Re: [PATCH] arm64: Provide an AMU-based version of arch_freq_get_on_cpu

[Beata Michalska <beata.michalska@arm.com>]

On Thu, Jul 27, 2023 at 10:56:05AM +0100, Will Deacon wrote:
> On Wed, Jun 07, 2023 at 03:00:49PM +0100, Beata Michalska wrote:
> > On Wed, Jun 07, 2023 at 10:58:56AM +0100, Sudeep Holla wrote:
> > > On Tue, Jun 06, 2023 at 04:57:54PM +0100, Beata Michalska wrote:
> > > > With the Frequency Invariance Engine (FIE) being already wired up with
> > > > sched tick and making use of relevant (core counter and constant
> > > > counter) AMU counters, getting the current frequency for a given CPU
> > > > on supported platforms, can be achieved by utilizing the frequency scale
> > > > factor which reflects an average CPU frequency for the last tick period
> > > > length.
> > > > 
> > > > With that at hand, arch_freq_get_on_cpu dedicated implementation
> > > > gets enrolled into cpuinfo_cur_freq policy sysfs attribute handler,
> > > > which is expected to represent the current frequency of a given CPU,
> > > > as obtained by the hardware. This is exactly the type of feedback that
> > > > cycle counters provide.
> > > > 
> > > > In order to avoid calling arch_freq_get_on_cpu from the scaling_cur_freq
> > > > attribute handler for platforms that do provide cpuinfo_cur_freq, and
> > > > yet keeping things intact for those platform that do not, its use gets
> > > > conditioned on the presence of cpufreq_driver (*get) callback (which also
> > > > seems to be the case for creating cpuinfo_cur_freq attribute).
> > > >
> > > 
> > > LGTM,
> > > 
> > > Reviewed-by: Sudeep Holla <sudeep.holla@arm.com>
> > > 
> > Thanks for the review.
> > > However I fail to understand if both the changes are dependent ?
> > > Can this be split into 2 patches ? I fail to see the dependency, what
> > > am I missing ? Even if there is some dependency to get arch value
> > > (arch_freq_get_on_cpu() from show_cpuinfo_cur_freq()), you can push
> > > that change first followed by the arm64 change as 2 different change.
> > > 
> > I guess I could split the patch into two parts:
> > 1. adding implementation for the arch_freq_get_on_cpu
> > 2. wiring it up with the cpufreq relevant attrib handlers
> > 
> > or the other way round (if that's what you have in mind).
> > 
> > Will wait a bit for any further comments before pushing new v.
> 
> Are you still planning on a v2?

Apologies for late reply, 've been away for a while and then got bit swamped.
I do not think there will be v2 unless I'll find reasonable way to handle cases
as one mentioned in [1].

---
BR
B.

[1] https://lore.kernel.org/linux-arm-kernel/691d3eb2-cd93-f0fc-a7a4-2a8c0d44262c@nvidia.com/T/#m4e74cb5a0aaa353c60fedc6cfb95ab7a6e381e3c
> 
> Will