[Intel-gfx] [PATCH 4/5] drm/i915: Suppress context restore w/a when using MI_RESTORE_INHIBIT

[Chris Wilson]

On Fri, Feb 24, 2017 at 12:45:54PM +0200, Mika Kuoppala wrote:
> Chris Wilson <chris at chris-wilson.co.uk> writes:
> 
> > If we are setting the context and do inhibit the restoration from the
> > context image, also forgo applying the set-context w/a.
> >
> > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > ---
> >  drivers/gpu/drm/i915/i915_gem_context.c | 12 ++++++------
> >  1 file changed, 6 insertions(+), 6 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c
> > index baceca14f5e0..25e8db73542a 100644
> > --- a/drivers/gpu/drm/i915/i915_gem_context.c
> > +++ b/drivers/gpu/drm/i915/i915_gem_context.c
> > @@ -577,17 +577,17 @@ void i915_gem_context_close(struct drm_device *dev, struct drm_file *file)
> >  }
> >  
> >  static inline int
> > -mi_set_context(struct drm_i915_gem_request *req, u32 hw_flags)
> > +mi_set_context(struct drm_i915_gem_request *req, u32 flags)
> >  {
> >  	struct drm_i915_private *dev_priv = req->i915;
> >  	struct intel_engine_cs *engine = req->engine;
> >  	enum intel_engine_id id;
> > -	u32 *cs, flags = hw_flags | MI_MM_SPACE_GTT;
> >  	const int num_rings =
> >  		/* Use an extended w/a on ivb+ if signalling from other rings */
> >  		i915.semaphores ?
> >  		INTEL_INFO(dev_priv)->num_rings - 1 :
> >  		0;
> > +	u32 *cs;
> >  	int len;
> >  
> >  	/* These flags are for resource streamer on HSW+ */
> > @@ -595,10 +595,10 @@ mi_set_context(struct drm_i915_gem_request *req, u32 hw_flags)
> >  		flags |= (HSW_MI_RS_SAVE_STATE_EN | HSW_MI_RS_RESTORE_STATE_EN);
> >  	else if (INTEL_GEN(dev_priv) < 8)
> >  		flags |= (MI_SAVE_EXT_STATE_EN | MI_RESTORE_EXT_STATE_EN);
> > -
> > +	flags |= MI_MM_SPACE_GTT;
> >  
> >  	len = 4;
> > -	if (INTEL_GEN(dev_priv) >= 7)
> > +	if (INTEL_GEN(dev_priv) >= 7 && !(flags & MI_RESTORE_INHIBIT))
> 
> If we are restoring from the context image, we dont need the
> workarounds. So should this be 

I merged up to the previous patch as we ran into an uncertainity as to
whether some of the w/a applied to restore or to save.

The next patch is still interesting, but merits some soak testing to see
if cures the last full-ppgtt corruption on hsw.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre