possible bugs in xfs-related packages
Hello,
I've just filled two bugs against xfs-xtt (#51082 and #51084).
The first one is that SIGHUP is being used instead of SIGUSR1 to cause the
xfs server to reload its configuration (in the /etc/init.d scripts). This
will cause the server to reset, closing all connections and therefore
causing all sort of misbehaviour in xservers which dislike such behaviour
(XFree86 *is* among those).
The second one deals with the xfs daemon dying when it is sent a SIGHUP or
SIGUSR1 signal, after logging to syslog it is going to either reset (SIGHUP)
or reload (SIGUSR1). This is probably related to file permission problems
when a xfs server is started as root and told to setuid to nobody.
Maybe the xfs packages will need to be started already suid to a xfs user
after all... :-(
Could someone test the other two xfs packages (xfs, xfstt) for the above two
bugs and report them to the maintainers, please ?
Do be careful when testing, Xfree86 may segfault if you cause a fontserver
in its fontpath to die (at least using unix sockets. I didn't try tcp). If
you are very lucky, it will sit around eating 100% cpu for a while and not
crash before you restart the font server, remove it from the old fontpath to
reset the fontpath state and add it back afterwards.
--
"One disk to rule them all, One disk to find them. One disk to bring
them all and in the darkness grind them. In the Land of Redmond
where the shadows lie." -- The Silicon Valley Tarot
Henrique Holschuh
Reply to: