[Drbd-dev] DRBD8: Ooops on reboot after WFBitMaps state deadlock

[Montrose, Ernest]

Hi all,
The two nodes were deadlocked in this WFBitMaps state.  I proceeded to
try to recover with drbdadm() that would hang.
I proceeded to reboot the nodes and I got an Ooops in one of the nodes:
Oops: 0000 [#1]
CPU:    0
EIP is at bd_release+0x9/0x30
eax: f12cbe44  ebx: 00000001  ecx: ec78ac80  edx: 00000000
esi: eebd7400  edi: 00000000  ebp: efc6be50  esp: efc6be50
ds: 007b  es: 007b  ss: 0069
Process cqueue/0 (pid: 776, threadinfo=efc6a000 task=efc06570)
Stack: <0>efc6bec4 f12c2428 ec78ac80 00000000 00000001 00000002 00000000
00000001
        00000000 00000001 00000004 00000000 00000000 00000001 000000d0
ebd59d80
        00000000 efc6beac c02e03c8 00000000 00000000 240bc3e4 00000000
efc6a000
Call Trace:
[<c010522a>] show_stack_log_lvl+0xaa/0xe0
[<c010543e>] show_registers+0x18e/0x210
[<c0105639>] die+0xd9/0x180
[<c011326c>] do_page_fault+0x3cc/0x6ce
[<c0104e3b>] error_code+0x2b/0x30
[<f12c2428>] drbd_set_role+0x178/0x4d0 [drbd]
[<f12c27ec>] drbd_nl_primary+0x6c/0x80 [drbd]
[<f12c4ff5>] drbd_connector_callback+0xb5/0x1d0 [drbd]
[<c026f3f3>] cn_queue_wrapper+0x13/0x40
[<c012a408>] run_workqueue+0x88/0x120
[<c012a5e8>] worker_thread+0x148/0x170
[<c012dde7>] kthread+0xa7/0xf0
[<c0102d7d>] kernel_thread_helper+0x5/0x18
Code: 00 00 8b 42 3c ff 40 38 c7 40 34 a0 1f 16 c0 89 5a 34 ff 42 38 5b
89 c8 5d c3 90 8d b4 26 00 00 00 00 55 89 e5 8b 4d 08 8b 51 3c <8b> 42
38 48 89 42 38 85 c0 75
 
This seems to imply that mdev->this_bdev is pointing to an invalid
location. 
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.linbit.com/pipermail/drbd-dev/attachments/20061128/23ca4e53/attachment.html