ZFS panic: existing znode

Kip Macy kmacy at freebsd.org
Tue May 12 21:47:39 UTC 2009


This is fairly easy to reproduce with stress2 + fsstress. I haven't
had time to track down the locking bug yet.


Cheers,
Kip


On Tue, May 12, 2009 at 12:39 PM, Matthew West <mwest at l.zeeb.org> wrote:
> FreeBSD 8-CURRENT, built from sources around 27/02/2009:
>
> FreeBSD foo.internal 8.0-CURRENT FreeBSD 8.0-CURRENT #5: Fri Apr 17 18:33:02 BST 2009 mwest at foo.internal:/usr/obj/usr/src/sys/DEBUGLOCK amd64
>
> The system is AMD64, with 16GB of RAM, serving a few hundred clients via
> NFS (v2 and v3) and Samba, from a 800GB ZFS pool; using hardware RAID
> (aac controller), not RAID-Z.
>
> Running a GENERIC kernel, but with the following options enabled:
>
> options DEBUG_LOCKS
> options DEBUG_VFS_LOCKS
> options DIAGNOSTIC
>
> I also have Jaakko Heinonen's patch to zfs_znode.c applied, from:
> http://www.freebsd.org/cgi/query-pr.cgi?pr=132068
>
> After almost a week of active usage, there was a system panic.
>
> ----------
> panic: existing znode 0xffffff007dfd3d38 for dbuf 0xffffff00aeeb9620
> cpuid = 2
> KDB: stack backtrace:
> db_trace_self_wrapper() at db_trace_self_wrapper+0x2a
> panic() at panic+0x182
> zfs_znode_dmu_init() at zfs_znode_dmu_init+0xb5
> zfs_znode_alloc() at zfs_znode_alloc+0xa0
> zfs_mknode() at zfs_mknode+0x205
> zfs_freebsd_create() at zfs_freebsd_create+0x617
> VOP_CREATE_APV() at VOP_CREATE_APV+0xb3
> nfsrv_create() at nfsrv_create+0x909
> nfssvc() at nfssvc+0x4af
> syscall() at syscall+0x1e7
> Xfast_syscall() at Xfast_syscall+0xab
> --- syscall (155, FreeBSD ELF64, nfssvc), rip = 0x800695c4c, rsp = 0x7fffffffe8e8, rbp = 0 ---
> ----------
>
> I see Kris Kennaway encountered a similar crash, but no one seems to
> have replied:
>
>  http://lists.freebsd.org/pipermail/freebsd-current/2009-January/002631.html
>
> I was unfortunately not able to generate a crash dump.
>
> Let me know if there's any further information I can provide which
> might be useful.
>
> Thanks,
>
> Matthew
>
> _______________________________________________
> freebsd-current at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-current
> To unsubscribe, send any mail to "freebsd-current-unsubscribe at freebsd.org"
>



-- 
When bad men combine, the good must associate; else they will fall one
by one, an unpitied sacrifice in a contemptible struggle.

    Edmund Burke


More information about the freebsd-current mailing list