Re: [PATCH 1/2] drm/i915: No need to put forcewake after a reset

[Daniel Vetter <daniel@ffwll.ch>]

On Wed, Mar 05, 2014 at 09:59:55AM -0800, Ben Widawsky wrote:
> On Wed, Mar 05, 2014 at 06:08:18PM +0200, Mika Kuoppala wrote:
> > As we now have intel_uncore_forcewake_reset() no need
> > to do explicit put after reset.
> > 
> > v2: rebase
> > 
> > Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
> > ---
> >  drivers/gpu/drm/i915/intel_uncore.c |   19 ++++++++-----------
> >  1 file changed, 8 insertions(+), 11 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/intel_uncore.c b/drivers/gpu/drm/i915/intel_uncore.c
> > index 6ca24ac..00320fd 100644
> > --- a/drivers/gpu/drm/i915/intel_uncore.c
> > +++ b/drivers/gpu/drm/i915/intel_uncore.c
> > @@ -952,6 +952,7 @@ static int gen6_do_reset(struct drm_device *dev)
> >  	struct drm_i915_private *dev_priv = dev->dev_private;
> >  	int	ret;
> >  	unsigned long irqflags;
> > +	u32 fw_engine = 0;
> >  
> >  	/* Hold uncore.lock across reset to prevent any register access
> >  	 * with forcewake not set correctly
> > @@ -971,25 +972,21 @@ static int gen6_do_reset(struct drm_device *dev)
> >  
> >  	intel_uncore_forcewake_reset(dev);
> >  
> > -	/* If reset with a user forcewake, try to restore, otherwise turn it off */
> > +	/* If reset with a user forcewake, try to restore */
> 
> Technically with the introduction of the deferred forcewaker_ put, it's
> not just user forcewake that this restores.
> 
> Wouldn't it be nice if we actually used something other than
> FORCEWAKE_KERNEL for debugfs?

mt forcewake doesn't exist on snb :(

> 
> >  	if (IS_VALLEYVIEW(dev)) {
> >  		if (dev_priv->uncore.fw_rendercount)
> > -			dev_priv->uncore.funcs.force_wake_get(dev_priv, FORCEWAKE_RENDER);
> > -		else
> > -			dev_priv->uncore.funcs.force_wake_put(dev_priv, FORCEWAKE_RENDER);
> > +			fw_engine |= FORCEWAKE_RENDER;
> >  
> >  		if (dev_priv->uncore.fw_mediacount)
> > -			dev_priv->uncore.funcs.force_wake_get(dev_priv, FORCEWAKE_MEDIA);
> > -		else
> > -			dev_priv->uncore.funcs.force_wake_put(dev_priv, FORCEWAKE_MEDIA);
> > +			fw_engine |= FORCEWAKE_MEDIA;
> >  	} else {
> >  		if (dev_priv->uncore.forcewake_count)
> > -			dev_priv->uncore.funcs.force_wake_get(dev_priv, FORCEWAKE_ALL);
> > -		else
> > -			dev_priv->uncore.funcs.force_wake_put(dev_priv, FORCEWAKE_ALL);
> > +			fw_engine = FORCEWAKE_ALL;
> >  	}
> 
> P
> 
> >  
> > -	/* Restore fifo count */
> > +	if (fw_engine)
> > +		dev_priv->uncore.funcs.force_wake_get(dev_priv, fw_engine);
> > +
> >  	if (IS_GEN6(dev) || IS_GEN7(dev))
> >  		dev_priv->uncore.fifo_count =
> >  			__raw_i915_read32(dev_priv, GTFIFOCTL) &
> 
> I for one would not be opposed to a vlv_do_reset() and a
> gen8_do_reset(). I implemented such a thing at some point, but threw it
> away because I didn't actually have a vlv diff at the time.
> 
> I had to re-review a lot of the uncore.lock stuff, but lgtm:
> Reviewed-by: Ben Widawsky <ben@bwidawsk.net>

Queued for -next, thanks for the patch.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch