Crashes due to freeing unused mutex

network manager crash log

is this a syslog from booting or druing/after the update?

Also, please give us info how to reproduce.

My interfaces file. I only want my wired connection managed with NM. Guess it is well configured for that purpose, but it might help diagnosing this.

To j^:

It happened on both. But I should add that I had the unnofficial 0.6 packages installed before. Yet I guess I updated all the needed packages.

To Mario:

To reproduce, only reboot :-)

Are you using master server archive for packages?

my sources.list

I've got a similar prolem with NM 0.6.1:
~$ sudo killall NetworkManager
NetworkManager: no process killed

:~$ sudo NetworkManager --no-daemon
NetworkManager: <information> starting...
NetworkManager: <WARNING> main (): nm_data_new: Setting up dbus filter
NetworkManager: <information> Deactivating device eth0.
NetworkManager: <WARNING> nm_signal_handler (): Caught signal 11. Generating backtrace...
NetworkManager: ******************* START **********************************
NetworkManager: (no debugging symbols found)
NetworkManager: Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1".
NetworkManager: (no debugging symbols found)
NetworkManager: (no debugging symbols found)
NetworkManager: (no debugging symbols found)
NetworkManager: (no debugging symbols found)
NetworkManager: (no debugging symbols found)
NetworkManager: (no debugging symbols found)
NetworkManager: [Thread debugging using libthread_db enabled]
NetworkManager: [New Thread -1211578688 (LWP 5612)]
NetworkManager: (no debugging symbols found)
NetworkManager: (no debugging symbols found)
NetworkManager: (no debugging symbols found)
NetworkManager: (no debugging symbols found)
NetworkManager: (no debugging symbols found)
NetworkManager: (no debugging symbols found)
NetworkManager: (no debugging symbols found)
NetworkManager: (no debugging symbols found)
NetworkManager: (no debugging symbols found)
NetworkManager: (no debugging symbols found)
NetworkManager: (no debugging symbols found)
NetworkManager: 0xffffe410 in __kernel_vsyscall ()
NetworkManager: ******************* END **********************************

seems to have the same problem on both of my comp

I was seeing the same thing until I removed the following line from /etc/network/interfaces:

iface eth1 inet dhcp

With this line removed, I no longer have this problem.

I had no auto eth1, if that makes any difference.

I have a similar crash (Bug 37091 -- marked as a dup of this one). I also have a line

iface eth1 inet static

(similar to the one above) in my /etc/network/interfaces.

I cannot remove this, however, because this how I block network-manager from using my internal wireless card, which is unreliable with the current dapper system. Instead, I use a PCMCIA card.

I can also confirm this bug and the above fix. Once I removed all lines in /etc/network/interfaces, except lo related, the above error disappears.

I get this error on a Centrino laptop with (lspci):
0000:02:01.0 Network controller: Intel Corporation PRO/Wireless 2915ABG MiniPCI Adapter (rev 05)

The ipw2200 driver.

The problem disappears when commenting out all configuration options in /etc/network/interfaces for eth0 and eth1.

Dapper should default to NetworkManager

This is a big problem for regular users, no errors are being displayed explaining the crash or that the crash occurs. To a normal user, the nm-applet just doesn't appear.

True. I spend a million hours to figure that out from various sources...
Now atleast i have the nm-applet working. But it doesnt work as promised:...it is supposed to connect to the wired with the static addresses when connected and turn the wireless off. I have to manually turn the wireless off to get the wired connection working.

Similar crash on a Centrino laptop (ipw2200 driver).
I have to commented out the line setting iface eth1 in /etc/network/interfaces to get NetworkManager working.

similarly...i havent been able to connect to the wireless with WEP encryption

Here is a backtrace:

Apr 11 02:32:46 localhost NetworkManager: ******************* START **********************************
Apr 11 02:32:47 localhost NetworkManager: Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1".
Apr 11 02:32:47 localhost NetworkManager: [Thread debugging using libthread_db enabled]
Apr 11 02:32:47 localhost NetworkManager: [New Thread -1211926848 (LWP 13973)]
Apr 11 02:32:47 localhost NetworkManager: 0xffffe410 in __kernel_vsyscall ()
Apr 11 02:32:47 localhost NetworkManager:
Apr 11 02:32:47 localhost NetworkManager: Thread 1 (Thread -1211926848 (LWP 13973)):
Apr 11 02:32:47 localhost NetworkManager: #0 0xffffe410 in __kernel_vsyscall ()
Apr 11 02:32:47 localhost NetworkManager: #1 0xb7f11463 in __waitpid_nocancel () from /lib/tls/i686/cmov/libpthread.so.0
Apr 11 02:32:47 localhost NetworkManager: #2 0x0806ac32 in nm_logging_backtrace () at nm-logging.c:72
Apr 11 02:32:47 localhost NetworkManager: #3 0x0806ae05 in nm_signal_handler (signo=11) at nm-logging.c:166
Apr 11 02:32:47 localhost NetworkManager: #4 <signal handler called>
Apr 11 02:32:47 localhost NetworkManager: #5 0xb7f0ce47 in pthread_mutex_destroy ()
Apr 11 02:32:47 localhost NetworkManager: from /lib/tls/i686/cmov/libpthread.so.0
Apr 11 02:32:47 localhost NetworkManager: #6 0x08059a3b in nm_device_802_11_wireless_dispose (object=0x0)
Apr 11 02:32:47 localhost NetworkManager: at nm-device-802-11-wireless.c:3015
Apr 11 02:32:47 localhost NetworkManager: #7 0xb7c66eac in g_object_unref () from /usr/lib/libgobject-2.0.so.0
Apr 11 02:32:47 localhost NetworkManager: #8 0x08055e50 in nm_device_new (iface=0x80aa5d8 "wlan0",
Apr 11 02:32:47 localhost NetworkManager: rg/freedesktop/Hal/devices/net_00_02_6f_06_a2_ea_0",
Apr 11 02:32:47 localhost NetworkManager: test_dev=0, test_dev_type=DEVICE_TYPE_UNKNOWN, app_data=0x8094110)
Apr 11 02:32:47 localhost NetworkManager: at nm-device.c:205
Apr 11 02:32:47 localhost NetworkManager: #9 0x080658aa in nm_create_device_and_add_to_list (data=0x8094110,
Apr 11 02:32:47 localhost NetworkManager: udi=0x80aaf40 "/org/freedesktop/Hal/devices/net_00_02_6f_06_a2_ea_0",
Apr 11 02:32:47 localhost NetworkManager: iface=0x80aa5d8 "wlan0", test_device=0,
Apr 11 02:32:47 localhost NetworkManager: test_device_type=DEVICE_TYPE_UNKNOWN) at NetworkManager.c:127
Apr 11 02:32:47 localhost NetworkManager: #10 0x08065edf in nm_add_initial_devices (data=0x8094110)
Apr 11 02:32:47 localhost NetworkManager: at NetworkManager.c:328
Apr 11 02:32:47 localhost NetworkManager: #11 0x0806655e in nm_hal_init (data=0x8094110) at NetworkManager.c:689
Apr 11 02:32:47 localhost NetworkManager: #12 0x080672ea in main (argc=1, argv=0xbff7e894) at NetworkManager.c:889
Apr 11 02:32:47 localhost NetworkManager: #0 0xffffe410 in __kernel_vsyscall ()
Apr 11 02:32:47 localhost NetworkManager: ******************* END **********************************

I tried to look at the code but couldn't come up with nothing (knowing about nothing about NM and glib). But I can see that #7 g_object_unref() is calling into #6 nm_device_802_11_wireless_dispose() with a NULL pointer w...

I can confirm a similar bug that occurs when I have a configuration for my wireless card in the interfaces file. Here's the relevant section of the syslog:

Apr 14 21:07:44 localhost NetworkManager: <information>^Istarting...
Apr 14 21:07:44 localhost NetworkManager: <WARNING>^I main (): nm_data_new: Setting up dbus filter
Apr 14 21:07:45 localhost NetworkManager: <WARNING>^I nm_signal_handler (): Caught signal 11. Generating backtrace...
Apr 14 21:07:45 localhost NetworkManager: ******************* START **********************************
Apr 14 21:07:45 localhost NetworkManager: Frame 0: /usr/sbin/NetworkManager [0x806ac58]
Apr 14 21:07:45 localhost NetworkManager: Frame 1: /usr/sbin/NetworkManager [0x806ae05]
Apr 14 21:07:45 localhost NetworkManager: Frame 2: [0xffffe420]
Apr 14 21:07:45 localhost NetworkManager: Frame 3: /usr/sbin/NetworkManager [0x8059a3b]
Apr 14 21:07:45 localhost NetworkManager: Frame 4: /usr/lib/libgobject-2.0.so.0(g_object_unref+0xc5) [0xb7cbaeac]
Apr 14 21:07:45 localhost NetworkManager: Frame 5: /usr/sbin/NetworkManager(nm_device_new+0x415) [0x8055e50]
Apr 14 21:07:45 localhost NetworkManager: Frame 6: /usr/sbin/NetworkManager(nm_create_device_and_add_to_list+0xb8) [0x80658aa]
Apr 14 21:07:45 localhost NetworkManager: Frame 7: /usr/sbin/NetworkManager(nm_add_initial_devices+0xea) [0x8065edf]
Apr 14 21:07:45 localhost NetworkManager: Frame 8: /usr/sbin/NetworkManager(nm_hal_init+0xef) [0x806655e]
Apr 14 21:07:45 localhost NetworkManager: Frame 9: /usr/sbin/NetworkManager(main+0x875) [0x80672ea]
Apr 14 21:07:45 localhost NetworkManager: Frame 10: /lib/tls/i686/cmov/libc.so.6(__libc_start_main+0xd2) [0xb7cfbea2]
Apr 14 21:07:45 localhost NetworkManager: Frame 11: /usr/sbin/NetworkManager [0x8052d91]
Apr 14 21:07:45 localhost NetworkManager: ******************* END **********************************

We've got 0.6.2 now.

No, this is not fixed as of the 0.6.2-0ubuntu4 version.

Similar crash on a Centrino laptop (ipw2200 driver).
I have to commented out the line setting iface eth1 in /etc/network/interfaces to get NetworkManager working.

Without commenting out I got this debug messages:

sudo NetworkManager --no-daemon
NetworkManager: <information> starting...
NetworkManager: <WARNING> main (): nm_data_new: Setting up dbus filter
NetworkManager: <information> eth0: Device is fully-supported using driver 'e100'.
NetworkManager: <information> nm_device_init(): waiting for device's worker thread to start
NetworkManager: <information> nm_device_init(): device's worker thread started, continuing.
NetworkManager: <information> Now managing wired Ethernet (802.3) device 'eth0'.
NetworkManager: <information> eth1: Device is fully-supported using driver 'ipw2200'.
NetworkManager: <information> nm_device_init(): waiting for device's worker thread to start
NetworkManager: <information> nm_device_init(): device's worker thread started, continuing.
NetworkManager: <information> Now managing wireless (802.11) device 'eth1'.
NetworkManager: <information> Updating allowed wireless network lists.
NetworkManager: <WARNING> nm_dbus_get_networks_cb (): nm-dbus-nmi.c:522 (nm_dbus_get_networks_cb): error received: org.freedesktop.NetworkManagerInfo.NoNetworks - org.freedesktop.NetworkManagerInfo.NoNetworks.
NetworkManager: <WARNING> nm_signal_handler (): Caught signal 2, shutting down normally.
NetworkManager: <information> Caught terminiation signal
NetworkManager: <debug info> [1146763008.632036] nm_print_open_socks (): Open Sockets List:
NetworkManager: <debug info> [1146763008.632711] nm_print_open_socks (): Open Sockets List Done.
NetworkManager: <information> Deactivating device eth0.
NetworkManager: <information> Deactivating device eth1.
user@uber:~$ sudo vim /etc/network/interfaces
user@uber:~$ sudo NetworkManager --no-daemon
NetworkManager: <information> starting...
NetworkManager: <WARNING> main (): nm_data_new: Setting up dbus filter
NetworkManager: <information> eth0: Device is fully-supported using driver 'e100'.
NetworkManager: <information> nm_device_init(): waiting for device's worker thread to start
NetworkManager: <information> nm_device_init(): device's worker thread started, continuing.
NetworkManager: <information> Now managing wired Ethernet (802.3) device 'eth0'.
NetworkManager: <WARNING> nm_signal_handler (): Caught signal 11. Generating backtrace...
NetworkManager: ******************* START **********************************
NetworkManager: Frame 0: NetworkManager [0x806ac48]
NetworkManager: Frame 1: NetworkManager [0x806adf5]
NetworkManager: Frame 2: [0xffffe420]
NetworkManager: Frame 3: NetworkManager [0x8059a3b]
NetworkManager: Frame 4: /usr/lib/libgobject-2.0.so.0(g_object_unref+0xc5) [0xb7cbceac]
NetworkManager: Frame 5: NetworkManager(nm_device_new+0x415) [0x8055e50]
NetworkManager: Frame 6: NetworkManager(nm_create_device_and_add_to_list+0xb8) [0x806589a]
NetworkManager: Frame 7: NetworkManager(nm_add_initial_devices+0xea) [0x8065ecf]
NetworkManager: Frame 8: NetworkManager(nm_hal_init+...

It seems that I have found the bug at least the reason for the crash. If /etc/network/interfaces consist configuration lines for wpa_supplicant NetworkManager crashes. After removing them it starts fine with both eth-devices listed in interfaces.

Using my /etc/networking/interfaces (which I use to tell NM5.1 *not* to use my builltin wireless card (eth1), so that it instead uses a pcmcia wireless card (eth0)):
--------------------------
iface lo inet loopback auto lo

iface eth1 inet static
address 192.168.1.114
netmask 255.255.255.0
wireless-essid ??????
wireless-key ?????????
gateway 192.168.1.1
--------------------

I tried NM6.2 from today, and go the following crash (from syslog):

May 4 10:33:29 localhost NetworkManager: <WARNING>^I main (): nm_data_new: Setting up dbus filter
May 4 10:33:29 localhost NetworkManager: <information>^Ieth0: Device is fully-supported using driver 'orinoco_cs'.
May 4 10:33:29 localhost NetworkManager: <information>^Inm_device_init(): waiting for device's worker thread to start
May 4 10:33:29 localhost NetworkManager: <information>^Inm_device_init(): device's worker thread started, continuing.
May 4 10:33:29 localhost NetworkManager: <information>^INow managing wireless (802.11) device 'eth0'.
May 4 10:33:29 localhost kernel: [4297908.926000] ADDRCONF(NETDEV_UP): eth1: link is not ready
May 4 10:33:29 localhost NetworkManager: <WARNING>^I nm_signal_handler (): Caught signal 11. Generating backtrace...
May 4 10:33:29 localhost NetworkManager: ******************* START **********************************
May 4 10:33:30 localhost NetworkManager: (no debugging symbols found)
May 4 10:33:30 localhost NetworkManager: Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1".
May 4 10:33:30 localhost NetworkManager: (no debugging symbols found)
May 4 10:33:30 localhost last message repeated 5 times
May 4 10:33:30 localhost NetworkManager: [Thread debugging using libthread_db enabled]
May 4 10:33:30 localhost NetworkManager: [New Thread -1211685184 (LWP 9990)]
May 4 10:33:30 localhost NetworkManager: [New Thread -1211688016 (LWP 10007)]
May 4 10:33:30 localhost NetworkManager: (no debugging symbols found)
May 4 10:33:31 localhost last message repeated 15 times
May 4 10:33:31 localhost NetworkManager: 0xffffe410 in __kernel_vsyscall ()
May 4 10:33:31 localhost NetworkManager: ******************* END **********************************

Got a gdb session going on the -dbg version during a crash and found some interesting things...

(gdb) ptype NMDevice80211WirelessPrivate {
  [...]
  double freqs[32];
}
(gdb) print *self->priv
... num_freqs = -1 '�', freqs = {1.3258229701962879e-310,
    -2.7516420520760352e-135, 1.7604077081861733e-310,
    -2.7516420517557019e-135, 2.1949924461760587e-310,
    -2.7516420514353686e-135, 2.6295771841659441e-310,
    -2.7516420511150353e-135, 3.0641619221558295e-310,
    -2.751642050794702e-135, 3.4987466601457149e-310,
    -2.7516420504743687e-135, 3.9333313981356003e-310,
    -2.7516420501540354e-135, 4.3679161361254857e-310,
    -2.7516420498337021e-135, 4.8025008741153711e-310,
    -2.7516420495133688e-135, 5.2370856121052565e-310,
    -2.7516420491930355e-135, 5.6716703500951419e-310,
    -2.7516420488727022e-135, 6.1062550880850273e-310,
    -2.7516420485523689e-135, 6.5408398260749127e-310,
    -2.7516420482320356e-135, 6.9754245640647981e-310,
    -2.7516420479117023e-135, 7.4100093020546834e-310,
    -2.751642047591369e-135, 7.8445940400445688e-310,
    -2.7516420472710357e-135}, scan_mutex = 0x58, ap_list = 0x9868,
...
[ scan_mutex is 0x58 ?!?!?! ]
(gdb) x/48fg self->priv->freqs
0x37ea0f56 <pthread_attr_setdetachstate@plt+2>: 1.3258229701962879e-310 -2.7516420520760352e-135
0x37ea0f66 <g_thread_self@plt+2>: 1.7604077081861733e-310 -2.7516420517557019e-135
...
0x37ea1046 <g_thread_error_quark@plt+2>: 7.8445940400445688e-310 -2.7516420472710357e-135
0x37ea1056 <pthread_attr_getschedparam@plt+2>: 8.2791787780344542e-310 -2.7516420469507024e-135
0x37ea1066 <g_thread_init@plt+2>: 8.7137635160243396e-310 -2.7516420466303691e-135
0x37ea1076 <g_strerror@plt+2>: 9.148348254014225e-310 -2.7516420463100358e-135
0x37ea1086 <g_malloc@plt+2>: 9.5829329920041104e-310 -2.7516420459897026e-135
0x37ea1096 <pthread_setschedparam@plt+2>: 1.0017517729993996e-309 -2.7516420456693693e-135
0x37ea10a6 <pthread_setspecific@plt+2>: 1.0452102467983881e-309 -2.751642045349036e-135
0x37ea10b6 <pthread_key_create@plt+2>: 1.0886687205973767e-309 -2.7516420450287027e-135

Looks like it was writing frequency values well past the end of the array, there...

Actually, digging deeper, it looks like this is coming from the handling of scan_mutex, and the early-out mechanism for devices that appear to be off limits because nm_system_device_get_disabled

The mutex is allocated in real_init. But in nm-device.c, the check for whether or not the device is disabled happens before the call to the type-specific init function. That means the cleanup code is getting called for an object that is not fully initialized. The cleanup code tries to free the scan_mutex, which was never allocated, and BOOM!

This patch changes nm-device-802-11-wireless.c so that the class keeps track not of whether dispose has been called, but whether the object is in an initialized state. If the object's class-specific initializer has not been called, then we shouldn't do any class-specific cleanup, either.

The assumption that dispose always has a fully initialized object to work with exists throughout this code. It is definitely backwards logic that the original author should correct to avoid this kind of problem in the future.

If everyone could test that patch and let us know whether it
a) fixes the bug
b) causes any new problems

Then I should be able to get an upload out with it asap.

Thanks!

The patch did indeed fix the crash!! txshtkckr rules!

Unfortunately, while network-manager is not crashing, DHCP is failied for me, so I'm back to 0.5.1. But I suppose that must be a separate issue (??).

Not sure. It had been working perfectly for me, but now my DHCPs are failing as well. Haven't dug into it far enough to figure that one out... :(

I reported this on the NM list and got an answer[1] from Dan Williams.

[1] http://mail.gnome.org/archives/networkmanager-list/2006-May/msg00032.html

Sorry, my first patch submission... didn't think about license concerns. GPL is fine by me for it. I'll submit it to the NM list as soon as I figure out where to subscribe to it at. ;)

crf

http://lists.gnome.org/mailman/listinfo/networkmanager-list

I got something similar:

(gdb) bt
#0 0xb7f58e47 in pthread_mutex_destroy () from /lib/tls/i686/cmov/libpthread.so.0
#1 0x08059a3b in nm_device_802_11_wireless_get_type ()
#2 0xb7cb2eac in g_object_unref () from /usr/lib/libgobject-2.0.so.0
#3 0x08055e50 in nm_device_new ()
#4 0x0806589a in nm_create_device_and_add_to_list ()
#5 0x08065ecf in nm_add_initial_devices ()
#6 0x0806654e in nm_hal_init ()
#7 0x080672da in main ()

Any idea if the proposed patch will fix this problem as well? This happened when HAL(?) started using the wired-network card as eth1 and the wireless-network card as eth0, and I had a static configuration for the wired card in /etc/network/interfaces.

Yes, that looks like it is probably the same problem, both by the trace and
your description of the circumstances. Although, it would also be nice to
find out why your eth0 and eth1 switched places, as well. That should really
be deterministic for a sane system...

crf

-----------------------------------
From: Celso Pinto <email address hidden>
Date: Monday 08 May 2006 19:24
Subj: [Bug 37084] Re: network-manager crashes because of old ways
configuration
To: <email address hidden>
Cc:

I got something similar:

(gdb) bt
#0 0xb7f58e47 in pthread_mutex_destroy ()
from /lib/tls/i686/cmov/libpthread.so.0
#1 0x08059a3b in nm_device_802_11_wireless_get_type ()
#2 0xb7cb2eac in g_object_unref () from /usr/lib/libgobject-2.0.so.0
#3 0x08055e50 in nm_device_new ()
#4 0x0806589a in nm_create_device_and_add_to_list ()
#5 0x08065ecf in nm_add_initial_devices ()
#6 0x0806654e in nm_hal_init ()
#7 0x080672da in main ()

Any idea if the proposed patch will fix this problem as well? This happened
when HAL(?) started using the wired-network card as eth1 and the
wireless-network card as eth0, and I had a static configuration for the wired
card in /etc/network/interfaces.

--
network-manager crashes because of old ways configuration
https://launchpad.net/bugs/37084

After testing, I'm convinced that this is the right patch for the job -- and that it's unrelated to any DHCP problems

What packages need to be installed to get the NM syslog trace to display proper debugging symbols?