[Intel-gfx] i915 next
Chris Wilson
chris at chris-wilson.co.uk
Wed Apr 13 09:26:35 CEST 2011
On Tue, 12 Apr 2011 21:31:28 +0100, Chris Wilson <chris at chris-wilson.co.uk> wrote:
> This is just the first batch of patches that look ready for testing and
> feedback.
>
> 1-9: Eric's modesetting refactor. This has met with unanimous approval.
> 10-14: Ben's rc6 fixes for Ironlake, and Jesse's module parameter for SNB.
> 15-22: Enabling LLC by default on SNB. There are a couple of new patches in
> there since Eric's posting to switch pwrite and mmap GTT to use the
> cached CPU domains, which may or may not be strictly necessary for
> earlier chipsets.
> 23: Cache GT fifo count. Short term performance gain for the ddx, but will
> probably be dropped in favour of Ben's GT read/write fixes. Hint, Ben,
> hint.
> 24-25: Some minor code refactoring
> 26-30: Pipelined fence fixes.
I've pushed this patchset to drm-intel-next-proposed
[git://git.kernel.org/pub/scm/linux/kernel/git/ickle/drm-intel.git] for
ease of integrated testing.
For those of you keeping track, that means we know have:
drm-intel-fixes - used by QA as their stable tree for nightly regression
testing
Bug fixes should be based on this branch (or Linus or
airlied/drm-fixes, they all should be equivalent).
drm-intel-staging - used by us for communication of patches to testers
(inc. QA) so that we have a single source for
testing before those patches are baked into the stable
trees.
drm-intel-next - used by QA for their unstable tree and regular
testing of feature development. The patches in this
tree are ready for pushing upstream in the next merge
window. (Remember this is Dave's merge window not
Linus's, which is ideally rc4-rc5 i.e. next week!)
Feature development should be based on this branch.
* drm-intel-next-proposed - used by me to track what patches I've sent to
the list that I feel are ready for applying,
and by you for convenience and integration
testing
[And the observant will have noticed a drm-intel-backport because there
are always those patches which are critical for stability but have not yet
made it into 2.6.x.y at the time of our Q release.]
Hope this helps,
-Chris
--
Chris Wilson, Intel Open Source Technology Centre
More information about the Intel-gfx
mailing list