[Intel-gfx] [PATCH 16/19] drm/i915: sanity check power well sw state against hw state

[Ville Syrjälä]

On Tue, Feb 18, 2014 at 07:37:01PM +0200, Imre Deak wrote:
> On Tue, 2014-02-18 at 18:55 +0200, Ville Syrjälä wrote:
> > On Tue, Feb 18, 2014 at 12:02:17AM +0200, Imre Deak wrote:
> > > Suggested by Daniel.
> > > 
> > > Signed-off-by: Imre Deak <imre.deak at intel.com>
> > > ---
> > >  drivers/gpu/drm/i915/intel_pm.c | 33 ++++++++++++++++++++++++++++++---
> > >  1 file changed, 30 insertions(+), 3 deletions(-)
> > > 
> > > diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
> > > index e81e7de..21ccf89 100644
> > > --- a/drivers/gpu/drm/i915/intel_pm.c
> > > +++ b/drivers/gpu/drm/i915/intel_pm.c
> > > @@ -5338,6 +5338,24 @@ static void hsw_power_well_disable(struct drm_i915_private *dev_priv,
> > >  	hsw_enable_package_c8(dev_priv);
> > >  }
> > >  
> > > +static void check_power_well_state(struct drm_i915_private *dev_priv,
> > > +				   struct i915_power_well *power_well)
> > > +{
> > > +	bool enabled;
> > > +
> > > +	if (!power_well->ops->is_enabled)
> > > +		return;
> > > +
> > > +	enabled = power_well->ops->is_enabled(dev_priv, power_well);
> > > +
> > > +	if (enabled != (power_well->count > 0 || !i915.disable_power_well)) {
> > 
> > Doesn't i915.disable_power_well==true mean "leave power wells always
> > enabled"? So I think the '!' needs to be removed.
> 
> No, i915.disable_power_well==true means disable power wells when the
> refcount goes to 0. Perhaps not the best name/semantics for this kind of
> option, the default for it should be 0 and mean normal operation, which
> is to disable power wells when possible.

Oh I had the impression it was the other way around, but you're right.
Seems I keep getting confused by this thing. It has happened before and
I'm guessing it will happen again after I've forgotten the details
again.

> 
> > > +		DRM_ERROR("state mismatch for '%s' (hw state %d use-count %d disable_power_well %d\n",
> > > +			  power_well->name, enabled, power_well->count,
> > > +			  i915.disable_power_well);
> > > +		WARN_ON(1);
> > > +	}
> > 
> > For an error message + backtrace, you could just use WARN().
> 
> Ok.
> 
> > > +}
> > > +
> > >  void intel_display_power_get(struct drm_i915_private *dev_priv,
> > >  			     enum intel_display_power_domain domain)
> > >  {
> > > @@ -5349,9 +5367,14 @@ void intel_display_power_get(struct drm_i915_private *dev_priv,
> > >  
> > >  	mutex_lock(&power_domains->lock);
> > >  
> > > -	for_each_power_well(i, power_well, BIT(domain), power_domains)
> > > -		if (!power_well->count++ && power_well->ops->enable)
> > > +	for_each_power_well(i, power_well, BIT(domain), power_domains) {
> > > +		if (!power_well->count++ && power_well->ops->enable) {
> > > +			DRM_DEBUG_KMS("enabling %s\n", power_well->name);
> > >  			power_well->ops->enable(dev_priv, power_well);
> > > +		}
> > > +
> > > +		check_power_well_state(dev_priv, power_well);
> > > +	}
> > >  
> > >  	power_domains->domain_use_count[domain]++;
> > >  
> > > @@ -5376,8 +5399,12 @@ void intel_display_power_put(struct drm_i915_private *dev_priv,
> > >  		WARN_ON(!power_well->count);
> > >  
> > >  		if (!--power_well->count && power_well->ops->disable &&
> > > -		    i915.disable_power_well)
> > > +		    i915.disable_power_well) {
> > > +			DRM_DEBUG_KMS("disabling %s\n", power_well->name);
> > >  			power_well->ops->disable(dev_priv, power_well);
> > > +		}
> > > +
> > > +		check_power_well_state(dev_priv, power_well);
> > >  	}
> > >  
> > >  	mutex_unlock(&power_domains->lock);
> > > -- 
> > > 1.8.4
> > > 
> > > _______________________________________________
> > > Intel-gfx mailing list
> > > Intel-gfx at lists.freedesktop.org
> > > http://lists.freedesktop.org/mailman/listinfo/intel-gfx
> > 
> 



-- 
Ville Syrjälä
Intel OTC