FS#26643 - xorg-xdm 1.1.11 segfaults consequently
Attached to Project:
Arch Linux
Opened by Martin Kühne (mar77i) - Thursday, 27 October 2011, 04:11 GMT
Last edited by Ionut Biru (wonder) - Saturday, 28 January 2012, 23:35 GMT
Opened by Martin Kühne (mar77i) - Thursday, 27 October 2011, 04:11 GMT
Last edited by Ionut Biru (wonder) - Saturday, 28 January 2012, 23:35 GMT
|
Details
Description:
Well, the title pretty much says it. Looking at the log, xdm causes the segfault in the X process, but I can't really tell what's going wrong at the moment, but I could spend an afternoon on it and probably will. Also there's been broken pointers leading to segfaults prior to 1.1.10, see Additional info: * xorg-server 1.11.1.901-1 * using nvidia proprietary driver Steps to reproduce: * launch xdm * try to log in [1] https://bugs.archlinux.org/task/26561 |
This task depends upon
well, thanks. It was early morning when I wrote that. :)
on a sidenote neither wiki [1] nor XDM manual [2] mention dbus anywhere. I just checked the PKGBUILD, it somehow does depend on dbus-core, but not explicitly. I think being able to fully read an error message weighs more than the fact that a lightweight DSM shouldn't need dbus.
On a further note, here is how it is "working" with 1.1.10:
martti@deepthought:~$ rc.d list | grep dbus
[STOPPED][AUTO] dbus
martti@deepthought:~$ pgrep -l -U 1000 dbus
2498 dbus-launch
2499 dbus-daemon
so, my user starts dbus at the moment. Despite this being a single user machine, I must say I'm a tiny bit more comfortable this way, although I can't really tell what is making it work. xorg-xdm?
Anyway, you mean I should upgrade and put dbus into the real daemons?
[1] https://bugs.archlinux.org/task/26561
[2] http://www.xfree86.org/current/xdm.1.html
FS#22671). Looking at the patch we got from SuSE for this, this patch enables consolekit by default, and bails out if it can't register itself with consolekit. I could change the return value of the case where dbus/consolekit is not installed/running so it runs as if consolekit was disabled. To test that, you could change line 100 of the patch to "return 1" instead of "return 0" and rebuild the package. In that case, it should just print some dbus/consolekit errors, but continue to work as if consolekit was disabled.The segfault you see is another bug that is somewhere in X. X seems to segfault on exit in some situations, but that segfault is totally unrelated to your problems.