Re: [PATCH v9 1/5] cpufreq: Allow arch_freq_get_on_cpu to return an error

From: Viresh Kumar
Date: Tue Jan 28 2025 - 03:18:36 EST


On 28-01-25, 09:09, Beata Michalska wrote:
> On Fri, Jan 24, 2025 at 09:03:33AM +0530, Viresh Kumar wrote:
> > On 23-01-25, 22:45, Beata Michalska wrote:
> > > That would mean we are opting for presenting '0' value (whatever that means)
> > > instead of trying alternative ways of getting 'current' frequency ?
> > > This is still the scaling_cur_freq.
> >
> > A return value of 0 should typically mean something went wrong
> > somewhere and didn't return the right value to us.
> If smth goes wrong, an error should be returned, shoulnd't it?

Right, but what if no error is detected and still a value of 0 is
returned somehow ? That's what I was talking about.

> > - For the print message, I think we should just print the value
> > instead of UNKNOWN. Let the user / developer decide what to do with
> > it.
> Are you refering to the x86 show_cpuinfo behaviour altered by this patch ?

Yes

> > - As for trying other mechanism to find the frequency now, maybe you
> > are right and looking for an alternate way is the right way to go.
> > And that would be consistent with existing behavior too.
> >
> That would mean that changes to show_scaling_cur_freq are fine ?

Yes.

--
viresh