[ QUOTE ]
Hi Dave,
Here are the patches I've talked about on irc for .39.
Patches 1-5 is the "make drm_mm_node embeddable" series. Just the drm core
part, without the patches for i915 (they need to be redone). Chris
Wilson very much likes what these will to do drm/i915. Thomas Hellstrom
cautioned not to go crazy and embed drm_mm_node everywhere (especially in
ttm_bo_object), because the drm_mm allocator is not adequate everwhere and
embedding might just create future obstacles for the Right Thing. Like
switching to idr, as vmgfx already has already done for a few things.
Patches 6-8 embed the gem bo object into the radeon bo object. Changes
since first submission:
- Resolve conflict with pageflip code (resulted in a null-pointer deref).
- Align bo size to PAGE_SIZE (the r600/evergreen kms blit code didn't do
this, upsetting the more paranoid gem code).
Patches 9-11 finish my radeon asic header cleanup that I've started a while
back. Reviewed by Alex Deucher, I've dropped the 3 small patches he
objected.
Tested on my usual array of machines (which now also includes an HD5750 and
a HD 4350 and a g73 from nvidia ;).
Please review and consider merging for next.
Thanks, Daniel
Daniel Vetter (11):
drm/nouveau: don't munge in drm_mm internals
drm: mm: track free areas implicitly
drm: mm: extract node insert helper functions
drm: mm: add api for embedding struct drm_mm_node
drm: mm: add helper to unwind scan state
drm/radeon: embed struct drm_gem_object
drm/radeon: introduce gem_to_radeon_bo helper
drm/radeon: kill radeon_bo->gobj pointer
radeon: consolidate asic-specific function decls for r600 & later
radeon: kill decls for inline functions
radeon: move blit functions to radeon_asic.h
...
[ /QUOTE ]
Tested successfully against linux-next (next-20110218) on a radeon
RV250 gfxcard.
It would be nice to see Daniel's (missing) patchset(s) finally in one
of Dave's drm-2.6 GIT branches.
I had the previous (splitted in multiple little) patchsets running
here againt linux-next for such a long time.
So go go go!
- Sedat -