linux-next: lockdep whinge in cgroup_rmdir

Nick Piggin npiggin at gmail.com
Thu Jan 13 19:35:17 PST 2011


On Fri, Jan 14, 2011 at 2:34 AM,  <Valdis.Kletnieks at vt.edu> wrote:
> Seen booting yesterday's linux-next, was not present in 2.6.37-rc7-mmotm1202.
>
> Not sure if it's an selinux or cgroup issue, so I'm throwing it at every
> address I can find for either.  This is easily replicatable and happens at
> every boot, so I can test patches if needed.  Am willing to bisect it down if
> nobody knows right off the bat what the problem is.
>
> The 'W' taint is from the already-reported kernel/workqueue.c worker_enter_idle issue.
>
> [   85.100795] systemd[1]: readahead-replay.service: main process exited, code=exited, status=1
> [   85.101530]
> [   85.101531] =============================================
> [   85.101796] [ INFO: possible recursive locking detected ]
> [   85.102002] 2.6.37-next-20110111 #1
> [   85.102009] ---------------------------------------------
> [   85.102009] systemd/1 is trying to acquire lock:
> [   85.102009]  (&(&dentry->d_lock)->rlock){+.+...}, at: [<ffffffff8107ca5c>] cgroup_rmdir+0x339/0x479
> [   85.102009]
> [   85.102009] but task is already holding lock:
> [   85.102009]  (&(&dentry->d_lock)->rlock){+.+...}, at: [<ffffffff8107ca54>] cgroup_rmdir+0x331/0x479
> [   85.102009]
> [   85.102009] other info that might help us debug this:
> [   85.102009] 4 locks held by systemd/1:
> [   85.102009]  #0:  (&sb->s_type->i_mutex_key#14/1){+.+.+.}, at: [<ffffffff810fea4d>] do_rmdir+0x7d/0x121
> [   85.102009]  #1:  (&sb->s_type->i_mutex_key#14){+.+.+.}, at: [<ffffffff810fd4bc>] vfs_rmdir+0x4a/0xbe
> [   85.102009]  #2:  (cgroup_mutex){+.+.+.}, at: [<ffffffff8107cb84>] cgroup_rmdir+0x461/0x479
> [   85.102009]  #3:  (&(&dentry->d_lock)->rlock){+.+...}, at: [<ffffffff8107ca54>] cgroup_rmdir+0x331/0x479
> [   85.102009]
> [   85.102009] stack backtrace:
> [   85.102009] Pid: 1, comm: systemd Tainted: G        W   2.6.37-next-20110111 #1
> [   85.102009] Call Trace:
> [   85.102009]  [<ffffffff81069f22>] ? __lock_acquire+0x929/0xd4e
> [   85.102009]  [<ffffffff8107c6f1>] ? cgroup_clear_directory+0xff/0x131
> [   85.102009]  [<ffffffff8107c6f1>] ? cgroup_clear_directory+0xff/0x131
> [   85.102009]  [<ffffffff8107ca5c>] ? cgroup_rmdir+0x339/0x479
> [   85.102009]  [<ffffffff8106a859>] ? lock_acquire+0x100/0x126
> [   85.102009]  [<ffffffff8107ca5c>] ? cgroup_rmdir+0x339/0x479
> [   85.102009]  [<ffffffff815521ef>] ? sub_preempt_count+0x35/0x48
> [   85.102009]  [<ffffffff8154e401>] ? _raw_spin_lock+0x36/0x45
> [   85.102009]  [<ffffffff8107ca5c>] ? cgroup_rmdir+0x339/0x479
> [   85.102009]  [<ffffffff8107ca5c>] ? cgroup_rmdir+0x339/0x479
> [   85.102009]  [<ffffffff810579cd>] ? autoremove_wake_function+0x0/0x34
> [   85.102009]  [<ffffffff811e1839>] ? selinux_inode_rmdir+0x15/0x17
> [   85.102009]  [<ffffffff810fd4eb>] ? vfs_rmdir+0x79/0xbe
> [   85.102009]  [<ffffffff810feaa0>] ? do_rmdir+0xd0/0x121
> [   85.102009]  [<ffffffff8100256c>] ? sysret_check+0x27/0x62
> [   85.102009]  [<ffffffff8106ac79>] ? trace_hardirqs_on_caller+0x117/0x13b
> [   85.102009]  [<ffffffff8154e201>] ? trace_hardirqs_on_thunk+0x3a/0x3f
> [   85.102009]  [<ffffffff8110040b>] ? sys_rmdir+0x11/0x13
> [   85.102009]  [<ffffffff8100253b>] ? system_call_fastpath+0x16/0x1b
> [   85.268272] systemd[1]: readahead-collect.service: main process exited, code=exited, status=1
>
> Any ideas?

It looks like it is just a missing parent->child lock order annotation, but
mainline cgroupfs code looks to be OK there. What does
cgroup_clear_directory() look like in mmotm?


More information about the Containers mailing list