Re: [PATCH v2 02/10] cpufreq: provide data for frequency-invariant load-tracking support

[Dietmar Eggemann]

On 13/07/17 13:40, Sudeep Holla wrote:
> 
> 
> On 11/07/17 16:21, Dietmar Eggemann wrote:
>> On 11/07/17 07:39, Viresh Kumar wrote:
>>> On 10-07-17, 14:46, Rafael J. Wysocki wrote:

[...]

>> Like I said in the other email, since for (future)
>> arm/arm64 fast-switch driver, the return value of
>> cpufreq_driver->fast_switch() does not give us the information that the
>> frequency value did actually change, we have to implement
> 
> I was under the impression that we strictly don't care about that
> information when I started exploring the fast_switch with the standard
> firmware interface on ARM platforms(until if and when ARM provides an
> instruction to achieve that).
> 
> If f/w failed to change the frequency, will that be not corrected in the
> next sample or instance. I would like to know the impact of absence of
> such notifications.

In the meantime we agreed that we have to invoke frequency invariance
from within the cpufreq driver.

For a fast-switch driver I would have to put the call to
arch_set_freq_scale() somewhere where I know that the frequency has been
set.

Without a notification (from the firmware) that the frequency has been
set, I would have to call arch_set_freq_scale() somewhere in the
driver::fast_switch() call assuming that the frequency has been actually
set.

[...]