Re: [PATCH V2 2/3] cpufreq: Panic if policy is active in cpufreq_policy_free()

[Rafael J. Wysocki]

On Wed, Jul 6, 2022 at 5:23 PM Viresh Kumar <viresh.kumar@xxxxxxxxxx> wrote:
>
> On 06-07-22, 15:49, Rafael J. Wysocki wrote:
> > WARN_ON() would be somewhat better, but then I'm not sure if having a
> > full call trace in this case is really useful, because we know when
> > cpufreq_policy_free() can be called anyway.
> >
> > Maybe just print a warning message.
>
> The warning will get printed, yes, but I am sure everyone will end up
> ignoring it, once it happens.
>
> One of the benefits of printing the call-stack is people will take it
> seriously and report it, and we won't miss a bug, if one gets in
> somehow.

I'd rather not go into discussing things that people may or may not do and why.

My point is that if WARN_ON() gets converted to panic(), they will not
see the message at all and if the message gets printed, they will have
a chance to see it even in that case.  Whether or not they use that
chance as desirable is beyond the scope of engineering IMV.