Re: [PATCH 08/25] media: atmel: properly get pm_runtime

From: Eugen.Hristev
Date: Mon Sep 06 2021 - 04:14:04 EST


On 9/6/21 11:03 AM, Mauro Carvalho Chehab wrote:
> Hi Eugen,
>
> Em Thu, 10 Jun 2021 12:00:42 +0000
> <Eugen.Hristev@xxxxxxxxxxxxx> escreveu:
>
>> On 6/10/21 12:38 PM, Mauro Carvalho Chehab wrote:
>>> EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe
>>>
>>> Em Thu, 10 Jun 2021 09:04:07 +0000
>>> <Eugen.Hristev@xxxxxxxxxxxxx> escreveu:
>>>
>>>>>> diff --git a/drivers/media/platform/atmel/atmel-isc-base.c b/drivers/media/platform/atmel/atmel-isc-base.c
>>>>>> index fe3ec8d0eaee..ce8e1351fa53 100644
>>>>>> --- a/drivers/media/platform/atmel/atmel-isc-base.c
>>>>>> +++ b/drivers/media/platform/atmel/atmel-isc-base.c
>>>>>> @@ -294,9 +294,13 @@ static int isc_wait_clk_stable(struct clk_hw *hw)
>>>>>> static int isc_clk_prepare(struct clk_hw *hw)
>>>>>> {
>>>>>> struct isc_clk *isc_clk = to_isc_clk(hw);
>>>>>> + int ret;
>>>>>>
>>>>>> - if (isc_clk->id == ISC_ISPCK)
>>>>>> - pm_runtime_get_sync(isc_clk->dev);
>>>>>> + if (isc_clk->id == ISC_ISPCK) {
>>>>>> + ret = pm_runtime_resume_and_get(isc_clk->dev);
>>>>>> + if (ret < 0)
>>>>>> + return ret;
>>>>>> + }
>>>>
>>>> Hi Mauro,
>>>>
>>>> With this patch, the ISC is broken on latest media tree. It looks like
>>>> pm_runtime_resume_and_get for the ISC_ISPCK clock returns -ENOACCESS and
>>>> thus, the probe of the driver fails:
>>>>
>>>> atmel-sama5d2-isc f0008000.isc: failed to enable ispck: -13
>>>> atmel-sama5d2-isc: probe of f0008000.isc failed with error -13
>
> What's the current status of this issue?

Hi Mauro,

Currently, as far as I know, the ISC is broken in 5.14 and current Linux
master. I plan to investigate this issue this week (or the next...),
together with some other things. I want to tryout the PM part of the
driver to see where is the problem.
When I come up with a fix or patch, I will send it on the mailing list.
If you have any ideas that I can try meanwhile, feel free to contact me
to send a patch that I can test.


Eugen


>
> If the bug still happens, we need a fix for it.
>
> We might revert this patch, but this would just be masking some other
> hidden bug.
>
> Regards,
> Mauro
>
>
>
>>>>
>>>>
>>>> Could you point out how I could fix this ? Maybe the isc_clk->dev is not
>>>> properly handled/initialized in some other part of the code ?
>>>
>>> Looking at RPM implementation:
>>>
>>> static int rpm_resume(struct device *dev, int rpmflags)
>>> {
>>> ...
>>> if (dev->power.runtime_error)
>>> retval = -EINVAL;
>>> else if (dev->power.disable_depth == 1 && dev->power.is_suspended
>>> && dev->power.runtime_status == RPM_ACTIVE)
>>> retval = 1;
>>> else if (dev->power.disable_depth > 0)
>>> retval = -EACCES;
>>> ...
>>>
>>> It sounds that RPM is disabled for this clock. Did you call
>>> pm_runtime_enable() before calling isc_clk_prepare()?
>>>
>>> Thanks,
>>> Mauro
>>>
>>
>> I tried to call pm_runtime_enable for the clk at clk_init time, but
>> doing that makes the runtime for the ISC fail like this:
>>
>> atmel-sama5d2-isc f0008000.isc: Unbalanced pm_runtime_enable!
>
>
>
> Thanks,
> Mauro
>