[lxc-devel] segfault on shutdown if containers running

[Adam Majer]

Cedric Le Goater wrote:
> On 06/29/2009 02:47 AM, Adam Majer wrote:
>> Daniel Lezcano wrote:
>>>> Is this a known problem? If not, I'll provide the backtrace.
>>> Yes, please, maybe I am missing something but I was not able to
>>> reproduce it.
>>
>> I see to be failing at being able to safe this segfault. The computer
>> has no serial port so I've been trying to get netconsole to work, and it
>> does, but the entire OOPS is only visible on local console, not
>> transmitted. Only part of the oops is transmitted :(
>>
>>
>> [  635.526580] BUG: unable to handle kernel NULL pointer dereference at
>> 000000000000009c
>> [  635.526692] IP: [<ffffffff803b26f1>] tty_open+0x1e6/0x411
>> [  635.526775] PGD 13dd44067 PUD 1379d5067 PMD 0
>> [  635.526905] Oops: 0000 [#1] SMP
>> [  635.526999] last sysfs file: /sys/devices/virtual/net/lo/operstate
>> [  635.527045] CPU 1
>> [  635.527110] Modules linked in: netconsole configfs veth nvidia(P)
>> binfmt_misc powernow_k8 cpufreq_powersave cpufreq_conservative
>> cpufreq_stats cpufreq_userspace tun bridge stp ext2 mbcache it87
>> hwmon_vid loop firewire_sbp2 snd_hda_codec_analog snd_hda_intel
>> snd_hda_codec snd_hwdep snd_pcm_oss snd_mixer_oss snd_pcm snd_seq_midi
>> snd_rawmidi snd_seq_midi_event snd_seq snd_timer snd_seq_device usblp
>> k8temp i2c_nforce2 pcspkr snd evdev asus_atk0110 i2c_core soundcore
>> snd_page_alloc button processor xfs exportfs raid1 md_mod
>>
>>
>>
>> So, I've copied the stacktrace on paper and typing it back,
>>
>> Process getty (pid:3940)<<<-- looks like a process in the container?
>>
>> chrdev_open + 0x148/0x167
>> chrdev_open + 0x0/0x167
>> __dentry_open + 0x148/0x260
>> do_flip_open + 0x468/0x85a
>> alloc_fd + ....
>> do_sys_gen + ...
>> system_call_fastpath + ....
>>
>> RIP tty_open
> 
> which kernel are you using ? could you run :
> 
>     $ addr2line -e <vmlinux> ffffffff803b26f1
> 

This is a stock Debian kernel 2.6.30-1-amd64

  linux-image-2.6.30-1-amd64

Unfortunately, there is no vmlinux, only the compressed boot image.

I have also recompiled the kernel myself and the tty_open  had the same
base address, but alas, the addr2line just returned,

  tty_io.c:0

so not much more helpful than the System.map.

- Adam

PS. The Debian kernel is compiled with,

# CONFIG_DEBUG_INFO is not set

Would setting this to yes produce a backtrace with line numbers?