Bug#695492: Info received (Bug#695492: [wheezy] Ctrl-C on program accessing CIFS mount breaks/hangs mount)
OK... I have some info on this. I loaded up a VM with a
stock install of squeeze with the following kernel:
ii linux-image-2.6.32-5-686 2.6.32-46 Linux 2.6.32 for modern PCs
After mounting the six CIFS drives, I ran "find" and "du"
command many, many... times and then inturrupting them wirh
CTRL-C and the mounts were stable. I did see a little cruft
in the kern.log as follows:
.
.
.
Dec 26 10:37:02 joe kernel: [ 6.639592] ADDRCONF(NETDEV_UP): eth0: link is not ready
Dec 26 10:37:02 joe kernel: [ 6.641458] e1000: eth0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX
Dec 26 10:37:02 joe kernel: [ 6.641953] ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
Dec 26 10:37:12 joe kernel: [ 16.956905] eth0: no IPv6 routers present
Dec 26 10:37:25 joe kernel: [ 29.932683] Registering the dns_resolver key type
Dec 26 10:37:25 joe kernel: [ 29.932690] Slow work thread pool: Starting up
Dec 26 10:37:25 joe kernel: [ 29.932707] Slow work thread pool: Ready
Dec 26 11:24:44 joe kernel: [ 2869.168062] CIFS VFS: No response for cmd 50 mid 24451
Dec 26 11:25:00 joe kernel: [ 2885.237227] CIFS VFS: No response for cmd 50 mid 29
Dec 26 11:25:00 joe kernel: [ 2885.237242] CIFS VFS: Unexpected lookup error -112
<log end>
I have seen the above CIFS log entries before but they seem
benign.
Next... I loaded up a VM with a stock install of wheezy with
the following kernel:
ii linux-image-3.2.0-4-486 3.2.35-2 i386 Linux 3.2 for older PCs
After mounting the six CIFS drives, I ran a "find" on one of
the drives and then hit CTRL-C. All six drives went dead and
were inaccessable, even though the mount command shows that
they are mounted. The following was in the kern.log:
.
.
.
Dec 26 11:22:04 jax kernel: [ 7.934294] e1000: eth0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX
Dec 26 11:22:04 jax kernel: [ 7.934743] ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
Dec 26 11:22:04 jax kernel: [ 9.192704] vboxsf: Successfully loaded version 4.1.18_Debian (interface 0x00010004)
Dec 26 11:22:13 jax kernel: [ 18.656029] eth0: no IPv6 routers present
Dec 26 11:22:16 jax kernel: [ 21.367504] FS-Cache: Netfs 'cifs' registered for caching
Dec 26 11:22:16 jax kernel: [ 21.400621] CIFS VFS: default security mechanism requested. The default security mechanism will be upgraded from ntlm to ntlmv2 in kernel release 3.3
Dec 26 11:22:42 jax kernel: [ 47.502297] CIFS VFS: Unexpected lookup error -512
Dec 26 11:22:43 jax kernel: [ 47.695430] CIFS VFS: No task to wake, unknown frame received! NumMids 0
Dec 26 11:22:43 jax kernel: [ 47.695433] Received Data is: : dump of 37 bytes of data at 0xf63b1040
Dec 26 11:22:43 jax kernel: [ 47.695436] 23000000 424d53ff 000022a4 c00580c0 . . . # ▒ S M B ▒ " . . ▒ . . ▒
Dec 26 11:22:43 jax kernel: [ 47.695438] 48df0000 857b2fd2 00003d1e 0a1e8007 . . H / { . . = . . . . . .
Dec 26 11:22:43 jax kernel: [ 47.695440] 0fd17000 f2000000 . p . .
Dec 26 11:22:44 jax kernel: [ 48.736690] CIFS VFS: Unexpected SMB signature
Dec 26 11:22:44 jax kernel: [ 48.737697] CIFS VFS: Unexpected SMB signature
Dec 26 11:22:44 jax kernel: [ 48.738947] CIFS VFS: Unexpected SMB signature
.
.
.
The "CIFS VFS: Unexpected SMB signature" errors stack up
quickly as if something is looping.
Next... I loaded a kernel into the wheezy VM as follows:
ii linux-image-3.6-trunk-486 3.6.9-1~experimental.1 i386 Linux 3.6 for older PCs
I then mounted the same six CIFS net drives and issued the
same "find" command and then hit CTRL-C after it had run for
about 10 seconds. Again, all of the CIFS mounts became
inaccessable. The following was in the kern.log:
.
.
.
Dec 26 11:11:15 jax kernel: [ 7.768716] e1000: eth0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX
Dec 26 11:11:15 jax kernel: [ 7.768995] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
Dec 26 11:14:07 jax kernel: [ 179.938244] FS-Cache: Netfs 'cifs' registered for caching
Dec 26 11:14:07 jax kernel: [ 179.938253] Key type cifs.spnego registered
Dec 26 11:14:07 jax kernel: [ 179.938259] Key type cifs.idmap registered
Dec 26 11:14:07 jax kernel: [ 179.938388] CIFS: no cache= option specified, using "cache=loose". This default will change to "cache=strict" in 3.7.
Dec 26 11:14:07 jax kernel: [ 179.959184] CIFS VFS: default security mechanism requested. The default security mechanism will be upgraded from ntlm to ntlmv2 in kernel release 3.3
Dec 26 11:15:03 jax kernel: [ 236.412231] CIFS VFS: SMB signature verification returned error = -13
Dec 26 11:15:03 jax kernel: [ 236.413194] CIFS VFS: SMB signature verification returned error = -13
.
.
.
I noted that in the above log there is no data "dump" as in
the previous log. Also the repeating error is now:
"CIFS VFS: SMB signature verification returned error = -13"
...which is different from the previous log.
In all cases the drives and be fixed by remounting.
let me know if you need any other information.
-- john
Reply to: