > + list_for_each_entry(policy, &cpufreq_policy_list,
> > > > policy_list) {
> > > > + freq_table =
> > > > cpufreq_frequency_get_table(policy->cpu);
> > > > + if (freq_table) {
> > > > + ret =
> > > > cpufreq_frequency_table_cpuinfo(policy,
> > > > +
> > > > freq_table);
> > > > + if (!ret) {
> > > > + policy->user_policy.max =
> > > > policy->max;
> > > > + __cpufreq_governor(policy,
> > > > CPUFREQ_GOV_LIMITS);
> > > > + }
> > > > + }
> > > > + }
> > > > +
> > > > + return ret;
> > > > +}
> > > > +
> > > > +int cpufreq_boost_trigger_state(int state)
> > > > +{
> > > > + unsigned long flags;
> > > > + int ret = 0;
> > > > +
> > > > + if (cpufreq_driver->boost_enabled == state)
> > > > + return 0;
> > > > +
> > > > + write_lock_irqsave(&cpufreq_driver_lock, flags);
> > > > + cpufreq_driver->boost_enabled = state;
> > > > + write_unlock_irqrestore(&cpufreq_driver_lock, flags);
> > ^^^^^^^^^^^^^^^^^^^^ [*]
> > >
> > > Not sure if we should leave the lock at this point of time, as we
> > > haven't enabled boost until now.
> >
> > The problem here is with the cpufreq_driver->set_boost() call.
> >
> > I tried to avoid acquiring lock at one function and release it at
> > another (in this case cpufreq_boost_set_sw), especially since the
> > __cpufreq_governor() acquires its own lock - good place for
> > deadlock.
> >
> > Is it OK for you to grab lock at one function
> > (cpufreq_boost_trigger_state()) and then at other function
> > (cpufreq_boost_set_sw) release it before calling
> > __cpufreq_governor() and grab it again after its completion?
>
> It generally is better to avoid doing that, although it is not
> unheard of.
In this particular case, one also needs to pass the "flags" parameter to
the set_boost() function. This looks a bit unnatural to mix lock layer
with the boost.
>
> Thanks,
> Rafael
>
>
--
Best regards,
Lukasz Majewski
Samsung R&D Institute Poland (SRPOL) | Linux Platform Group
--
To unsubscribe from this list: send the line "unsubscribe cpufreq" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
