UFS lock order reversal stack trace with r264677 on i386

[Benjamin Kaduk]

On Sat, 19 Apr 2014, R. Tyler Croy wrote:

> I've noticed this as of late on my i386 -CURRENT Thinkpad T43 when I perform 
> some file operations, but an exact reproduction case I've not yet stumbled 
> upon:
>
> Apr 20 01:29:32 lemon kernel: lock order reversal:
> Apr 20 01:29:32 lemon kernel: 1st 0xc5832358 bufwait (bufwait) @ 
> /usr/home/tyler/source/github/freebsd/sys/kern/vfs_bio.c:3081
> Apr 20 01:29:32 lemon kernel: 2nd 0xc6f1d600 dirhash (dirhash) @ 
> /usr/home/tyler/source/github/freebsd/sys/ufs/ufs/ufs_dirhash.c:284
[...]
> Apr 20 01:29:54 lemon kernel: lock order reversal:
> Apr 20 01:29:54 lemon kernel: 1st 0xc699e388 ufs (ufs) @ 
> /usr/home/tyler/source/github/freebsd/sys/kern/vfs_subr.c:2101
> Apr 20 01:29:54 lemon kernel: 2nd 0xc5859e98 bufwait (bufwait) @ 
> /usr/home/tyler/source/github/freebsd/sys/ufs/ffs/ffs_vnops.c:262
> Apr 20 01:29:54 lemon kernel: 3rd 0xc7d27c68 ufs (ufs) @ 
> /usr/home/tyler/source/github/freebsd/sys/kern/vfs_subr.c:2101

These are "well known", #261 and #285 at 
http://sources.zabbadoz.net/freebsd/lor.html .

-Ben Kaduk