Re: [PATCH v2 3/3] drm/i915: Move to new PM core fields

[Ulf Hansson]

On Mon, 17 Dec 2018 at 15:22, Vincent Guittot
<vincent.guittot@xxxxxxxxxx> wrote:
>
> On Fri, 14 Dec 2018 at 15:36, Ulf Hansson <ulf.hansson@xxxxxxxxxx> wrote:
> >
> > On Fri, 14 Dec 2018 at 15:22, Vincent Guittot
> > <vincent.guittot@xxxxxxxxxx> wrote:
> > >
> > > With jiffies been replaced by raw ns in PM core accounting, 915 driver is
> > > updated to use this new time infrastructure.
> > >
> > > Signed-off-by: Vincent Guittot <vincent.guittot@xxxxxxxxxx>
> > > ---
> > >  drivers/gpu/drm/i915/i915_pmu.c | 12 ++++++------
> > >  drivers/gpu/drm/i915/i915_pmu.h |  4 ++--
> > >  2 files changed, 8 insertions(+), 8 deletions(-)
> > >
> > > diff --git a/drivers/gpu/drm/i915/i915_pmu.c b/drivers/gpu/drm/i915/i915_pmu.c
> > > index d6c8f8f..cf6437d 100644
> > > --- a/drivers/gpu/drm/i915/i915_pmu.c
> > > +++ b/drivers/gpu/drm/i915/i915_pmu.c
> > > @@ -493,14 +493,14 @@ static u64 get_rc6(struct drm_i915_private *i915)
> > >                  */
> > >                 if (kdev->power.runtime_status == RPM_SUSPENDED) {
> > >                         if (!i915->pmu.sample[__I915_SAMPLE_RC6_ESTIMATED].cur)
> > > -                               i915->pmu.suspended_jiffies_last =
> > > -                                                 kdev->power.suspended_jiffies;
> > > +                               i915->pmu.suspended_time_last =
> > > +                                       kdev->power.suspended_time;
> > >
> >
> > Huh, so patch 2 introduces a complier error because of removing the
> > old fields. We can't have that.
>
> I agree
> The patch was mainly to raise discussion
> >
> > Ideally the driver shouldn't touch these fields in the first place,
> > but because the fields are defined in a public header, there is always
> > a risk that they becomes abused. I would appreciate if we can change
> > the driver move away from using these fields and make that a patch
> > preceding patch 2.
>
> In fact, the driver doesn't only use some internal fields but also
> takes the power.lock
> IIUC, the driver wants the current pm runtime suspended time to
> estimate a metric when perf events are not accessible

Huh, so it's even worse. :-(

I had a brief look at the runtime PM deployment for these drivers and
it's really messy. Unfortunate, I have limited knowledge about the
gpu/drm drivers, so I can't tell without a investing in a deeper
analyze, of how to convert to something more generic.

> I'm going to send a new version with a runtime pm interface proposal to fix this

Yeah, please go ahead and try something, I am willing to help review.

> >
> > > -                       val = kdev->power.suspended_jiffies -
> > > -                             i915->pmu.suspended_jiffies_last;
> > > -                       val += jiffies - kdev->power.accounting_timestamp;
> > > +                       val = kdev->power.suspended_time -
> > > +                               i915->pmu.suspended_time_last;
> > > +                       val += ktime_to_ns(ktime_get()) -
> > > +                               kdev->power.accounting_timestamp;
> > >
> > > -                       val = jiffies_to_nsecs(val);
> > >                         val += i915->pmu.sample[__I915_SAMPLE_RC6].cur;
> > >
> > >                         i915->pmu.sample[__I915_SAMPLE_RC6_ESTIMATED].cur = val;
> > > diff --git a/drivers/gpu/drm/i915/i915_pmu.h b/drivers/gpu/drm/i915/i915_pmu.h
> > > index 7f164ca..3dc2a30 100644
> > > --- a/drivers/gpu/drm/i915/i915_pmu.h
> > > +++ b/drivers/gpu/drm/i915/i915_pmu.h
> > > @@ -95,9 +95,9 @@ struct i915_pmu {
> > >          */
> > >         struct i915_pmu_sample sample[__I915_NUM_PMU_SAMPLERS];
> > >         /**
> > > -        * @suspended_jiffies_last: Cached suspend time from PM core.
> > > +        * @suspended_time_last: Cached suspend time from PM core.
> > >          */
> > > -       unsigned long suspended_jiffies_last;
> > > +       u64 suspended_time_last;
> > >         /**
> > >          * @i915_attr: Memory block holding device attributes.
> > >          */
> > > --
> > > 2.7.4
> > >

Kind regards
Uffe