Liferea (Hardy) crashes on startup

[Tim Price]

Binary package hint: liferea

Liferea 1.4.14-0ubuntu4 on fully updated Hardy crashes almost immediately on startup (ie manages to draw window, but then disappears). Running liferea from console produces the following output:

"Obtaining the module object from Python failed.

Traceback (most recent call last):
cant import cStringIO
<type 'exceptions.ImportError'>: /usr/lib/python2.5/lib-dynload/time.so: undefined symbol: PyExc_ValueError

Liferea did receive signal 11 (Segmentation fault)."

This is my first bug report, so please let me know if more information is required or would be helpful. Thank you.

[Emilio Pozuelo Monfort (pochu)]

Tim Price wrote:
> Liferea did receive signal 11 (Segmentation fault)."

Can you please get a backtrace of the crash? You can find instructions on how to
do so at https://wiki.ubuntu.com/Backtrace

Thanks

[Tim Price (timprice)]

Thanks Emilio, I've attached a backtrace as requested. When I followed the instructions to which you linked, Liferea remained running (as the debugger noted at the end of the backtrace), but its window was greyed and unresponsive to any input I tried to make.

Thanks again

[Emilio Pozuelo Monfort (pochu)]

Thanks for the backtrace. It's a bit incomplete though, so could you install liferea-dbg from the repositories, and xulrunner-1.9-dbgsym from the ddebs repo (instructions are at https://wiki.ubuntu.com/DebuggingProgramCrash, or you can get it directly from http://ddebs.ubuntu.com/pool/main/x/xulrunner-1.9 )

This time when you start gdb use the following command to set LD_LIBRARY_PATH properly:
LD_LIBRARY_PATH=/usr/lib/xulrunner-1.9b5 gdb /usr/bin/liferea-bin
Then run it as before.

Also, if you move out (make first a backup since that contains all your feeds) ~/.liferea_1.4 and start Liferea, does it work or still crashes?
How are you starting liferea, via liferea or liferea-bin?

Thanks

[Tim Price (timprice)]

New backtrace attached, which I hope is more helpful (at least I can see that it's bigger this time, which I hope means it's more complete). I had already installed liferea-dbg prior to sending the earlier backtrace; prior to sending this new one I installed xulrunner-1.9-dbgsym directly (trying from the ddebs repo produced an unresolvable dependency).

I started gdb using:
LD_LIBRARY_PATH=/usr/lib/xulrunner-1.9b5 gdb /usr/bin/liferea-bin 2>&1 | tee gdb-liferea-bin.txt (which I hope was what your instructions intended)

Deleting the ~/.liferea_1.4 folder makes no obvious difference - the program still crashes (after regenerating the folder).

Running either liferea or liferea-bin from the console produces the crash just the same (I've been using liferea-bin with gdm since gdm advised me that liferea was not an executable).

Thanks again

[Emilio Pozuelo Monfort (pochu)]

Thanks, that one looks much better.

Alexander, could you look at this crash? I can't reproduce it, but the backtrace looks complete now.

I see this:
(gdb) backtrace full
#0 0x00000000 in ?? ()
No symbol table info available.

And you once said "this means that it calls a not loaded function.", so it would be great if you could have a look at it.
https://bugs.edge.launchpad.net/ubuntu/+source/liferea/+bug/207410/comments/17

[Emilio Pozuelo Monfort (pochu)]

What's the output of
dpkg -l "liferea*" "xulrunner-1.9*"

Your system is up-to-date as of this moment, right? Have you rebooted your computer lately? If not, does rebooting it make any difference?

[Emilio Pozuelo Monfort (pochu)]

Also, please attach the output of
strace -f -eopen liferea

[Alexander Sack (asac)]

and ... are you running any special accessibility tools like screen reader or such?

[Tim Price (timprice)]

Thanks Emilio and Alexander.

Output of
dpkg -l "liferea*" "xulrunner-1.9*":

Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Installed/Config-f/Unpacked/Failed-cfg/Half-inst/t-aWait/T-pend
|/ Err?=(none)/Hold/Reinst-required/X=both-problems (Status,Err: uppercase=bad)
||/ Name Version Description
+++-==============-==============-============================================
ii liferea 1.4.14-0ubuntu feed aggregator for GNOME
ii liferea-dbg 1.4.14-0ubuntu liferea debug symbols
un liferea-gtkhtm <none> (no description available)
un liferea-mozill <none> (no description available)
un liferea-xulrun <none> (no description available)
ii xulrunner-1.9 1.9~b5+nobinon XUL + XPCOM application runner
ii xulrunner-1.9- 1.9~b5+nobinon debug symbols for package xulrunner-1.9
ii xulrunner-1.9- 1.9~b5+nobinon Support for Gnome in xulrunner-1.9 applicati
ii xulrunner-1.9- 1.9~b5+nobinon debug symbols for package xulrunner-1.9-gnom

Do I perhaps need one of the three uninstalled ones, such as liferea-xulrun?

My system is up to date (I've been keeping up with Hardy packages as they appear, so just installed another load, which were mostly KDE PIM stuff). I reboot after each new batch of packages, whether this is required or not, though this has not made any difference to this problem.

I've attached the output of strace -f -eopen liferea.

I'm not running any accessibility tools, but this question is interesting, as I did recently install orca (though as I say I'm not running it). Perhaps I should uninstall it and see if that makes a difference (?).

[Tim Price (timprice)]

A quick follow-up - I removed orca, but doing so made no difference to this crash.

[Tim Price (timprice)]

1. I noticed this morning that new bug #215658 describes the same error message I noted receiving at the beginning of this bug report.

2. I apologise if I'm pointing out something obvious and irrelevant here, but I also discovered today that running liferea using sudo (either "liferea" or "liferea-bin") allows it to run properly (and also resolves the minor problem of transparency of the icon in the notification bar, which I can see is not transparent when not using sudo, in the brief moment before the liferea window crashes).

[Jean-Baptiste Lallement (jibel)]

I can reproduce a similar case. Not a crash at startup but same effect (segfault) and may be same root case.

I can see:
(gdb) backtrace full
[...]
#0 0x00000000 in ?? ()
No symbol table info available.

To reproduce it :
1. Launch liferea
2. Identify an article with a flash video (a video stream not an animation, something like youtube or dailymotion )
3. Select the feed publishing this article
4. Right click on the headline and select "Launch Item In Tab"
5. Go to the tab
6. Start the video
7. Close the tab
=> Segfault

Example:
Feed: http://www.rue89.com/homepage/feed
Item : http://www.rue89.com/2008/04/24/demander-plus-pour-gagner-plus-le-retour-des-greves-salariales

It happens :
- only when closing the first tab.
- only when setting LD_LIBRARY_PATH=/usr/lib/xulrunner-1.9b5/

$ dpkg -l "liferea*" "xulrunner*"

ii liferea 1.4.14-0ubuntu4
ii liferea-dbg 1.4.14-0ubuntu4
ii xulrunner-1.9 1.9~b5+nobinonly-0ubuntu3
ii xulrunner-1.9-dbgsym 1.9~b5+nobinonly-0ubuntu39
ii xulrunner-1.9-dev 1.9~b5+nobinonly-0ubuntu3
ii xulrunner-1.9-gnome-support 1.9~b5+nobinonly-0ubuntu3
ii xulrunner-1.9-gnome-support-dbgsym 1.9~b5+nobinonly-0ubuntu3

Let me know if you think this is a totaly different case and I'll open a new report.

[Emilio Pozuelo Monfort (pochu)]

Jean-Baptiste, please check this comment, https://bugs.launchpad.net/ubuntu/+source/liferea/+bug/215016/comments/3, and make a new backtrace. Thanks!

[Jean-Baptiste Lallement (jibel)]

Sorry, some missing dbgsym packages :(

This backtrace looks better. But this case is apparently totally different from the one Tim is facing.

[Leon (leonbo)]

I've got the same problems. Yesterday it worked fine it seemed.

[Megaqwerty (matthewb)]

I am also experiencing the exact same problem, but it was working for me a day ago.

[Kunzahe (kunzahe+ubuntu)]

I have the same problem, but only if I have Compiz activated.

Sorry about my English, probably I have writing mistakers.

[Kunzahe (kunzahe+ubuntu)]

My last message isn't true.

If I have Compiz activated I have the same error, but Liferea doesn't crash.

[Emilio Pozuelo Monfort (pochu)]

Could someone verify if this is fixed with xulrunner-1.9 rc1 from hardy-proposed?

[Kunzahe (kunzahe+ubuntu)]

It isn't fixed with xulrunner-1.9 rc1.

[primefalcon (primefalcon)]

having same problem, autocloses on open from gnome, I tried the terminal and got this

Obtaining the module object from Python failed.

Traceback (most recent call last):
cant import cStringIO
<type 'exceptions.ImportError'>: /usr/lib/python2.5/lib-dynload/time.so: undefined symbol: PyExc_ValueError

Liferea did receive signal 11 (Segmentation fault).

[primefalcon (primefalcon)]

also rebooting make sno difference and my computer is fully updated

[Sig (sig-akasig)]

Possible work-around :
I confirm what tim said above (https://bugs.launchpad.net/ubuntu/+source/liferea/+bug/215016/comments/11) : running liferea using sudo avoids the crash :

- without sudo => it crashes at startup

- if I just "sudo liferea" then it does not crash but it does not display my feeds either (I guess it uses or even creates some root profile which starts with default example feeds)

- if I "sudo -u myusername liferea" then it does not crash and it uses my feeds ! and the liferea icon remains visible

[primefalcon (primefalcon)]

gksudo -u myusername liferea works for me, sudo for some reason actually still crashed me

[primefalcon (primefalcon)]

Oh I added that line to my main menu and it seems to be ok for now too

[gil (gilbert-laycock)]

For me the "gksudo -u <uid> liferea" technique worked once, but only once.
On a couple of other machines I have liferea working fine; I'm not sure what is different about the one that doesn't work.

[primefalcon (primefalcon)]

the gksudo stopped working for me too but I just changed the gksudo to sudo and well, it's working again, sigh

[Leon (leonbo)]

I enabled the proposed repository and update everything. Now I get this output:

leon@polly:~$ liferea
Obtaining the module object from Python failed.

Traceback (most recent call last):
cant import cStringIO
<type 'exceptions.ImportError'>: /usr/lib/python2.5/lib-dynload/time.so: undefined symbol: PyExc_ValueError
Segmentation fault
leon@polly:~$

[Leon (leonbo)]

Sorry, here's the backtrace. It ends with:

0x00007f7578ff80fc in _nss_wins_gethostbyname_r () from /lib/libnss_wins.so.2

[Leon (leonbo)]

When I remove the "wins" option from /etc/nsswitch.conf it works! Weird :s

[Emilio Pozuelo Monfort (pochu)]

There is 3 different kind of backtraces here...

The original one, a Xulrunner issue (either Xulrunner's fault or Liferea's fault), comments 4, 18.
A crash from flashplugin-nonfree, comment 16.
A crash from checking /etc/nsswitch.conf, comments 17, 31. This is bug #214192, so please if you have this bug comment there and not here.

Now I wonder what bug workarounds using sudo (probably 214192).

[Jean-Baptiste Lallement (jibel)]

Can someone check if the original bug is fixed with the latest version in Intrepid ?

Thanks

[primefalcon (primefalcon)]

I have no idea about intrepid, but I'd forgotten all about this until I got this email notification about a post here. But it seems fixed in Hardy

I honestly forgot I had that workaround after all this time lol

[jedioetzi (jedioetzi)]

in Hardy I don't have any "wins" option in /etc/nsswitch.conf, and with/without compiz liferea crashes with

Liferea did receive signal 11 (Segmentation fault).
Obtaining the module object from Python failed.

Traceback (most recent call last):
cant import cStringIO
<type 'exceptions.ImportError'>: /usr/lib/python2.5/lib-dynload/time.so: undefined symbol: PyExc_ValueError

there is a wordkaround?
is not a python problem?

[jedioetzi (jedioetzi)]

maybe is important: I have both python 2.5 and python 2.4 installed.

[Alexander Sack (asac)]

can you try to build the intrepid liferea and see if its gone?

[Domenico Urbano (domurb)]

> When I remove the "wins" option from /etc/nsswitch.conf it works! Weird :s

Thanks Leon, that solved the problem for me. Starting liferea from console it gives the same errors than before now, but it doesn't segfault. The errors are:
-------------
libnm_glib_nm_state_cb: dbus returned an error.
  (org.freedesktop.DBus.Error.ServiceUnknown) The name org.freedesktop.NetworkManager was not provided by any .service files
Obtaining the module object from Python failed.

Traceback (most recent call last):
cant import cStringIO
<type 'exceptions.ImportError'>: /usr/lib/python2.5/lib-dynload/time.so: undefined symbol: PyExc_ValueError
-------------

I think in the first error it complains for the absense of network manager (i use wicd) and the second i have no idea. Anyway the two errors are not related with the segfault, in my case it depends only from wins beeing enabled in nsswitch.

[Alexander Sack (asac)]

On Thu, Sep 11, 2008 at 11:44:35PM -0000, Domenico wrote:
>
> > When I remove the "wins" option from /etc/nsswitch.conf it works!
> Weird :s
>
> Thanks Leon, that solved the problem for me. Starting liferea from console it gives the same errors than before now, but it doesn't segfault. The errors are:
> -------------
> libnm_glib_nm_state_cb: dbus returned an error.
> (org.freedesktop.DBus.Error.ServiceUnknown) The name org.freedesktop.NetworkManager was not provided by any .service files
> Obtaining the module object from Python failed.
>
> Traceback (most recent call last):
> cant import cStringIO
> <type 'exceptions.ImportError'>: /usr/lib/python2.5/lib-dynload/time.so: undefined symbol: PyExc_ValueError
> -------------
>
> I think in the first error it complains for the absense of network
> manager (i use wicd) and the second i have no idea. Anyway the two
> errors are not related with the segfault, in my case it depends only
> from wins beeing enabled in nsswitch.
>

Please open an individual bug about liferea being broken when no
network manager is installed.

 - Alexander

[Godlike (godlike-panos-hot)]

Hello,

In my case the crash happened after inserting "wins" to /etc/nsswitch.conf (just like jedioetzi)

I solve the crash by changing the position of "wins" inside the /etc/nsswitch.conf:

[...]
hosts: files mdns4_minimal [NOTFOUND=return] dns mdns4 wins
[...]

[busymind (busymind)]

Same here with wins
Today I inserted "wins" in my nsswitch.conf and my lifierea got crash. The solution of Godlike seems working good.

[Emilio Pozuelo Monfort (pochu)]

This is a duplicate of bug #215016. I don't have any clue on how it can be fixed. Any help is appreciated.

[Emilio Pozuelo Monfort (pochu)]

Or at least the last two comments are the same as bug 215016. I'm not sure if all the other backtraces in this report are about nsswitch :-)

[busymind (busymind)]

Sorry about to fast bug reporting. I wrote about changing the nsswitch.conf the position of wins. Seems yesterday working ok about 5 houres. Today i have same problems reported by others - crash after starting or random.

[Alexander Sack (asac)]

On Thu, Nov 13, 2008 at 07:29:19PM -0000, Emilio Pozuelo Monfort wrote:
> This is a duplicate of bug #215016. I don't have any clue on how it can
> be fixed. Any help is appreciated.
>

Do we know how to reproduce this?

 - Alexander

[Emilio Pozuelo Monfort (pochu)]

Alexander Sack wrote:
> On Thu, Nov 13, 2008 at 07:29:19PM -0000, Emilio Pozuelo Monfort wrote:
>> This is a duplicate of bug #215016. I don't have any clue on how it can
                               ^^^^^^
                    Sorry, I meant bug 214192.

>> be fixed. Any help is appreciated.
>>
>
> Do we know how to reproduce this?

There are two different backtraces in this bug report. The one in comment 4,
which looks xulrunner related, the one in comment 12, likely a flash bug, and
the one in comment 31, which is bug 214192.

I've had Liferea crashes when opening articles with embedded flash videos. The
other two, I haven't had them, but perhaps the nsswitch can be reproducing by
adding wins to /etc/nsswitch.conf (see bug 214192, comment 8)

[Jean-Baptiste Lallement (jibel)]

The flash bug in comment 12 is probably a dup of bug 255393 and bug 255595. It is still an issue and easily reproduced in Jaunty.

[Alexander Sack (asac)]

my guess is that comment 4 crash should be gone by now. the other cases (like flash) are dealt elsewhere already.