[linux-cifs-client] (no subject)

[Dave Kleikamp]

On Thu, 2009-01-08 at 13:35 +0100, Eric Sesterhenn wrote:
> hi,
> 
> with current -git i can trigger a box lockup with samba and iozone.
> 
> The relevant .config entries:
> root at whiterabbit:/usr/src/linux# grep CIFS .config
> CONFIG_CIFS=y
> # CONFIG_CIFS_STATS is not set
> CONFIG_CIFS_WEAK_PW_HASH=y
> # CONFIG_CIFS_UPCALL is not set
> CONFIG_CIFS_XATTR=y
> CONFIG_CIFS_POSIX=y
> CONFIG_CIFS_DEBUG2=y
> CONFIG_CIFS_EXPERIMENTAL=y
> # CONFIG_CIFS_DFS_UPCALL is not set
> root at whiterabbit:/usr/src/linux# grep SMB .config
> # CONFIG_SMB_FS is not set
> 
> To reproduce:
> 
> mkdir /media/samba > /dev/null 2>&1
> mount -t cifs //WHITERABBIT/data /media/samba -o password=root
> touch /media/samba/stress/nfs2
> chown nobody:nobody /media/samba/stress/nfs2
> /usr/bin/iozone -a -s 1024 -f /media/samba/stress/nfs2
> 
> The exported directory lies on an ext3 partition
> 
> Kernel reports the following:
> 
> [  238.935942] BUG: scheduling while atomic: iozone/5035/0x00000002
> [  238.936214] INFO: lockdep is turned off.
> [  238.936330] Modules linked in:
> [  238.936524] Pid: 5035, comm: iozone Tainted: G        W
> 2.6.28-07485-g9e42d0c #177
> [  238.936706] Call Trace:
> [  238.936853]  [<c0127137>] __schedule_bug+0x63/0x6a
> [  238.936986]  [<c07f6f51>] schedule+0x99/0x947
> [  238.937170]  [<c014c9d2>] ? print_lock_contention_bug+0x11/0xb2
> [  238.937300]  [<c014e553>] ? trace_hardirqs_on_caller+0x18/0x15f
> [  238.937450]  [<c07f9df4>] ? _spin_unlock_irqrestore+0x47/0x5d
> [  238.937591]  [<c01401bd>] ? prepare_to_wait+0x46/0x4c
> [  238.937745]  [<c0145031>] async_synchronize_cookie_special+0xac/0x103
> [  238.937873]  [<c0140060>] ? autoremove_wake_function+0x0/0x35
> [  238.938038]  [<c01450d0>] async_synchronize_full_special+0x1a/0x1c
> [  238.938172]  [<c01a1e6d>] sync_filesystems+0x4f/0xf9
> [  238.938323]  [<c01b965d>] do_sync+0x2f/0x5f
> [  238.938442]  [<c01b969f>] sys_sync+0x12/0x17
> [  238.938584]  [<c0102f21>] sysenter_do_call+0x12/0x31

etc.

This isn't a cifs problem.  I followed up to another report of the
problem here: http://marc.info/?l=linux-kernel&m=123142559722824&w=2

Thanks.  I probably wouldn't have seen that one if you hadn't reported
it here.

Shaggy
-- 
David Kleikamp
IBM Linux Technology Center