[Intel-gfx] [PATCH 1/6] drm/i915: Force the switch to the i915->kernel_context

[Chris Wilson]

Quoting Mika Kuoppala (2017-11-02 15:40:50)
> Chris Wilson <chris at chris-wilson.co.uk> writes:
> 
> > In the next few patches, we will have a hard requirement that we emit a
> > context-switch to the perma-pinned i915->kernel_context (so that we can
> > save the HW state using that context-switch). As the first context
> > itself may be classed as a kernel context, we want to be explicit in our
> > comparison. For an extra-layer of finesse, we can check the last
> > unretired context on the engine; as well as the last retired context
> > when idle.
> >
> > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > ---
> >  drivers/gpu/drm/i915/intel_engine_cs.c | 16 ++++++++++++++--
> >  1 file changed, 14 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/intel_engine_cs.c b/drivers/gpu/drm/i915/intel_engine_cs.c
> > index ddbe5c9bf45a..c15e288bed02 100644
> > --- a/drivers/gpu/drm/i915/intel_engine_cs.c
> > +++ b/drivers/gpu/drm/i915/intel_engine_cs.c
> > @@ -1587,8 +1587,20 @@ bool intel_engines_are_idle(struct drm_i915_private *dev_priv)
> >  
> >  bool intel_engine_has_kernel_context(const struct intel_engine_cs *engine)
> >  {
> > -     return (!engine->last_retired_context ||
> > -             i915_gem_context_is_kernel(engine->last_retired_context));
> > +     const struct i915_gem_context *kctx = engine->i915->kernel_context;
> > +     struct drm_i915_gem_request *rq;
> > +
> > +     lockdep_assert_held(&engine->i915->drm.struct_mutex);
> > +
> > +     /* An unretired request? */
> > +     rq = __i915_gem_active_peek(&engine->timeline->last_request);
> > +     if (rq)
> > +             return rq->ctx == kctx;
> > +
> > +     if (!engine->last_retired_context)
> > +             return true;
> > +
> > +     return engine->last_retired_context == kctx;
> 
> 
> Conside that we would have intel_engine_loaded_context(engine)
> {
>         rq = __i915_gem_active_peek(&engine->timeline->last_request);
>         if (rq)
>            return rq->ctx;
> 
>         if (engine->last_retired_context)
>                 return engine->last_retired_context;
> 
>         return NULL;        
> }
> 
> and then have simple intel_engine_has_kernel_context_loaded()

Could, just needs refinement on the name first. I'll keep one clumsy
name for the moment ;)
 
> Also further on the patch series we do switch to kernel context
> on init, to grab the hw state. So would it be possible to completely
> get rid of notion that !engine->last_retired_context means kernel
> context? I would rather assert that we have always last retired
> context after init and resume.

Yup, that seems possible.
-Chris