another buildworld -j4 panic
Bosko Milekic
bmilekic at technokratis.com
Fri Aug 8 06:56:50 PDT 2003
On Thu, Aug 07, 2003 at 05:36:38PM -0500, Mike Silbersack wrote:
>
> ok, here goes:
>
> duplicate free from zone FFS1 dinode
>
> traceback:
> Debugger
> panic
> uma_dbg_free
> uma_zfree_arg
> ffs_ifree
> ufs_reclaim
> ufs_vnoperate
> vclean
> gdonel
> getnewvnode
> ffs_vget
> ufs_lookup
> ufs_vnoperate
> vfs_cache_lookup
> ufs_vnoperate
> lookup
> namei
> stat
> syscall
> Xint0x80_syscall
>
> Whee!
geez, you're really hitting a lot of double frees. Can the other
people running -current under high load turn on INVARIANTS and see if
they start hitting the double free problems, too? I'd like some more
evidence.
Judging from the things we've seen from your traces I would guess at
this point that there is a path somewhere that ends up in the ffs code
without Giant, and that code fails to assert Giant requirements most
of the time. Maybe it's time we start sprinkling the GIANT_REQUIREDs
where we know they belong... that one we talked about adding the
other day, did it catch anything?
--
Bosko Milekic * bmilekic at technokratis.com * bmilekic at FreeBSD.org
TECHNOkRATIS Consulting Services * http://www.technokratis.com/
More information about the freebsd-current
mailing list