[Intel-gfx] [PATCH 3/3] drm/i915: Check the ring is empty when declaring the engines are idle

[Chris Wilson]

On Tue, May 30, 2017 at 03:33:41PM +0300, Mika Kuoppala wrote:
> Chris Wilson <chris at chris-wilson.co.uk> writes:
> 
> > As another precaution when testing whether the CS engine is actually
> > idle, also inspect the ring's HEAD/TAIL registers, which should be equal
> > when there are no commands left to execute by the GPU.
> >
> > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > Cc: Mika Kuoppala <mika.kuoppala at intel.com>
> > ---
> >  drivers/gpu/drm/i915/intel_engine_cs.c | 5 +++++
> >  1 file changed, 5 insertions(+)
> >
> > diff --git a/drivers/gpu/drm/i915/intel_engine_cs.c b/drivers/gpu/drm/i915/intel_engine_cs.c
> > index 699f2d3861c7..bc38bd128b76 100644
> > --- a/drivers/gpu/drm/i915/intel_engine_cs.c
> > +++ b/drivers/gpu/drm/i915/intel_engine_cs.c
> > @@ -1212,6 +1212,11 @@ static bool ring_is_idle(struct intel_engine_cs *engine)
> >  
> >  	intel_runtime_pm_get(dev_priv);
> >  
> > +	/* First check that no commands are left in the ring */
> > +	if ((I915_READ_HEAD(engine) & HEAD_ADDR) !=
> > +	    (I915_READ_TAIL(engine) & TAIL_ADDR))
> > +		idle = false;
> > +
> 
> You are already certain that is not idle so why not goto out?

In this case I could argue that extra path for the jump is not worth it.
It saves a mmio read, yes, but will any one notice?

It boils down to is it easier to read as:

static bool ring_is_idle(struct intel_engine_cs *engine)
{
        struct drm_i915_private *dev_priv = engine->i915;
        bool idle = false;

        intel_runtime_pm_get(dev_priv);

        /* First check that no commands are left in the ring */
        if ((I915_READ_HEAD(engine) & HEAD_ADDR) !=
            (I915_READ_TAIL(engine) & TAIL_ADDR))
                goto busy;

        /* No bit for gen2, so assume the CS parser is idle */
        if (INTEL_GEN(dev_priv) > 2 && !(I915_READ_MODE(engine) & MODE_IDLE))
                goto busy;

        idle = true;
busy:
        intel_runtime_pm_put(dev_priv);

        return idle;
}

First instinct would be no because of the goto.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre