[PATCH v7 00/11] Stop monitoring disabled devices

Andrzej Pietrasiewicz andrzej.p at collabora.com
Fri Jul 3 06:45:44 EDT 2020


Hi,

W dniu 03.07.2020 o 08:38, Daniel Lezcano pisze:
> On 03/07/2020 03:49, Zhang Rui wrote:
>> On Thu, 2020-07-02 at 19:49 +0200, Daniel Lezcano wrote:
> 
> [ ... ]
> 
>>>> So the reason is that ->get_temp() is called while the mutex is
>>>> held and
>>>> thermal_zone_device_is_enabled() wants to take the same mutex.
>>>
>>> Yes, that's correct.
>>>
>>>> Is adding a comment to thermal_zone_device_is_enabled() to never
>>>> call
>>>> it while the mutex is held and adding another version of it which
>>>> does
>>>> not take the mutex ok?
>>>
>>> The thermal_zone_device_is_enabled() is only used in two places, acpi
>>> and this imx driver, and given:
>>>
>>> 1. as soon as the mutex is released, there is no guarantee the
>>> thermal
>>> zone won't be changed right after, the lock is pointless, thus the
>>> information also.
>>>
>>> 2. from a design point of view, I don't see why a driver should know
>>> if
>>> a thermal zone is disabled or not
>>>
>>> It would make sense to end with this function and do not give the
>>> different drivers an opportunity to access this information.
>>
>> I agree.
>>>
>>> Why not add change_mode for the acpi in order to enable or disable
>>> the
>>> events
>>
>> thermal_zone_device_is_enabled() is invoked in acpi thermal driver
>> because we only want to do thermal_zone_device_update() when the acpi
>> thermal zone is enabled.
>>
>> As thermal_zone_device_update() can handle a disabled thermal zone now,
>> we can just remove the check.
> 
> Ah yes, good point!
> 
> 
> 

I sent a short series with fixes. Daniel, can you kindly test it?

Regards,

Andrzej



More information about the linux-arm-kernel mailing list