Re: fsck issue?

On Wed, 5 Jan 2005, Randy Bush wrote:

> a box running happily over a month
>
> FreeBSD foo.psg.com 6.0-CURRENT FreeBSD 6.0-CURRENT #13: Tue Nov 30 20:38:02 GMT 2004     root_at_foo.psg.com:/usr/obj/usr/src/sys/FOO  i386
>
> locked up sufficiently to require the reset button.  as it is remote, and
> i was not logged into the console, i do not know why.
>
> on restart, it seems to work, but i get
>
> KDB: stack backtrace:
> kdb_backtrace(c0680fb8,2,cbbea330,0,22) at kdb_backtrace+0x2e
> getdirtybuf(d6654bc0,0,1,cbbea330,1) at getdirtybuf+0x2b
> flush_deplist(c26ead4c,1,d6654be4,d6654be8,0) at flush_deplist+0x47
> flush_inodedep_deps(c1634800,1562b,3,d6654c34,c054b6fa) at flush_inodedep_deps+0x9e
> softdep_sync_metadata(d6654ca4,3,0,d6654c5c,c155f2d0) at softdep_sync_metadata+0x54
> ffs_fsync(d6654ca4,0,0,0,0) at ffs_fsync+0x47c
> fsync(c1937900,d6654d14,4,d6654d3c,c0516c66) at fsync+0x1a1
> syscall(2f,2f,bfbf002f,0,8125c90) at syscall+0x300
> Xint0x80_syscall() at Xint0x80_syscall+0x1f
> --- syscall (95, FreeBSD ELF32, fsync), eip = 0x283c59af, esp = 0xbfbfdaac, ebp = 0xbfbfe898 ---

These backtraces are generated by this code in getdirtybuf():
5847                 if (bp->b_vp == NULL)
5848                         kdb_backtrace();

Somehow there's a buf floating around without a vnode attached. I'd
suggest booting single user and fsck -y'ing all your filesystems to verify
there is no residual damage, and remove any snapshots from your ufs2
filesystems.

This isn't the first time I've seen this, so I'd check the archive too.

-- 
Doug White                    |  FreeBSD: The Power to Serve
dwhite_at_gumbysoft.com          |  www.FreeBSD.org