| Summary: | Bubble click callbacks are not fired | ||
|---|---|---|---|
| Product: | Mageia | Reporter: | Colin Guthrie <mageia> |
| Component: | RPM Packages | Assignee: | Thierry Vignaud <thierry.vignaud> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | Normal | CC: | mageia |
| Version: | Cauldron | ||
| Target Milestone: | --- | ||
| Hardware: | i586 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Source RPM: | libnotify, perl-Gtk3-Notify, mgaonline | CVE: | |
| Status comment: | |||
| Bug Depends on: | |||
| Bug Blocks: | 11778 | ||
| Attachments: | WIP but glib/gtk create threads behind our back and segfault | ||
|
Description
Colin Guthrie
2013-12-18 18:27:56 CET
Thierry Vignaud
2013-12-19 10:02:07 CET
Source RPM:
mgaonline =>
libnotify, perl-Gtk3-Notify, mgaonline Any message if you run mgaapplet from the command line? Keywords:
(none) =>
NEEDINFO Will have a look, but normally I just check "journalctl -f" as a user to see all error output from processes started in the user session and it's got nothing in it, so I suspect not. One observation I did make is that perl-Gtk3-Notify package has no binary data in it unlike perl-Gtk2-Notify. As a slightly educated guess I'd say this is due to the gobject introspection that is possible in Gtk3 and thus the binary bit isn't needed... I got lost looking at the code from then on - perl is not my strong point as you know :) Will let you know tho'. Yup, as suspected, there is no further output on the terminal if run manually. Testing under gnome, didn't try other desktops. But how could have such notifications in cauldron? We explicitely disable them on Cauldron... I misread updates for upgrade
Thierry Vignaud
2013-12-21 00:08:45 CET
Status:
NEW =>
ASSIGNED Created attachment 4652 [details]
WIP but glib/gtk create threads behind our back and segfault
Which I've a fix for: since we cannot prevent glib/gtk to spawn threads behind our back, we can at least try to prevent segfaults due to mixing threads with secular forks by exec()ing immediately I though I'd a fix but even using run_program::get_stdout instead of `ip addr` in network::tools cause the segfault. Once commented, mgaapplet no more crash. I suspect this can explain quite some net_applet crashes too (same issue regarding libnotify and glib/gtk: those create threads and we mix fork()s with them...) CC:
(none) =>
mageia Too bad https://developer.gnome.org/gio/unstable/GSubprocess.html will only be in next major version of glib commit 67cbacd2c1cc61fcc8a4cb8ca2d567bebec7a814
Author: Thierry Vignaud <thierry.vignaud@...>
Date: Sat Dec 21 03:09:24 2013 +0100
prevent segfaulting b/c of fork vs threads (mga#12041)
the issue is that glib/gtk create threads behind us
however threads & fork() don't behave well when mixed...
---
Commit Link:
http://gitweb.mageia.org/software/drakx-net/commit/?id=67cbacd2c1cc61fcc8a4cb8ca2d567bebec7a814
Segfault fixed in drakx-net-2.6. I suspect that I've closed quite a lot of the stange net_applet segfaults we saw along the years. See https://bugzilla.redhat.com/show_bug.cgi?id=835112 & http://permalink.gmane.org/gmane.comp.bug-tracking.abrt.devel/2633 for another fork() vs hidden glib threads segfault. I'll issue a new mgaonline too. Fixed in git Status:
ASSIGNED =>
RESOLVED commit 95b9cd06f14a9817090584d72830df870c591acc
Author: Thierry Vignaud <thierry.vignaud@...>
Date: Sat Dec 21 03:17:55 2013 +0100
fix doing nothing when clicking on notifications buttons (mga#12041)
---
Commit Link:
http://gitweb.mageia.org/software/mgaonline/commit/?id=95b9cd06f14a9817090584d72830df870c591acc
Did you mean to do that "More Information2" thing in the above commit? Seems like it was just for testing... |