(ZFS?): panic: lockmgr: locking against myself

From: Peter Schuller <peter.schuller_at_infidyne.com>
Date: Sat, 28 Jul 2007 20:28:36 +0200
Hello,

I have a machine with root on ZFS and /boot on gmirror. Version is 7-CURRENT 
from about a week ago or so (can't check since system won't boot). After a 
certain sequence of events that I will describe below, I now get this on boot 
(typed manually, so there may be some mistakes):

Trying to mount root from zfs:tank/root
panic: lockmgr: locking against myself
cpuid = 0
KBD: enter: panic
[thread pid 1 tid 100002 ]
Stopped at kbd_enter+0x31: leave
db>bt
kbd_enter() at kbd_enter+0x31
panic() at panic+0x173
_lockmgr() at _lockmgr+0x085a
VOP_LOCK1_APV() at VOP_LOCK1_APV+0x46
_vn_lock() at _vn_lock+0x83
vrele() at vrele+0xf5
mountcheckdirs() at mountcheckdirs+0x1e8
vfs_donmount() at vfs_donmount+0x111c
kernel_mount() at kernel_mount+0x88
kernel_vmount() at kernel_vmoun+0xcb
vfs_mountroot_try() at vfs_mountroot_try+0x10c
vfs_mountroot() at vfs_mountroot+0x324
start_init() at start_init+0x4d
fork_exit() at fork_exit+0x11f
fork_trampoline() at fork_trampoline+0xe
--- trap 0, rip = 0, rsp = 0xfffffffffac357d30, rbp = 0 ---

This is on a Dell 2950, with two SATA drives exposed as individual 
non-redundant volumes through the PERC 5/i controller (mfi driver).

The sequence of events were:

(1) Boot the system.

(2) Yank one of the drives live; watch errors flash by, zfs detecting the 
corruption of one of the drives.

(3) Reboot with the drive missing, confirming booting still works (I am 
*pretty* sure I did this).

(4) Shutdown, insert drive again, enable it in RAID controller config, and 
boot.

(5) Gmirror refuses to use the swapped drive because it is "broken" (not sure 
why this happened; I was assuming it would detect it as out of date and 
rebuild). I manually forget and re-insert the drive.

(6) Meanwhile, ZFS has resilvered and is reporting some checksum mismatches. I 
scrub the pool and heal some more checksum mismatches (corruption/bitflips on 
hotswap is consistent with some other experience with a Marvell controller 
and attempted hotswapping). Second scrub completes without errors.

(7) All is fine. I shut the machine down physically, remove one drive, and try 
to boot for the purpose of testing this particular failure mode.

(8) I now get the above panic on trying to mount root.

(9) Shutdown, re-insert drive again per above, and try to boot. Still the same 
panic. Both drives are now being detected in the gmirror though.

-- 
/ Peter Schuller

PGP userID: 0xE9758B7D or 'Peter Schuller <peter.schuller_at_infidyne.com>'
Key retrieval: Send an E-Mail to getpgpkey_at_scode.org
E-Mail: peter.schuller_at_infidyne.com Web: http://www.scode.org
Received on Sat Jul 28 2007 - 14:49:24 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:15 UTC