[Xcb] xcb recursive write not protected in all cases?

[Keith Packard]

Uli Schlachter <psychon at znc.in> writes:

> Thus, the send_sync() requests are already atomic in the current code.

so you think the only problem is that the code doesn't wait for both
external owners and out.writing to be ready at the same time? Do we need
additional checks for the same condition 

(I think I prefer my simple while loop over the two separate checks; it
makes it much clearer to me that the conditions are necessarily correct
after the loop terminates)

> After having figured all of this out today, who wants to write this down? And
> where? Some Wiki page called "ThreadSafetyInternals"?

Sounds like we've got a good start here...

-- 
keith.packard at intel.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 827 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/xcb/attachments/20131226/2e5682e1/attachment.pgp>