[Patch 00/02] mieq threading prep

[Tiago Vignatti]

Hi,

Jeremy Huddleston escreveu:
> These are two fairly straight forward patches I'd like to give someone a 
> chance to comment on.  In a single-thread Xserver, these patches should 
> have no effect, but they make the code a little more friendly to Tiago's 
> threading patches (and this is what we're already doing in the apple 
> branches).

As I already pointed in other occasion, I'm not comfortable enough to 
push my series of patches to upstream right now. Peter told in a private 
mail that feels the same.

There are probably more critical regions (regarding input properties 
mainly) that I'm not covered yet (and I'm in the process to finish my 
masters soon. So I don't want to put another piano on my head). 
Therefore the real reason now to not push upstream is very simple: 
multi-thread programming sucks.


> The first only increments mieqEventQueue.tail when the data is in the 
> queue (since mieqProcessInputEvents assumes the data to be valid if tail 
> is incremented).
> 
> The second moves a hunk in mieqProcessInputEvents after we increment 
> mieqEventQueue.head, so we don't need to hold a mutex for an unnecessarily.
> 
> If nobody pipes up in the next few days, I'll be pushing them into master.

Right. If the patches are useful for xquartz then both are justified 
even in the lack of xorg's input thread:

     Signed-off-by: Tiago Vignatti <vignatti at c3sl.ufpr.br>


Thanks Jeremy,

-- 
Tiago Vignatti
C3SL - Centro de Computação Científica e Software Livre
www.c3sl.ufpr.br