[Intel-gfx] [PATCH] drm/i915: Invariably invalidate before ctx switch

[Daniel Vetter]

On Thu, Apr 03, 2014 at 01:21:38PM +0300, Ville Syrjälä wrote:
> On Thu, Apr 03, 2014 at 08:05:35AM +0100, Chris Wilson wrote:
> > On Wed, Apr 02, 2014 at 10:30:23PM -0700, Ben Widawsky wrote:
> > > We have been setting the bit which was originally BIOS dependent since:
> > > commit f05bb0c7b624252a5e768287e340e8e45df96e42
> > > Author: Chris Wilson <chris at chris-wilson.co.uk>
> > > Date:   Sun Jan 20 16:33:32 2013 +0000
> > > 
> > >     drm/i915: GFX_MODE Flush TLB Invalidate Mode must be '1' for scanline waits
> > > 
> > > Therefore, we do not need to try to figure it out dynamically and we can
> > > just always invalidate the TLBs.
> > > 
> > > It's a partial revert of:
> > > commit 12b0286f49947a6cdc9285032d918466a8c3f5f9
> > > Author: Ben Widawsky <ben at bwidawsk.net>
> > > Date:   Mon Jun 4 14:42:50 2012 -0700
> > > 
> > >     drm/i915: possibly invalidate TLB before context switch
> > > 
> > > The original commit attempted to only invalidate when necessary
> > > (very much a relic from the old days). Now, we can just always invalidate.
> > > 
> > > I guess the old TODO still exists. Since we seem to have abandoned ILK
> > > contexts however, there isn't much point in even remembering.
> > > 
> > > Cc: Chris Wilson <chris at chris-wilson.co.uk>
> > > Signed-off-by: Ben Widawsky <ben at bwidawsk.net>
> > 
> > Seems reasonable, except in most cases (execbuffer) there will be
> > a following cache-invalidate as part of the move-to-gpu.
> 
> Except we still move_to_gpu() before the context switch. My fbc related
> patch to change that order never got merged.

Hm, pointer to that thing? I guess we need to sign up someone to dig out
all your fbc patches and get them in line in prep for bdw. At least
another opportunity to beat all that into better shape ...
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch