The following reply was made to PR kern/29670; it has been noted by GNATS.

From: Ken Raeburn <raeburn@raeburn.org>
To: gnats-bugs@netbsd.org
Cc: 
Subject: Re: kern/29670: "release of unlocked lock" panic with null fs
Date: Mon, 14 Mar 2005 08:47:24 -0500

 Had another hang this morning.
 
 "vnode: table is full - increase kern.maxvnodes or NVNODE" logged at 
 04:21:01
 
 followed quickly by this same panic
 lockmgr+0xcb
 layer_unlock+0x4a
 VOP_UNLOCK+0x28
 vput+0x35
 lookup+0x283
 namei+0x138
 vn_open+0x6d
 sys_open+0xc7
 syscall_plain+0x7e
 
 hung "syncing disks"
 
 "ps" in kdb indicates the running process is ksh, called from sh, from 
 sh, from cron, uid 1000; a "rm" is also running.  There's a cron job 
 that fires each day at this time that fetches some code by ftp and then 
 builds it, but the job starts by deleting the previous night's 
 source+build tree; perhaps that causes high demand on the vnode table, 
 if entries aren't freed up fast enough.  This build tree is also 
 mentioned in one of the aforementioned null-mounted file systems.  
 (I've switched it to go to the underlying path, to see if that stops 
 these panic problems.)
 
 "c" doesn't seem to do anything; then ctl-alt-esc then "trace" shows 
 keyboard interrupt called from
 
 ltsleep+0x244
 getblk+0x91
 bread+0x37
 ffs_update+0x2bf
 VOP_UPDATE+0x34
 ffs_full_fsync+0x192
 ffs_fsync+0x4c
 VOP_FSYNC+0x4c
 ffs_sync+0x1ea
 sys_sync+0xca
 vfs_shutdown+0x5f
 cpu_reboot+0x103
 panic+0x108
 lockmgr...