[PULL] Xorg/darwin related build fixes, Xquartz crash fix

[Jeremy Huddleston]

On Jan 16, 2012, at 3:18 PM, Keith Packard wrote:

> On Mon, 16 Jan 2012 14:35:12 -0800, Jeremy Huddleston <jeremyhu at apple.com> wrote:
> 
>> Sorry if the comment was ambiguous, but I did not mean to imply that
>> the fix was present already or provided.  Reverting this patch does
>> return the memory leak when using glxdri, and a followup needs to be
>> provided to fix it properly in __glXDRIscreenDestroy.  I do not use
>> glxdri, so I wanted to leave that to someone else (Paulo) to
>> implement.
> 
> Sounds good. I'm somewhat hesitant to merge your sequence in before I
> know that it won't cause a regression for other users though, or at
> least have some promise that a fix for the impending regression is on
> the way.

Well the change that is being reverted *does* cause a regression, so it should be reverted on that grounds alone.  It will be no worse than our released servers which didn't free that memory.  Of course this happens in screen destroy, so it's not like it was happening anywhere other than shutdown in most use cases.