[PATCH 1/2] watchdog/perf: Provide function for adjusting the event period

[Yicong Yang]

On 2025/3/8 0:53, Doug Anderson wrote:
> Hi,
> 
> On Thu, Mar 6, 2025 at 6:18 PM Yicong Yang <yangyicong at huawei.com> wrote:
>>
>> @@ -211,6 +211,27 @@ void hardlockup_detector_perf_cleanup(void)
>>         cpumask_clear(&dead_events_mask);
>>  }
>>
>> +/**
>> + * hardlockup_detector_perf_adjust_period - Adjust the events period due
>> + *                                          to cpu frequency change
>> + */
>> +void hardlockup_detector_perf_adjust_period(int cpu, u64 period)
> 
> Running `scripts/kernel-doc -v kernel/watchdog_perf.c > /dev/null`
> after your patch shows:
> 
> kernel/watchdog_perf.c:219: warning: Function parameter or struct
> member 'cpu' not described in 'hardlockup_detector_perf_adjust_period'
> kernel/watchdog_perf.c:219: warning: Function parameter or struct
> member 'period' not described in
> 'hardlockup_detector_perf_adjust_period'
> 

so the script will check whether parameters are documented as well,
need to add description.

I just followed the document convention like hardlockup_detector_perf_cleanup().

> 
> Other than that, this seems reasonable to me.
> 
> ...but I'd also have to ask: is there a reason you're using the "perf"
> hard-lockup detector instead of the buddy one? In my mind, the "buddy"
> watchdog is better in almost all ways (I believe it's lower power,
> doesn't waste a "perf" controller, and doesn't suffer from frequency
> issues).

ok, sounds reasonable to me especially for save one perf counter.

> It's even crossed my mind whether the "perf" lockup detector
> should be deprecated. ;-)
> .

I think it's mainly due to the policy of the distribution. watchdog_perf
is used by default on openEuler which is used on my board.

Thanks.