Lock order assertion failure

[Lukasz Zalewski]

On 22/09/2012 11:40, Lukasz Zalewski wrote:
> On 22/09/2012 00:17, Andrew Bartlett wrote:
>> On Fri, 2012-09-21 at 16:11 +0100, Lukasz Zalewski wrote:
>>> On 21/09/12 16:07, Andrew Bartlett wrote:
>>>> On Fri, 2012-09-21 at 14:22 +0100, Lukasz Zalewski wrote:
>>>>> On 19/09/12 19:05, David Rivera wrote:
>>>>>> Hi Ricky,
>>>>>>
>>>>>> Yep you're right, it is a different panic than the one you had. I
>>>>>> posted
>>>>>> the info in my previous email but I should have included it again.
>>>>>> Here is the backtrace:  http://pastebin.com/Z5R0sXHZ . Samba Version
>>>>>> 4.1.0pre1-GIT-ea96d79.
>>>>>
>>>>> Hi all,
>>>>> I'm also experiencing smbd crashes (more less daily) using
>>>>> 4.1.0pre1-GIT-ea96d79.
>>>>>
>>>>> Here is an sniplet from the log file:
>>>>> [2012/09/21 12:23:24.576002,  0]
>>>>> ../lib/dbwrap/dbwrap.c:193(dbwrap_check_lock_order)
>>>>>      Lock order violation: Trying
>>>>> /usr/local/samba/var/lock/smbXsrv_session_global.tdb at 1 while
>>>>> /usr/local/samba/var/lock/locking.tdb at 1 is locked
>>>>> [2012/09/21 12:23:24.576148,  0]
>>>>> ../lib/dbwrap/dbwrap.c:133(debug_lock_order)
>>>>>      lock order:  1:/usr/local/samba/var/lock/locking.tdb 2:<none>
>>>>> 3:<none>
>>>>> [2012/09/21 12:23:24.576236,  0]
>>>>> ../source3/lib/util.c:810(smb_panic_s3)
>>>>>      PANIC (pid 25048): invalid lock_order
>>>>
>>>> Thanks Lukasz,
>>>>
>>>> Can you please file a bug with this.  This is different to the previous
>>>> crashes, as this is a deliberate assertion attempting to avoid
>>>> deadlocks.  As such, at least we understand why we assert, if not
>>>> why we
>>>> got there.
>>>
>>> Hi Andrew,
>>> Would you like me to do it now with the above info, or with complete
>>> backtrace (i need to wait for it to happen again)?
>>
>> It looks like metze just fixed this in master.  Can you check out master
>> and confirm?
>
> Hi Andrew,
> Already up and running will let you know in a few days

Hi Andrew,
Unfortunately crashes still happen (possibly in a different place now). 
I have uploaded a backtrace from the crashed smbd process to the bug report

L