Possible system crash running Ubuntu LTS kernel 3.13.0-77 and Samba

[Thomas Bork]

Am 10.02.2016 um 12:18 schrieb Karolin Seeger:

> this is a heads-up that we have seen some system crashes after updating
> to Ubuntu LTS kernel 3.13.0-77 on systems running Samba.
>
> It looks like a kernel bug triggered by Samba calls.
> A bug report has been created [1].
>
> Downgrading to kernel 3.13.0-76 solves the problem.
>
> [1] https://bugs.launchpad.net/ubuntu/+source/linux-lts-trusty/+bug/1543980

I want to let you know that some of our samba users have a similar 
problem after switching the kernel from 3.2.74 to 3.2.76:

Feb 12 13:27:36 myeis kernel: INFO: rcu_sched detected stall on CPU 1 
(t=15000 jiffies)
Feb 12 13:27:36 myeis kernel: Pid: 3652, comm: smbd Tainted: G  O 
3.2.75-eisfair-1-VIRT #1
Feb 12 13:27:36 myeis kernel: Call Trace:
Feb 12 13:27:36 myeis kernel:  [__rcu_pending+0x64/0x28f] 
__rcu_pending+0x64/0x28f
Feb 12 13:27:36 myeis kernel:  [account_process_tick+0x104/0x15a] ? 
account_process_tick+0x104/0x15a
Feb 12 13:27:36 myeis kernel:  [rcu_check_callbacks+0x6d/0x98] 
rcu_check_callbacks+0x6d/0x98
Feb 12 13:27:36 myeis kernel:  [update_process_times+0x2d/0x58] 
update_process_times+0x2d/0x58
Feb 12 13:27:36 myeis kernel:  [tick_sched_timer+0x0/0x16b] ? 
tick_init_highres+0x11/0x11
Feb 12 13:27:36 myeis kernel:  [tick_sched_timer+0x144/0x16b] 
tick_sched_timer+0x144/0x16b
Feb 12 13:27:36 myeis kernel:  [tick_sched_timer+0x0/0x16b] ? 
tick_init_highres+0x11/0x11
Feb 12 13:27:36 myeis kernel:  [__run_hrtimer.isra.27+0x4d/0x9c] 
__run_hrtimer.isra.27+0x4d/0x9c
Feb 12 13:27:36 myeis kernel:  [hrtimer_interrupt+0xe2/0x1dd] 
hrtimer_interrupt+0xe2/0x1dd
Feb 12 13:27:36 myeis kernel:  [smp_apic_timer_interrupt+0x67/0x7a] 
smp_apic_timer_interrupt+0x67/0x7a
Feb 12 13:27:36 myeis kernel:  [apic_timer_interrupt+0x2a/0x30] 
apic_timer_interrupt+0x2a/0x30
Feb 12 13:27:36 myeis kernel:  [link_path_walk+0xfb/0x61b] ? 
link_path_walk+0xfb/0x61b
Feb 12 13:27:36 myeis kernel:  [try_to_merge_with_ksm_page+0x2d0/0x451] 
? try_to_merge_with_ksm_page+0x2d0/0x451
Feb 12 13:27:36 myeis kernel:  [__ticket_spin_lock+0x16/0x1c] ? 
__ticket_spin_lock+0x16/0x1c
Feb 12 13:27:36 myeis kernel:  [_raw_spin_lock+0x8/0x0b] 
_raw_spin_lock+0x8/0xb
Feb 12 13:27:36 myeis kernel:  [unix_state_double_lock+0x3d/0x41] 
unix_state_double_lock+0x3d/0x41
Feb 12 13:27:36 myeis kernel:  [unix_dgram_connect+0x83/0x153] 
unix_dgram_connect+0x83/0x153
Feb 12 13:27:36 myeis kernel:  [sys_connect+0x63/0x88] sys_connect+0x63/0x88
Feb 12 13:27:36 myeis kernel:  [sys_socketcall+0x76/0x192] 
sys_socketcall+0x76/0x192
Feb 12 13:27:36 myeis kernel:  [syscall_after_call+0x0/0x04] 
syscall_call+0x7/0x7
Feb 12 13:27:36 myeis kernel:  [get_cpu_leaves+0x1dd/0x28a] ? 
get_cpu_leaves+0x1dd/0x28a


This stalls later on seams to leading to a memory leak till the 
oom-killer kills processes and the machines crashes.

Downgrading to kernel 3.2.74 solves the problem.

After reading

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1543980
https://forge.univention.org/bugzilla/show_bug.cgi?id=40558
https://patchwork.ozlabs.org/patch/582017/

I think the patch

> https://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/commit/net/unix/af_unix.c?id=a3b0f6e8a21ef02f69a15abac440572d8cde8c2a

in 3.2.75 is the problem.

-- 
der tom
[eisfair-team]