network-manager segfaulting after bringing up Wi-Fi link

[Roland Dreier]

On my system after updating today, I see network manager bring up my wifi link and then immediately start reconnecting.

In my dmesg, I see many sequences of messages like

[ 2027.559748] NetworkManager[7265]: segfault at 0 ip 00007f03d9939f7d sp 00007fff1f62cbb0 error 4 in libc-2.15.so[7f03d9833000+1b3000]
[ 2027.561881] init: network-manager main process (7265) killed by SEGV signal
[ 2027.561942] init: network-manager main process ended, respawning

my /var/log/apport.log shows

ERROR: apport (pid 7120) Mon Jul 16 18:50:02 2012: called for pid 6970, signal 11, core limit 0
ERROR: apport (pid 7120) Mon Jul 16 18:50:02 2012: executable: /usr/sbin/NetworkManager (command line "Network Manager")
ERROR: apport (pid 7120) Mon Jul 16 18:50:02 2012: is_closing_session(): no DBUS_SESSION_BUS_ADDRESS in environment
ERROR: apport (pid 7120) Mon Jul 16 18:50:02 2012: this executable already crashed 2 times, ignoring

(I lost some reports because of course apport fails to upload while the network is down)

Not sure what's going on, but I do note that there is a new network-manager version as of today.

ProblemType: Bug
DistroRelease: Ubuntu 12.10
Package: network-manager 0.9.6.0~git201207161259.00297f4-0ubuntu1
ProcVersionSignature: Ubuntu 3.5.0-4.4-generic 3.5.0-rc6
Uname: Linux 3.5.0-4-generic x86_64
ApportVersion: 2.3-0ubuntu4
Architecture: amd64
Date: Mon Jul 16 18:50:41 2012
EcryptfsInUse: Yes
IfupdownConfig:
 auto lo
 iface lo inet loopback
InstallationMedia: Ubuntu 12.04 LTS "Precise Pangolin" - Alpha amd64 (20120104)
IpRoute:
 default via 10.1.0.1 dev wlan0 proto static
 10.1.0.0/16 dev wlan0 proto kernel scope link src 10.1.0.198 metric 9
 169.254.0.0/16 dev wlan0 scope link metric 1000
NetworkManager.state:
 [main]
 NetworkingEnabled=true
 WirelessEnabled=true
 WWANEnabled=true
 WimaxEnabled=true
SourcePackage: network-manager
UpgradeStatus: Upgraded to quantal on 2012-06-30 (16 days ago)
nmcli-con:
 Error: command ['nmcli', '-f', 'all', 'con'] failed with exit code 9:
 ** (process:8260): WARNING **: Could not initialize NMClient /org/freedesktop/NetworkManager: The name org.freedesktop.NetworkManager was not provided by any .service files
 Error: nmcli (0.9.5.95) and NetworkManager (unknown) versions don't match. Force execution using --nocheck, but the results are unpredictable.
nmcli-dev: Error: command ['nmcli', '-f', 'all', 'dev'] failed with exit code 8: Error: NetworkManager is not running.
nmcli-nm:
 RUNNING VERSION STATE NET-ENABLED WIFI-HARDWARE WIFI WWAN-HARDWARE WWAN
 not running unknown unknown unknown unknown unknown unknown unknown

[Thomas Hood (jdthood)]

I am running the same version of n-m with both wired and Wi-Fi connections and don't get a segfault.

The segfault occurs after an IPv6-related message. Does the segfault go away if you disable IPv6?

[Mathieu Trudel-Lapierre (cyphermox)]

Looks like a fun crash ;)

Roland, would you be able to run NM in a debugger so that we can figure out where it crashes?

Please see http://wiki.ubuntu.com/DebuggingProgramCrashes for how you can do this; but you can also do so quickly by installing:

network-manager-dbg

And starting NM as so (after running 'sudo stop network-manager' to stop it):

sudo gdb --args /usr/sbin/NetworkManager --no-daemon --log-level=debug

That will start NM with debug logs and in a debugger; once it crashes, try to use the command 'bt full' to get a stacktrace of the failure. Hopefully with this we'll have sufficient data despite missing other debugging symbols.

[kossmac (kosmala)]

Hi all,

I'm also affected and wanted to contribute my NetworkManager output.
As I cannot run the above gdb command, because I don't know how to connect to my wifi without nm-applet(which is not working when NetworkManager is started by gdb) I will provide the output of sudo /usr/sbin/NetworkManager --no-daemon --log-level=debug.

[kossmac (kosmala)]

@jdthood: no improvement by disabling ipv6 in sysctl.conf:
net.ipv6.conf.all.disable_ipv6 = 1
net.ipv6.conf.default.disable_ipv6 = 1
net.ipv6.conf.lo.disable_ipv6 = 1

[Roland Dreier (roland.dreier)]

OK, here's the output of

sudo gdb --args /usr/sbin/NetworkManager --no-daemon --log-level=debug |& tee nm.gdb.txt

looks like we crash in the dns management code because priv->last_iface is NULL. I haven't tried to understand what NM is doing here yet.

[Mathieu Trudel-Lapierre (cyphermox)]

Thanks -- Setting to In Progress/Critical; I'm looking at it, trying to figure out what is going wrong.

[Launchpad Janitor (janitor)]

This bug was fixed in the package network-manager - 0.9.6.0~git201207161259.00297f4-0ubuntu2

---------------
network-manager (0.9.6.0~git201207161259.00297f4-0ubuntu2) quantal; urgency=low

  * debian/tests/nm: update test to only start a private DBus bus if we're not
    running in adt-virt-null; which would have its own instance of NM started
    properly on install.
  * debian/patches/lp1025498_link-local_dns_with_null_iface.patch: don't crash
    if iface is NULL when adding link-local nameservers for dnsmasq.
    (LP: #1025498)
 -- Mathieu Trudel-Lapierre <email address hidden> Fri, 20 Jul 2012 15:03:12 -0400