[Xcb] Xlib/XCB bug fix: XID allocation

[Ian Osgood]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Apr 2, 2006, at 1:15 AM, Jamey Sharp wrote:

> If anyone's trying out Xlib/XCB, you might want to upgrade your Xlib
> bits. Yesterday I fixed a bug in Xlib's XID allocation: if a client  
> ran
> out of the initial batch of XIDs provided by the server at connection
> setup, then a request for more got sent while Xlib was already in the
> middle of constructing a request, and an assertion failed.
>
> On my system I apparently get about two million XIDs in that initial
> batch, so only heavy and prolonged usage of Firefox demonstrated  
> the bug
> in practice. I've written a tiny test app that demonstrated the bug  
> in a
> fraction of a second (allocating XIDs is way faster than sending
> requests) and the fix I've committed both seems correct and works  
> in the
> test app. No sign of assertion death from Firefox in 24 hours of
> testing, either. :-)
>
> I'm disturbed, though, that Firefox seemed to have consumed those two
> million XIDs after only twelve million requests. Is it really  
> allocating
> an XID every six requests? Is that normal?
>
> --Jamey

I didn't see the test app in the commit.  Is there a place for  
regression tests to live in the XCL or XCB trees?

Do fixes like this need a bugzilla bug?

Ian
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (Darwin)

iD8DBQFEL/lGHDwFgzc3zyIRAhFOAKC4QMHKlgInIAXZmfX7+EEOfojtgwCgxqtu
wXrMxC8+IH1b5vx0IF1+rXI=
=LNF2
-----END PGP SIGNATURE-----