[igt-dev] [PATCH i-g-t] pm_rpm: Require DMC loaded before testing runtime_pm for gen9+

[Chris Wilson]

Quoting Chris Wilson (2018-08-17 22:43:03)
> Quoting Rodrigo Vivi (2018-08-17 22:35:53)
> > Since we block runtime PM if DMC is not loaded, let's skip
> > the test.
> > 
> > v2: Use i915_dmc_info presence to detect dmc requirement
> >     instead of gen check as Chris suggested.
> > v3: Add missing \0 before using buf. (Chris)
> > 
> > Cc: Chris Wilson <chris at chris-wilson.co.uk>
> > Cc: Imre Deak <imre.deak at intel.com>
> > Cc: Anusha Srivatsa <anusha.srivatsa at intel.com>
> > Signed-off-by: Rodrigo Vivi <rodrigo.vivi at intel.com>
> > ---
> >  tests/pm_rpm.c | 16 ++++++++++++++++
> >  1 file changed, 16 insertions(+)
> > 
> > diff --git a/tests/pm_rpm.c b/tests/pm_rpm.c
> > index bbe36e59..fa0d98dc 100644
> > --- a/tests/pm_rpm.c
> > +++ b/tests/pm_rpm.c
> > @@ -693,6 +693,21 @@ static void setup_pc8(void)
> >         has_pc8 = true;
> >  }
> >  
> > +static bool dmc_loaded(void)
> > +{
> > +       char buf[15];
> > +       int len;
> > +
> > +       len = igt_sysfs_read(debugfs, "i915_dmc_info", buf, sizeof(buf));
> 
> sizeof(buf) - 1 :)
> 
> or else len will be sizeof(buf) and buf[len] not where it belongs.

I was debating whether or not having a big red fail was a good idea or
not, but at the end of the day it is an external failure (e.g. the user
may purposely not provide the dmc firmware).

With the final fix,
Reviewed-by: Chris Wilson <chris at chris-wilson.co.uk>
-Chris