[Intel-gfx] [PATCH 2/2] drm/i915: Reorder media/render reset on g4x

Chris Wilson chris at chris-wilson.co.uk
Wed May 17 12:05:40 UTC 2017


On Tue, May 16, 2017 at 04:38:01PM +0300, Mika Kuoppala wrote:
> Chris Wilson <chris at chris-wilson.co.uk> writes:
> 
> > Ville found a reference to WaMediaResetBeforeFullReset which we presume
> > means that we should simply do the media reset first.
> 
> Yesterday I reordered the resets but I recall it didnt help.
> I will retry but regardless yeah resetting media first makes
> sense.
> 
> >
> > References: https://bugs.freedesktop.org/show_bug.cgi?id=100942
> > Suggested-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
> > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > Cc: Ville Syrjälä <ville.syrjala at linux.intel.com>
> > Cc: Mika Kuoppala <mika.kuoppala at intel.com>
> > ---
> >  drivers/gpu/drm/i915/intel_uncore.c | 12 ++++++------
> >  1 file changed, 6 insertions(+), 6 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/intel_uncore.c b/drivers/gpu/drm/i915/intel_uncore.c
> > index fc3da0a6fdbb..c3d0d81b50e4 100644
> > --- a/drivers/gpu/drm/i915/intel_uncore.c
> > +++ b/drivers/gpu/drm/i915/intel_uncore.c
> > @@ -1504,12 +1504,6 @@ static int g4x_do_reset(struct drm_i915_private *dev_priv, unsigned engine_mask)
> >  	struct pci_dev *pdev = dev_priv->drm.pdev;
> >  	int ret;
> >  
> > -	pci_write_config_byte(pdev, I915_GDRST,
> > -			      GRDOM_RENDER | GRDOM_RESET_ENABLE);
> > -	ret =  wait_for(g4x_reset_complete(pdev), 500);
> > -	if (ret)
> > -		goto out;
> > -
> >  	/* WaVcpClkGateDisableForMediaReset:ctg,elk */
> >  	I915_WRITE(VDECCLK_GATE_D, I915_READ(VDECCLK_GATE_D) | VCP_UNIT_CLOCK_GATE_DISABLE);
> >  	POSTING_READ(VDECCLK_GATE_D);
> > @@ -1517,11 +1511,17 @@ static int g4x_do_reset(struct drm_i915_private *dev_priv, unsigned engine_mask)
> >  	pci_write_config_byte(pdev, I915_GDRST,
> >  			      GRDOM_MEDIA | GRDOM_RESET_ENABLE);
> >  	ret =  wait_for(g4x_reset_complete(pdev), 500);
> > +	if (ret)
> > +		goto out;
> >
> 
> We should restore the WaVcp... state if we fail. Not that it
> was right to begin with.

Did it right in the previous patch, then promptly undid it. I was
thinking if we should just do the clock gate tweak for the whole
function, it will look slightly easier to follow.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre


More information about the Intel-gfx mailing list