Re: [PATCH v2] PM / Domains: Call driver's noirq callbacks

[Mikko Perttunen <mperttunen@nvidia.com>]

On 20.06.2017 17:18, Ulf Hansson wrote:
> On 20 June 2017 at 15:38, Mikko Perttunen <mperttunen@nvidia.com> wrote:
>> Currently genpd installs its own suspend_noirq, resume_noirq,
>> and poweroff_noirq callbacks, but never calls down to the driver's
>> corresponding callbacks. Add these calls.
>>
>> Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
>> ---
>> v2:
>> - Moved pm_generic_suspend_noirq to before pm_runtime_force_suspend,
>>   and correspondingly pm_generic_resume_noirq after
>>   pm_runtime_force_resume
>> - Added new pm_genpd_poweroff_noirq callback that is identical to
>>   pm_genpd_suspend_noirq but calls the appropriate driver callback
>>
>>  drivers/base/power/domain.c | 50 ++++++++++++++++++++++++++++++++++++++++++++-
>>  1 file changed, 49 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c
>> index d3f1d96f75e9..b070ee58186d 100644
>> --- a/drivers/base/power/domain.c
>> +++ b/drivers/base/power/domain.c
>> @@ -919,6 +919,10 @@ static int pm_genpd_suspend_noirq(struct device *dev)
>>         if (dev->power.wakeup_path && genpd_dev_active_wakeup(genpd, dev))
>>                 return 0;
>>
>> +       ret = pm_generic_suspend_noirq(dev);
>> +       if (ret)
>> +               return ret;
>> +
>>         if (genpd->dev_ops.stop && genpd->dev_ops.start) {
>>                 ret = pm_runtime_force_suspend(dev);
>>                 if (ret)
>> @@ -961,6 +965,10 @@ static int pm_genpd_resume_noirq(struct device *dev)
>>         if (genpd->dev_ops.stop && genpd->dev_ops.start)
>>                 ret = pm_runtime_force_resume(dev);
>>
>> +       ret = pm_generic_resume_noirq(dev);
>> +       if (ret)
>> +               return ret;
>> +
>>         return ret;
>>  }
>>
>> @@ -1015,6 +1023,46 @@ static int pm_genpd_thaw_noirq(struct device *dev)
>>  }
>>
>>  /**
>> + * pm_genpd_poweroff_noirq - Completion of hibernation of device in an
>> + *   I/O PM domain.
>> + * @dev: Device to poweroff.
>> + *
>> + * Stop the device and remove power from the domain if all devices in it have
>> + * been stopped.
>> + */
>> +static int pm_genpd_poweroff_noirq(struct device *dev)
>> +{
>> +       struct generic_pm_domain *genpd;
>> +       int ret;
>> +
>> +       dev_dbg(dev, "%s()\n", __func__);
>> +
>> +       genpd = dev_to_genpd(dev);
>> +       if (IS_ERR(genpd))
>> +               return -EINVAL;
>> +
>> +       if (dev->power.wakeup_path && genpd_dev_active_wakeup(genpd, dev))
>> +               return 0;
>> +
>> +       ret = pm_generic_poweroff_noirq(dev);
>
> The only difference between  pm_genpd_suspend_noirq() and
> pm_genpd_poweroff_noirq() is the above line. Can we re-factor the code
> so we avoid open code here, please.

I wasn't sure if the functions' complexity warranted adding a helper 
function, but sure, I'll refactor this with a helper function.

>
>> +       if (ret)
>> +               return ret;
>> +
>> +       if (genpd->dev_ops.stop && genpd->dev_ops.start) {
>> +               ret = pm_runtime_force_suspend(dev);
>> +               if (ret)
>> +                       return ret;
>> +       }
>> +
>> +       genpd_lock(genpd);
>> +       genpd->suspended_count++;
>> +       genpd_sync_power_off(genpd, true, 0);
>> +       genpd_unlock(genpd);
>> +
>> +       return 0;
>> +}
>> +
>> +/**
>>   * pm_genpd_restore_noirq - Start of restore of device in an I/O PM domain.
>>   * @dev: Device to resume.
>>   *
>> @@ -1493,7 +1541,7 @@ int pm_genpd_init(struct generic_pm_domain *genpd,
>>         genpd->domain.ops.resume_noirq = pm_genpd_resume_noirq;
>>         genpd->domain.ops.freeze_noirq = pm_genpd_freeze_noirq;
>>         genpd->domain.ops.thaw_noirq = pm_genpd_thaw_noirq;
>> -       genpd->domain.ops.poweroff_noirq = pm_genpd_suspend_noirq;
>> +       genpd->domain.ops.poweroff_noirq = pm_genpd_poweroff_noirq;
>>         genpd->domain.ops.restore_noirq = pm_genpd_restore_noirq;
>
> The pm_genpd_restore_noirq() doesn't invokes the lower level
> ->restore_noirq() callbacks. If you are going to change that for the
> *poweroff* callback, certainly we should change that also for the
> *restore* callbacks as well. Don't you think?
>
> Moreover, what about the freeze and thaw callbacks, should these also
> walk the lower level callbacks?

Yes, I'll add the calls to the rest of the ops as well.

Thanks,
Mikko.

>
>>         genpd->domain.ops.complete = pm_genpd_complete;
>>
>> --
>> 2.1.4
>>
>
> Kind regards
> Uffe
>