[RFC] xkb: processInputProc wrapping and FreezeThaw

[Daniel Stone]

On Fri, Mar 28, 2008 at 12:55:40AM -0400, Thomas Jaeger wrote:
> If FreezeThaw is called inside a UNWRAP_PROCESS_INPUT_PROC /
> COND_WRAP_PROCESS_INPUT_PROC pair, the X server might be left in a
> broken state.  This problem was addressed in commit
> 50e80c39870adfdc84fdbc00dddf1362117ad443 [1], but the workaround doesn't
> cover all possible cases.  There is a bug that is fairly easy to
> reproduce [2,3] where the server will autorepeat a key forever if it is
> released under load and compiz is running.
> 
> The attached patch tries to offer a complete solution to the problem,
> implemented as follows.  The UNWRAP/COND_WRAP macros set both
> realInputProc and processInputProc, unless "regular" input processing is
> suspended currently suspended (this is the case if processInputProc !=
> realInputProc), in which case only realInputProc is updated.

Okay, this looks fine to me.  The reasoning's sound and wow, comments!
Novel.

I can't think of a case where we'd possibly want to smash the enqueue
proc, so yeah, looks good.  ajax, do you want to merge this to both
master and 1.5?

Cheers,
Daniel
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.x.org/archives/xorg/attachments/20080328/b28d4c23/attachment.pgp>