Re: [PATCH v1] PM / Domains: Add tracepoints

[Matthias Kaehlcke]

On Wed, Oct 16, 2019 at 03:47:44PM +0200, Ulf Hansson wrote:
> On Tue, 15 Oct 2019 at 19:19, Matthias Kaehlcke <mka@xxxxxxxxxxxx> wrote:
> >
> > Hi Ulf,
> >
> > On Tue, Oct 15, 2019 at 02:37:42PM +0200, Ulf Hansson wrote:
> > > On Tue, 1 Oct 2019 at 18:35, Matthias Kaehlcke <mka@xxxxxxxxxxxx> wrote:
> > > >
> > > > On Fri, Sep 27, 2019 at 04:42:39AM -0400, Steven Rostedt wrote:
> > > > > On Thu, 26 Sep 2019 15:04:38 -0700
> > > > > Matthias Kaehlcke <mka@xxxxxxxxxxxx> wrote:
> > > > >
> > > > > > Define genpd_power_on/off and genpd_set_performance_state
> > > > > > tracepoints and use them.
> > > > >
> > > > > I agree with Greg about adding a "why" you need this. But, in case
> > > > > there's a good reason to have this, I have comments about the code
> > > > > below.
> > > >
> > > > Thanks Greg and Steven for your comments.
> > > >
> > > > How about this instead:
> > > >
> > > >   Add tracepoints for genpd_power_on, genpd_power_off and
> > > >   genpd_set_performance_state. The tracepoints can help with
> > > >   understanding power domain behavior of a given device, which
> > > >   may be particularly interesting for battery powered devices
> > > >   and suspend/resume.
> > >
> > > Apologize for the delay, no excuse!
> > >
> > > I don't mind adding trace events, as long as it's for good reasons -
> > > and to me, that seems a bit questionable here.
> > >
> > > According to the above, I believe the information you need is already
> > > available via genpd's debugfs interface, no?
> >
> > Not in all cases, e.g. you can't peek at sysfs while the device is
> > suspended.
> 
> Not sure I get this right. If a device that is attached to a genpd
> that is runtime suspended, for sure you can have a look at the genpd
> debugfs to see its current status.

Sorry, I used an ambiguous terminology by talking about 'the device',
I was referring to system suspend.

> > Also sysfs doesn't help much with seeing that a PD is
> > toggling between on an off for some (possibly legitimate) reason.
> 
> Well, you could look at the "active_time" and the "total_idle_time"
> nodes for the genpd in question. Those should change accordingly.

Ok, thanks for the pointer!