[Intel-gfx] [PATCH] drm/i915/cmdparser: Limit clflush to active cachelines

[Chris Wilson]

On Fri, Mar 10, 2017 at 10:04:16AM +0000, Chris Wilson wrote:
> On Fri, Mar 10, 2017 at 11:58:44AM +0200, Mika Kuoppala wrote:
> > Chris Wilson <chris at chris-wilson.co.uk> writes:
> > 
> > > We only need to clflush those cachelines that we have validated to be
> > > read by the GPU. Userspace typically fills the batch length in
> > > correctly, the exceptions tend to be explicit tests within igt.
> > >
> > > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > > ---
> > >  drivers/gpu/drm/i915/i915_cmd_parser.c | 3 ++-
> > >  1 file changed, 2 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/drivers/gpu/drm/i915/i915_cmd_parser.c b/drivers/gpu/drm/i915/i915_cmd_parser.c
> > > index 21b1cd917d81..b9ce9a6881ea 100644
> > > --- a/drivers/gpu/drm/i915/i915_cmd_parser.c
> > > +++ b/drivers/gpu/drm/i915/i915_cmd_parser.c
> > > @@ -1331,7 +1331,8 @@ int intel_engine_cmd_parser(struct intel_engine_cs *engine,
> > >  	}
> > >  
> > >  	if (ret == 0 && needs_clflush_after)
> > > -		drm_clflush_virt_range(shadow_batch_obj->mm.mapping, batch_len);
> > > +		drm_clflush_virt_range(shadow_batch_obj->mm.mapping,
> > > +				       (void *)cmd - shadow_batch_obj->mm.mapping);
> > 
> > (void *)cmd - ptr_mask_bit(shadow_batch_obj->mm.mapping)
> > 
> > We get away as the wb mapping being zero but for correctness.
> 
> The low bits of mm.mapping cannot change the cacheline and so doesn't
> affect the clflush. Same as before.

Although, not quite the same as before as before we didn't use a fixed
end-point and so it could overshoot by a cacheline, even across a page
boundary.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre