Bug 2679

This bug doesn't appears now it seems it has appeared when there was a bad driver for my graphic card. See bug 3031

*** Bug 2696 has been marked as a duplicate of this bug. ***

CC: (none) => grenoya

@ Jacques

Thanks for the feedback, but I'll leave this bug report open for the reporter of a duplicate bug 2696

CC: (none) => marja11

*** Bug 2915 has been marked as a duplicate of this bug. ***

CC: (none) => pietrotux88

@ grenoya
@ pietrotux

We have version 0.99-1.mga2  for drakx-net-applet now. is this bug still valid?

Keywords: (none) => NEEDINFO

I tried to do the same thing than the first time : changing style of the xfce desktop.

The problem still exists !
net_applet and perl version have changed.
I really think it has nothing to do with net_applet: I was changing xfce style, Jacques put it's cursor over an icone, did not have time to clic and the bug report was there.
Two different DE are concerned : xfce and kde.

@ pietrotux : could you please tell us what you were doing went the bug report appeared ?

@ marja : what about include some packager of DE and DE style in the circle ?

If one of you can reproduce this bug, then please:

1) enable the core debug medium, then

2) install at least the following packages: glibc-debug perl-debug perl-Glib-debug perl-Gtk2-debug glib2.0-debug gtk+2.0-debug gdb

3) open a terminal

4) run "killall net_applet; gdb -q --args perl /usr/bin/net_applet"

5) type "run" 

6) once it crashed, type "bt" and copy the stack trace in a file you'll attach here

CC: (none) => thierry.vignaud

Ok, I did what asked Tv and here is the backtrace. Hope it will help you :)

#0  0x00007fffe38648fa in std::_Rb_tree_decrement(std::_Rb_tree_node_base*) () from /usr/lib64/libstdc++.so.6
#1  0x00007fffe3bc2a29 in std::_Rb_tree<_GtkWidget*, std::pair<_GtkWidget* const, Oxygen::ShadowHelper::WidgetData>, std::_Select1st<std::pair<_GtkWidget* const, Oxygen::ShadowHelper::WidgetData> >, std::less<_GtkWidget*>, std::allocator<std::pair<_GtkWidget* const, Oxygen::ShadowHelper::WidgetData> > >::_M_insert_unique(std::pair<_GtkWidget* const, Oxygen::ShadowHelper::WidgetData> const&) () from /usr/lib64/gtk-2.0/2.10.0/engines/liboxygen-gtk.so
#2  0x00007fffe3bc22bf in Oxygen::ShadowHelper::registerWidget(_GtkWidget*) ()
   from /usr/lib64/gtk-2.0/2.10.0/engines/liboxygen-gtk.so
#3  0x00007fffe3bc23b3 in Oxygen::ShadowHelper::realizeHook(_GSignalInvocationHint*, unsigned int, _GValue const*, void*) ()
   from /usr/lib64/gtk-2.0/2.10.0/engines/liboxygen-gtk.so
#4  0x00007ffff2be9d12 in signal_emit_unlocked_R (node=<optimized out>, detail=0, instance=0x4010b30, emission_return=0x0, 
    instance_and_params=0x16bf8e0) at gsignal.c:3238
#5  0x00007ffff2bf1a87 in g_signal_emit_valist (instance=<optimized out>, signal_id=<optimized out>, detail=0, var_args=
    0x7fffffffd230) at gsignal.c:3003
#6  0x00007ffff2bf1c43 in g_signal_emit (instance=<optimized out>, signal_id=<optimized out>, detail=<optimized out>)
    at gsignal.c:3060
#7  0x00007fffef1d6b3e in IA__gtk_widget_realize (widget=0x4010b30 [GtkWindow]) at gtkwidget.c:3543
#8  0x00007fffef1e17c8 in gtk_window_show (widget=0x4010b30 [GtkWindow]) at gtkwindow.c:4532
#9  0x00007ffff2bd8fb4 in g_closure_invoke (closure=0x1491510, return_value=0x0, n_param_values=1, param_values=0x42f1ae0, 
    invocation_hint=<optimized out>) at gclosure.c:774
#10 0x00007ffff2bea50e in signal_emit_unlocked_R (node=<optimized out>, detail=0, instance=0x4010b30, emission_return=0x0, 
    instance_and_params=0x42f1ae0) at gsignal.c:3202
#11 0x00007ffff2bf1a87 in g_signal_emit_valist (instance=<optimized out>, signal_id=<optimized out>, detail=0, var_args=
    0x7fffffffd6e0) at gsignal.c:3003
#12 0x00007ffff2bf1c43 in g_signal_emit (instance=<optimized out>, signal_id=<optimized out>, detail=<optimized out>)
    at gsignal.c:3060
#13 0x00007fffef1d7b36 in IA__gtk_widget_show (widget=0x4010b30 [GtkWindow]) at gtkwidget.c:3225
#14 0x00007fffef18dedc in gtk_tooltip_position (new_tooltip_widget=<optimized out>, display=<optimized out>, tooltip=
    0x16d4a80 [GtkTooltip]) at gtktooltip.c:1211
#15 gtk_tooltip_show_tooltip (display=<optimized out>) at gtktooltip.c:1288
#16 0x00007fffef18e22d in tooltip_popup_timeout (data=<optimized out>) at gtktooltip.c:1371
#17 0x00007fffeece7356 in gdk_threads_dispatch (data=0x423f620) at gdk.c:512
#18 0x00007ffff271b1cb in g_timeout_dispatch (source=0x4299b00, callback=<optimized out>, user_data=<optimized out>)
    at gmain.c:3731
#19 0x00007ffff271a59a in g_main_dispatch (context=0x1058b20) at gmain.c:2387
#20 g_main_context_dispatch (context=0x1058b20) at gmain.c:2924
#21 0x00007ffff271a958 in g_main_context_iterate (context=0x1058b20, block=1, dispatch=1, self=<optimized out>)
    at gmain.c:2995
#22 0x00007ffff271ad3a in g_main_loop_run (loop=0x16d0bd0) at gmain.c:3189
#23 0x00007fffef0b1427 in IA__gtk_main () at gtkmain.c:1256
#24 0x00007fffef6aba49 in XS_Gtk2_main () from /usr/lib/perl5/vendor_perl/5.14.1/x86_64-linux-thread-multi/auto/Gtk2/Gtk2.so
#25 0x00007ffff7b17680 in Perl_pp_entersub (my_perl=0x602010) at pp_hot.c:3046
#26 0x00007ffff7b0e7e6 in Perl_runops_standard (my_perl=0x602010) at run.c:41
#27 0x00007ffff7aae0b6 in S_run_body (oldscope=<optimized out>, my_perl=<optimized out>) at perl.c:2345
#28 perl_run (my_perl=0x602010) at perl.c:2268
#29 0x0000000000400f34 in main (argc=2, argv=0x7fffffffdd88, env=0x7fffffffdda0) at perlmain.c:120
quit
A debugging session is active.

        Inferior 1 [process 15684] will be killed.

Quit anyway? (y or n)

No you did not :-(
I said "a file you'll attach here" (not paste)
But thanks for the GDB trace :-)
At least, for once in a while, a segfault that can be reproduce with a reporter that can give the needed data :-)

Wherewas we're good to pinpoint the bugs when we crashed due to perl errors thanks to the drakbug trace, when we segfault, the trace is not usefull because gdb get attached too late and the stack has been "cleaned" up.
I wonder how many of the segfault reports we got are actually theme engines bugs?
BTW I should add the theme info into drakbug reports...

Since we're crashing in libstdc++, could you install gcc-debug too and get a new back trace (to attach, not paste please)?

Thanks.

Summary: net_applet segfaulted => net_applet segfaulted in Oxygen
Source RPM: drakx-net-applet-0.99-1.mga2 => oxygen-gtk

Created attachment 1099 [details]
liboxygen crash

Created attachment 1100 [details]
libstdc++ crash

I installed gcc-debug and made some more crash-tests.
After making net_applet, mgaapplet, xfce4-apprearance and xfce4-panel crash all together and after several X reboot and global reboot,
I realized that I have 2 different crash messages => 2 backtraces.

maybe they are redundant but better 2 than none ! :)

and maybe there are more crash messages, but I will wait for you to be really really sure that you need them before searching for them :D

I hope it will help

Don't forget to fill in the version of lib(64)oxygen-gtk you're using

(In reply to comment #12)
> Don't forget to fill in the version of lib(64)oxygen-gtk you're using

Yes please, I updated it to 1.1.5 this week.

Here it is, sorry to have forgotten that

[grenoya@silenus ~]$ rpm -qa | grep oxygen
lib64oxygen-gtk-1.1.5-1.mga2
oxygen-gtk-1.1.5-1.mga2
oxygen-gtk3-1.1.50-0.20110922.1.mga2
oxygen-icon-theme-4.7.3-3.mga2

Hi, 
is there any news ?
oxygen-gtk and oxygen-icon-theme have changed version but I still have mgaapplet and net_applet crashs.
The only improvement I can see, but not the least, is that xfce4-* do not crashing anymore :)

I just rebuild the theme with new gcc-4.6, this might help since it's crashing into libstdc++ (from gcc)

Juan, could you see with upstream about the traces?

Assignee: bugsquad => juan.baptiste

(In reply to comment #17)
> Juan, could you see with upstream about the traces?

I took a look at upstream bugzilla and found this report 

https://bugs.kde.org/show_bug.cgi?id=285213

that seems to have the same trace as the one on "libstdc++_crash" attachment. There are two traces on that report, look at the second one.

I just updated oxygen-gtk to 1.2.0, can someone please test and see if the issue persist ?

Thanks

(sorry for the noise don't know what happen :/)

Blocks: (none) => 4300

Hi,
sorry to answer so late...

today in my other laptop (cauldron too), while trying to change appearence in xfce (xfce was really ugly) it made crash : xfce-appearence, mgaapllet and net_applet :\
No automatic report pops up and as it was graphical manipulation, I have no backtrace nor error message.

Here is was is installed on this laptop :
[grenoya@enee ~]$ rpm -qa | grep oxygen
oxygen-icon-theme-4.8.0-1.mga2
lib64oxygenstyle4-4.8.0-1.mga2
oxygen-gtk-1.2.0-2.mga2
lib64oxygenstyleconfig4-4.8.0-1.mga2
lib64oxygen-gtk-1.2.0-2.mga2


feel free to ask more if needed :)

Please test new oxygen rpm.

CC: (none) => dmorganec

Hi, tested the evening after update and reboot ... it's still crashing :\

@claire
Just to make sure I understand things properly, 
Your multiple crash occurs only when you try changing the appearance to oxygen-gtk. Correct ? 
If you logout and login after doing that, is the theme being used oxygen-gtk ? 
(and then you do not have any more crashes) ? 
Or is it still the old theme ? 

Thanks,

Hugo (Oxygen Dev)

Hi Hugo, happy to see upstream guy here :)

(In reply to comment #24)
> @claire
> Just to make sure I understand things properly, 
> Your multiple crash occurs only when you try changing the appearance to
> oxygen-gtk. Correct ? 
No it occurs when I try to change from oxygen-gtk to another one (Ia ora or Raleigh). At the moment I click on the theme, xcfe-appereance disapeared, as mgaapllet and net_applet icons (plus other little graphical pb).
I can restart xfce-appereance and choose back oxygen-gtk without any problem.(even without reloging or rebooting)

> If you logout and login after doing that, is the theme being used oxygen-gtk ? 
> (and then you do not have any more crashes) ? 
> Or is it still the old theme ? 
What ever is the choosen theme, it is still there after reboot or reloging.

The point I may not have been clear is that "xfce-apereance" is just the interface for choosing theme. xfce-panel and xfwm are not crashing anymore since 10 of December.

> 
> Thanks,
Thank *you* for your work on this bug :)

Claire

*** Bug 4694 has been marked as a duplicate of this bug. ***

CC: (none) => gruescubogdan

*** Bug 4743 has been marked as a duplicate of this bug. ***

CC: (none) => herbert

*** Bug 4657 has been marked as a duplicate of this bug. ***

CC: (none) => liste

*** Bug 4658 has been marked as a duplicate of this bug. ***

*** Bug 4790 has been marked as a duplicate of this bug. ***

CC: (none) => bigdavesr

@Hugo,

Have had the time to take a look at this bug ?

In fact I'am not sure that the 4-5 latest duplicate are really duplicate or related to oxygen-gtk but the crash was the same as another one. :/

Theme name: bubble for the 4790
Theme name: Ia Ora Smooth for the 4658, 4657
Theme name: Adwaita for the 4743

(In reply to comment #32)
> In fact I'am not sure that the 4-5 latest duplicate are really duplicate or
> related to oxygen-gtk but the crash was the same as another one. :/
> 
> Theme name: bubble for the 4790
> Theme name: Ia Ora Smooth for the 4658, 4657
> Theme name: Adwaita for the 4743

They may, as when i try to change xfce theme from oxygen to something else, it crashes and drakbug gives me the new theme name.

my 2 cents

I'm quite certain its related to oxygen-gtk.
Crash happen when changing from oxygen-gtk to "something else".
My guess is that there is something "wrong" (or not properly handled) at *unloading* the oxygen theme, and in the deletion chain. 
Sadly enough, I cannot reproduce here, though I am unable to change theme with gnome-control-center. I use KDE, and changes theme either using gtk-chtheme, or kde's system-setting. When doing so, I experience no issue when switching back and forth between various themes ... 
Will investigate further, but so far, no clue ...

Hugo

PS: what's the name of the xfce application used to change widget theme ?

(In reply to comment #35)
> PS: what's the name of the xfce application used to change widget theme ?

xfce4-appearance

mmm. here, running xfce4-appearance (in my kde session) and selecting another style does not change anything :( (and doesn't crash either, but ...)
I'll try to log in xfce later today. 
Will keep you posted.

Ok thanks for the info, so we have still one or two duplicate
/me will try to find them.

I am running gnome classic just did update. I am now stuck on oxygen theme and can not change it at all. I also cant change display settings.Both system settings and advanced settings or not working.

@dave
Well I guess you can always edit manually the file $HOME/.gtkrc-2.0 (or similar)
and change the loaded theme here.

As for display settings, although I am not sure what you refer to, this is probably totally unrelated to this bug report.
Also, actually, does the "stuck to", and "cannot change" ... means that whenever you try to change, things start crashing ? Or is it just that the change is not applied ? 
If the latter is true, then it is likely unrelated too, and you should file another bug report.

Cheers,

Hugo

(In reply to comment #34)
> Sadly enough, I cannot reproduce here, though I am unable to change theme with
> gnome-control-center. 

the theme (when using gnome3) is changed with the 'gnome-tweak-tool' and not 'gnome-control-center' ... the tool is labeled 'Advanced Settings' in the GUI

*** Bug 4823 has been marked as a duplicate of this bug. ***

see https://bugs.kde.org/show_bug.cgi?id=275366
(the bug is somewhat different, but I believe the underlying issue is the same).

some commits have been made that might improve the situations, though it is not yet fully completed.
I hope to have something ready in the comming days, and pushed to the next minor releases of oxygen-gtk (namely 1.2.2 and gtk3-0.0.2)

Can someone give a shot to oxygen-gtk2-1.2.2 
(http://download.kde.org/download.php?url=stable/oxygen-gtk2/1.2.2/src/oxygen-gtk2-1.2.2.tar.bz2)

and/or oxygen-gtk3-1.0.2
(http://download.kde.org/download.php?url=stable/oxygen-gtk3/1.0.2/src/oxygen-gtk3-1.0.2.tar.bz2)

Problems may (or may not) have been fixed in these releases
(at least I tried).

Thanks,

Hugo

It get better with the new oxygen-gtk packages (tested in Gnome): no segfault in mgaapplet and net_applet, but after switching back from oxygen-gtk the theme doesn't revert cleanly (a log out and log in is needed). Thanks Hugo, hope you will succeed to find a fix for this glitch as well.

It looks perfect for me ! 
No more crashing \o/

For me the bug can be closed, as it was about crashes of applets and xfce-appearence.

I didn't see any problem of them not coming back completely, but i am not difficult with that as long as it doesn't block freeze anything :)

@Hugo :
Thank you very much for all the time you spend on mageia's bugs about oxygen. It was a very good job and I can't imagine all the time you spent on it. :)

Created attachment 1771 [details]
Screenshot of Oxygen Theme

This is with Mageia 2 Beta 1 updated to latest Cauldron.

switching to it crashed gnome-shell ... I let you be the judge whether the theme is perfect or not :)

@Herbert.

This is a different bug as far as I know.
Only related to oxygen-gtk3 
and depending on the gtk3 version installed.

The style looking admittingly ugly is due to regressions in latest gtk3 releases (minor versions, not even major)
Since a large number of regressions come and go from one gtk3 minor release to the other, it is impossible for us to keep the pace, and have oxygen work for all of these. (things basically keep getting broken then fixed upstream).

So as far as I am concerned, I can only focus on having oxygen-gtk3 working on gtk3 +stable+ release, which, for the moment, is gtk 3.2 (see: http://www.gtk.org/download/linux.php)

I think none of the bugs seen on your screenshot are actually visible with this release. (and not visible either with several sub-releases of gtk3.3)

so ... Mageia 2 will ship with gtk 3.2?
If so, why are we testing 3.3.18 in Cauldron?
If not, what relevance has gtk 3.2 for Cauldron/Mageia 2?

Hugo, for me oxygen-gtk2-1.2.2 is not a disaster in terms of regressions (no crash in mcc when installing or removing rpms and I don't experience Herbert's problems either). I've got no crash whatsoever, although my test was limited.
I use systemd and the radeon free driver, if that matters.
gtk 3.3.x is needed as build require for new gnome stuff and gnome 3.4 will soon be available (probably accompanied by a new gtk stable version).
However, for me there are still some visual inconveniences, most notable are that nautilus looks very bad (almost unusable) and when switching back from oxygen-gtk a few decorations does not revert.
I don't say that the problems reported by others are not legitimate, only that from me, except the nautilus issue (which might be fixed with a new nautilus version), oxygen-gtk works acceptable.

oxygen-gtk has been reverted to 1.2.1. It fixes crash in drakconnect

There will be a new release of oxygen gtk (either 1.2.3, or 1.2.2.1) hopefully by friday. It should fix both the crash with drakconnect and this report.
(at least that's what my tests here say)

(In reply to comment #52)
> oxygen-gtk has been reverted to 1.2.1. It fixes crash in drakconnect

That doesn't solve anything.
You're just trading one issue for others.

See last comment before yours

Could you please test with last version of oxygen-gtk updated this morning ?

I've done some very simple tests and with the new oxygen-gtk package I was not able to crash anything (including drakconnect, mgaapplet, net_applet) - it's looking good.
In comparison with the previous version, the visual experience didn't change too much.

As a side note, could someone please update the package gnome-tweak-tool once a stable version will be available? (Jani Valimaa is officially its maintainer)

no crashes here, but I'd like to emphasize that once you selected oxygen as theme, you cannot properly go back to the default (Adwaita) because even if you set all theme selectors back, most oxygen attributes remain (strange colors, odd sized buttons, etc). restarting the gnome theme engine (ALT-F2 'rt') or the entire gnome shell (ALT-F2 'r') doesn't fix the issues, only logging out and back in will.

ok closing it for now then. Please open a new bug report for any other issue

Status: NEW => RESOLVED
Resolution: (none) => FIXED