Re: [PATCH v3 1/5] PM: sleep: Resume children after resuming the parent

[Jon Hunter <jonathanh@nvidia.com>]

On 07/05/2025 15:25, Rafael J. Wysocki wrote:

...

>>>> So are the devices in question "async"?  To check this, please see the
>>>> "async" attribute in the "power" subdirectory of the sysfs device
>>>> directory for each of them.
>>>
>>> I checked for both the I2C controller and ina3221 and don't see any
>>> 'async' files ...
>>>
>>> $ ls /sys/class/i2c-dev/i2c-2/device/2-0040/power/
>>> autosuspend_delay_ms  runtime_active_time  runtime_suspended_time
>>> control               runtime_status
>>> $ ls /sys/class/i2c-dev/i2c-2/device/2-0041/power/
>>> autosuspend_delay_ms  runtime_active_time  runtime_suspended_time
>>> control               runtime_status
>>> $ ls /sys/class/i2c-dev/i2c-2/power/
>>> autosuspend_delay_ms  runtime_active_time  runtime_suspended_time
>>> control               runtime_status
>>
>> You need to set CONFIG_PM_ADVANCED_DEBUG to see those (and other debug
>> attributes).
>>
>>>> If they are "async", you can write "disable" to this attribute to turn
>>>> them into "sync" devices.  I'd do this and see what happens.
>>
>> You may also turn off async suspend altogether:
>>
>> # echo 0 > /sys/power/pm_async
>>
>> and see if this helps.

This does indeed help!

>>>> Overall, it looks like some dependencies aren't properly represented
>>>> by device links on this platform.
>>>
>>> Yes that would appear to be the case, but at the moment, I don't see
>>> what it is. The ina3221 devices appear to suspend fine AFAICT, but hangs
>>> when suspending I2C controller. Exactly where is still a mystery.
> 
> I checked in the meantime and found that the i2c subsystem enables
> async suspend/resume for all devices, clients and controllers, so the
> devices in question are "async" AFAICS.

So that would make sense given that the above works.

When it fails it appears to hang in dpm_wait_for_subordinate() when 
calling dpm_wait_for_children() from what I can tell.

I will enable the PM_ADVANCED_DEBUG and confirm that making the I2C 
itself non-async works.

Jon

-- 
nvpublic