Re: "recursive lock for object" & "unlocking unheld lock" in smb

From: Sergey Zaharchenko <doublef-ctm_at_yandex.ru>
Date: Mon, 8 Jan 2007 12:04:14 +0300
Hello Alexander!

Wed, Sep 27, 2006 at 08:34:38PM +0200 you wrote:

> Quoting John Baldwin <jhb_at_freebsd.org> (Wed, 27 Sep 2006 11:44:23 -0400):
> 
> > On Wednesday 27 September 2006 10:01, Alexander Leidinger wrote:
> > > Hi,
> > > 
> > > yesterday I rsynced a smb share from a samba-3.0.x server (FreeBSD 4)  
> > > via a smb mount (current from Sep 23) "locally" (mount -t smbfs from  
> > > the samba server and rsync a/ b/; most easy solution to convert some  
> > > ISO-8859-1 filenames to UTF-8 ("dos charset = UTF8" in smb.conf!)  
> > > while moving to another system).
> > > 
> > > Today I noticed the following in the daily mail on the -current system:
> > 
> > If you are willing to panic your box, can you try the patch below and get a 
> 
> Not this one. But I try to reproduce it with another one later this
> week (not today and not tomorrow).
> 

I'm just writing to note that I've observed the same:

smb_iod_recvall: drop resp with mid 1
smb_co_lock: recursive lock for object 1
lockmgr: thread 0xc3a78c40 unlocking unheld lock
KDB: stack backtrace:
db_trace_self_wrapper(c0939b3c) at db_trace_self_wrapper+0x25
kdb_backtrace(c0934cf7,c3a78c40) at kdb_backtrace+0x29
_lockmgr(c4213408,2006,c4213434,c3a78c40,c41e22b0,...) at _lockmgr+0x5ca
smb_co_put(c4213400,d63a1b8c,c41fe800,c4213400,0,...) at smb_co_put+0x50
smb_sm_lookup(d63a1b18,d63a1af8,d63a1b8c,d63a1af4,d63a1af8,...) at smb_sm_lookup+0x11a
smb_usr_lookup(c3397c00,d63a1b8c,d63a1b88,d63a1b84,c0a86358,...) at smb_usr_lookup+0x76
nsmb_dev_ioctl(c420bb00,82fc6e6a,c3397c00,3,c3a78c40,c0a391e8,0,c0932a36,131) at nsmb_dev_ioctl+0x1e5
giant_ioctl(c420bb00,82fc6e6a,c3397c00,3,c3a78c40,...) at giant_ioctl+0x33
devfs_ioctl_f(c3f77828,82fc6e6a,c3397c00,c3ceae80,c3a78c40) at devfs_ioctl_f+0xaf
kern_ioctl(c3a78c40,4,82fc6e6a,c3397c00) at kern_ioctl+0x296
ioctl(c3a78c40,d63a1d04) at ioctl+0xf1
syscall(3b,3b,3b,bfbfe780,bfbfe280,...) at syscall+0x256
Xint0x80_syscall() at Xint0x80_syscall+0x1f
--- syscall (54, FreeBSD ELF32, ioctl), eip = 0x281562db, esp = 0xbfbfe26c, ebp = 0xbfbfe598 ---
smb_iod_recvall: drop resp with mid 3

with -CURRENT from Fri Dec  1 19:31:56 MSK 2006 and windows xp as the
server when copying large files to/from the server. So it's still there
and the server being FreeBSD is probably irrelevant. Has anything been
done since or should I try the patch and send the backtrace?

P.S. A word about my PR's:

PR 102747 has been sitting there for a while, all my mail to either
brian_at_freebsd.org or brian_at_Awfulhak.org was rejected. I've been running
with the patch for all this time without problems. Anyone?

PR 104406 should probably be closed; it's a bit weird case of a bug
which looks inherent to idprio when there's no priority propagation (all
processes get to a sleep queue and will only wake up when the
low-priority process has run, which is almost never).

-- 
DoubleF
No virus detected in this message. Ehrm, wait a minute...
/kernel: pid 56921 (antivirus), uid 32000: exited on signal 9
Oh yes, no virus:)

Received on Mon Jan 08 2007 - 08:18:49 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:04 UTC