Re: [PATCH 3/2] drm/i915: Streamline VLV forcewake handling

[Daniel Vetter <daniel@ffwll.ch>]

On Fri, Feb 28, 2014 at 04:51:16PM +0200, Ville Syrjälä wrote:
> On Fri, Feb 28, 2014 at 07:56:56PM +0530, S, Deepak wrote:
> > 
> > 
> > On 2/28/2014 1:37 AM, ville.syrjala@linux.intel.com wrote:
> > > From: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > >
> > > It occured to me that when we're trying to wake up both render
> > > and media wells on VLV, we might end up calling the low level
> > > force_wake_get/put two times even though one call would be
> > > enough. Make that happen by figuring out which wells really
> > > need to be woken up based on the forcewake counts.
> > >
> > > Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > > ---
> > >   drivers/gpu/drm/i915/intel_uncore.c | 70 +++++++++++++++----------------------
> > >   1 file changed, 29 insertions(+), 41 deletions(-)
> > >
> > > diff --git a/drivers/gpu/drm/i915/intel_uncore.c b/drivers/gpu/drm/i915/intel_uncore.c
> > > index dacb751..4119ddc 100644
> > > --- a/drivers/gpu/drm/i915/intel_uncore.c
> > > +++ b/drivers/gpu/drm/i915/intel_uncore.c
> > > @@ -251,16 +251,16 @@ void vlv_force_wake_get(struct drm_i915_private *dev_priv,
> > >   	unsigned long irqflags;
> > >
> > >   	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
> > > -	if (FORCEWAKE_RENDER & fw_engine) {
> > > -		if (dev_priv->uncore.fw_rendercount++ == 0)
> > > -			dev_priv->uncore.funcs.force_wake_get(dev_priv,
> > > -							FORCEWAKE_RENDER);
> > > -	}
> > > -	if (FORCEWAKE_MEDIA & fw_engine) {
> > > -		if (dev_priv->uncore.fw_mediacount++ == 0)
> > > -			dev_priv->uncore.funcs.force_wake_get(dev_priv,
> > > -							FORCEWAKE_MEDIA);
> > > -	}
> > > +
> > > +	if (fw_engine & FORCEWAKE_RENDER &&
> > > +	    dev_priv->uncore.fw_rendercount++ != 0)
> > > +		fw_engine &= ~FORCEWAKE_RENDER;
> > > +	if (fw_engine & FORCEWAKE_MEDIA &&
> > > +	    dev_priv->uncore.fw_mediacount++ != 0)
> > > +		fw_engine &= ~FORCEWAKE_MEDIA;
> > 
> > Should  we add WARN_ON? I think it will help us if we have forcewake 
> > count mismatch?
> > 
> > Other than this. Patch looks good.
> > Reviewed-by:Deepak S <deepak.s@intel.com>
> 
> I dropped the WARNs since we didn't have them on other platforms either.
> But if people think they might help, I'm not opposed to keeping them.

Especially with the multiple fw domains I think a WARN for a refcount
underflow would be useful. btw for vlv fw unification: The delayed put is
also still on the table ...

Entire series merged, thanks for patches&review.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch