FS#34936 - [glib2] 2.36 causes hard hangs, and gdm wont start after reverting it

Attached to Project: Arch Linux
Opened by ... (spider007) - Wednesday, 24 April 2013, 16:59 GMT
Last edited by Jan de Groot (JGC) - Thursday, 23 May 2013, 06:47 GMT
Task Type Bug Report
Category Packages: Core
Status Closed
Assigned To Jan de Groot (JGC)
Jan Alexander Steffens (heftig)
Architecture x86_64
Severity High
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 5
Private No

Details

Description:

After upgrading my laptop it would hang hard after a reboot; even caps-lock light didn't work. After reverting glib2, glibmm and glib-networking from 2.36 to 2.34 the machine doesn't hang anymore, but gdm still wont start. The journal says:

Apr 24 18:52:35 spider007.spider007.net gdm[287]: GLib-GObject: gtype.c:2720: You forgot to call g_type_init()
Apr 24 18:52:35 spider007.spider007.net gdm[287]: GLib: g_once_init_leave: assertion `result != 0' failed
Apr 24 18:52:35 spider007.spider007.net gdm[287]: GLib-GObject: g_object_new: assertion `G_TYPE_IS_OBJECT (object_type)' failed
Apr 24 18:52:35 spider007.spider007.net gdm[287]: GLib-GObject: g_object_add_weak_pointer: assertion `G_IS_OBJECT (object)' failed
Apr 24 18:52:35 spider007.spider007.net kernel: gdm[287]: segfault at 18 ip 0000000000424d83 sp 00007fff911cad60 error 4 in gdm[400000+3c000]
Apr 24 18:52:35 spider007.spider007.net systemd[1]: gdm.service: main process exited, code=dumped, status=11/SEGV
Apr 24 18:52:35 spider007.spider007.net systemd-coredump[289]: Process 287 (gdm) dumped core.

Gdb reveals no useful info:

Core was generated by `/usr/sbin/gdm'.
Program terminated with signal 11, Segmentation fault.
#0 0x0000000000424d83 in ?? ()
(gdb) bt
#0 0x0000000000424d83 in ?? ()
#1 0x0000000000408891 in ?? ()
#2 0x00007fda09ddaa15 in __libc_start_main () from /usr/lib/libc.so.6
#3 0x0000000000408e65 in ?? ()
(gdb)

This might very well be related to  FS#34892 , but I didn't want to polute that since this seems different. How should I debug this further?
This task depends upon

Closed by  Jan de Groot (JGC)
Thursday, 23 May 2013, 06:47 GMT
Reason for closing:  Fixed
Additional comments about closing:  Recent versions of glib2 work fine here. Downgrading major versions without downgrading all other packages is not supported.
Comment by Jan de Groot (JGC) - Wednesday, 24 April 2013, 18:26 GMT
gdm needs 2.36.0, period. The message tells you that g_type_init is needed, since 2.35.x g_type_init is no longer needed and deprecated, so anything using glib has removed those calls.
Comment by Uğur (tty0) - Wednesday, 24 April 2013, 19:23 GMT
I've voted mistakenly. I've same problem with gdm -hard freeze- but i think this bug report doesn't suit me.
Comment by Martin Ziel (zilluss) - Wednesday, 24 April 2013, 19:44 GMT
I had a similar problem after upgrading my notebook. System froze at boot with gdm as the cause after upgrading to Gnome 3.8. There is a discussion about this on the forum. You may try the rc_6=0 fix suggested in the forum. This solved the problem for me, at least for now.

Link to the discussion regarding the bug: https://bbs.archlinux.org/viewtopic.php?id=161900&p=1
Comment by ... (spider007) - Thursday, 25 April 2013, 07:24 GMT
@jgc that explains the errors; but why doesn't any package have a dependency on >2.35? That would at least have prevented me from downgrading (or would have forced me to downgrade more)
@zilluss, thanks; I am reading that thread now
Comment by Jan de Groot (JGC) - Thursday, 25 April 2013, 12:16 GMT
Because we don't do versioned dependencies usually. The problem with versioned dependencies is that they don't work in reverse direction. If you do pacman -Sy eog on an outdated system for example, it will install a broken eog package because your gnome-desktop package wasn't updated. If it would be a versioned dependency, then eog would technically work, but the rest of your gnome installation would be fubar because old packages don't work with the new gnome-desktop package.
glib2 is an exception though, upstream wants to maintain backwards compatibility there.
Comment by ... (spider007) - Thursday, 25 April 2013, 12:58 GMT
Ah, I understand, thanks. Although I perform -Syu daily, issues like this are not easily resolved by the user because of these 'hidden dependencies' which forces me to wait for a new release from you guys or upstream :)
Comment by Jason Russell (jrussell) - Friday, 26 April 2013, 17:23 GMT
@Jan de Groot This might be a silly question, but how does arch get around that problem then? Or does it not? (Like in this case?)

Loading...