Re: [git pull] drm fixes

[Chris Wilson <chris@chris-wilson.co.uk>]

On Wed, 23 Feb 2011 15:58:02 -0800, Linus Torvalds <torvalds@linux-foundation.org> wrote:
> On Wed, Feb 23, 2011 at 3:17 PM, Dave Airlie <airlied@linux.ie> wrote:
> >
> > Nothing too major,
> >
> > Two regression fixers (one revert that got fixes properly elsewhere), some
> > timestamp fixes and an agp module reload fix.
> 
> Pulled. However, what about the report from	Pavel Machek <pavel@ucw.cz>:
> 
> > >      drm/i915: Completely disable fence pipelining.
> >
> > Reverting this commit helps in v2.6.38-rc5+ when screen is not fully updated,
> > or has a corrupted picture like horizontal black or white stripes. Using a
> > compositor like compiz may help to avoid the problem.
> >
> > See bug https://bugzilla.kernel.org/show_bug.cgi?id=27572
> 
> Any update on that one?

No, reverting that will cause just another bug elsewhere. I need to
work out how the gpu is not being flushed with a non-pipelined fence
change.

> There's that whole "return -EINVAL for
> I915_PARAM_HAS_RELAXED_FENCING" patch thing?

As it turns out this is a bug in the userspace components of the stack for
gen2 hardware, with lax kernel side enforcement. Daniel has a fix for both.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre