Re: Still Deadlock problem with pbbuttonsd 0.6.10?
On Wed, 22 Jun 2005 11:44:33 +0200
Johannes Berg <email@example.com> wrote:
> Matthias Grimm wrote:
> >Pbbuttonsd 0.6.10 has a problem if you have 8 input devices attached
> >and disconnect the last one. In best case pbbuttonsd will exit with
> >an error code. Other possibilities are busy loop or segmentation fault.
> >You wouldn't have any problem if you didn't use USB input devices or
> >if you had less than 8 input devices attached (including built in ones).
> >A fixed version will be released soon. For the impatient out there
> >you are invited to use the code in CVS.
> Ohh! I just debugged and found the same issue :)
Then you have some spare time to debug your CDROM issue, I gess :-) Did
you make any progress there?
> Note that on the newest PowerBook, there are 8 devices (if you have the
> trackpad driver) of which one will be automatically disconnected. I
> think the problem is some USB stuff and the kernel bounces the device
> off the bus. Not exactly sure.
The problem is caused by some outdated assumptions of pbbuttonsd.
Pbbuttonsd has a static array for 8 input devices and furthermore it
crashed if you really have 8 devices connected. This value is too low
for recent Powerbooks.
The Current CVS version (which is _not_ the debian package 0.6.10-2)
have 16 entries now and handles the array boundaries correctly.
A word to the fan problematic: Pbbuttonsd is a user space daemon. It
might consume 100% CPU time but it is not able to disturb or freeze
kernel modules. If this happens, the kernel module has a bug and you
should file a bug report to the module maintainer. Pbbuttonsd is only a
customer of kernel services, not more.