Bell notifications stop being reported in Fedora 14 Xorg server 1.9.4

Davide Bolcioni davide_bolcioni at
Wed Mar 30 04:20:57 PDT 2011

this is my first post on this list so I hope this is the right place for reporting the problem.

First, a bit about context: a Fedora 14 installation is used as the X terminal for a legacy industrial control application running on old Solaris servers.

The application uses a mix of raw Xlib calls and early Motif code. It crashes if no PseudoColor visual can be found, and for reasons far above my head must not ever be modified.

I have been providing ad-hoc support for the beast using slightly modified installs or Live CDs of Fedora for some time now, but now I am facing a thorny problem with the bell.

Fedora was chosen in the hope that support for graphics adapters would be fairly up to date within the framework of a standard X11 setup, since procurement rules here say that I can ask for a box and get an "equivalent" box with totally different hardware if the purchasing department feels like it. 

I know about Wayland, but that's a different story.

The legacy application, it turns out, also uses the X bell, and the Xkb extension, as a means to alert the operator to various
conditions of varying degrees of significance.

In this context, I wrote a small C program which listens for bell notifications and, upon receiving one, forks to play a sine wave of the given pitch and duration using PulseAudio, since its introduction apparently caused this behavior to disappear from Fedora, especially the live variants.

Neither that listener nor the legacy application let go of their X11 connection. In the case of the listener, I have been killing
and restarting it periodically in an attempt to work around the
problem and also to handle the cause of PulseAudio exiting for no reason I could discern.

No matter what I try, after a time between about 29 and 43 hours the listener stops receiving bell notifications although the legacy application is still running happily.

Firing up an xterm which runs a shell script echoing a bunch of BELs results in the appropriate cacophony. Every time.

Restarting the legacy application, which I can confirm closes the X11 connection, does not help. Notice that the application consists of an active instance on a server and a standby instance on a twin, each with its own X11 connection, so it is possible to terminate each in turn without disrupting anything.

I am not sure I ever terminated them both, as this would disrupt service, but I am going to explicitly try in a week or so.

Restarting PulseAudio does not help (just in case).

Restarting the listener does not help, nor apparently does it impact the interval between occurrences of the problem.

Does the above ring a bell (pun intended), anyone ?

Thank you in advance,
Davide Bolcioni
There is no place like /home.


More information about the xorg mailing list