[Samba] Cannot mount a remote volume after system upgrade

Günter Kukkukk linux at kukkukk.com
Mon Jun 9 18:25:33 MDT 2014


Am 10.06.2014 01:19, schrieb James Moe:
> On 06/09/2014 12:28 PM, Günter Kukkukk wrote:
> 
>   I tried adding the "ip" and "port" options. It made no difference; it
> fails to mount.
> 
>>
>> echo 7 > /proc/fs/cifs/cifsFYI
>> This enables the highest cifs module debug level, which you then can watch (during mount) with
>>    tail -f /var/log/messages
>>
>   Here is an excerpt of the trace; I have the complete trace if you
> would like to see that.
> 
> .../connect.c: build_unc_path_to_root: full_path=\\sma-server1\pub-data
> .../transport.c: For smb_command 117
> .../transport.c: Sending smb: smb_len=72
> .../connect.c: RFC1002 header 0x2d
> .../misc.c: checkSMB Length: 0x31, smb_buf_length: 0x2d
> .../transport.c: cifs_sync_mid_result: cmd=117 mid=4 state=4
> .../connect.c: Tcon rc = 0 ipc_tid = 33
> .../cifssmb.c: In GetDFSRefer the path \sma-server1\pub-data
> .../transport.c: For smb_command 50
> .../transport.c: Sending smb: smb_len=92
> .../connect.c: RFC1002 header 0x21
> 
> .../misc.c: checkSMB Length: 0x25, smb_buf_length: 0x21
> .../smb1ops.c: invalid transact2 word count
> 
> .../transport.c: cifs_sync_mid_result: cmd=50 mid=5 state=4
> .../netmisc.c: Mapping smb error code 0x10002 to POSIX err -5
> .../cifssmb.c: Send error in GetDFSRefer = -5

The cifs kernel module code usually uses recent smb calls, which
often fail. But it then falls back to try legacy cmds.
The GetDFSRefer call must fail, because OS/2 doesn't even know about it.
(this might also explain "the [Malformed Packet] response.)

> 
> This corresponds with the Wireshark trace excerpt:
> 
> 192.168.69.249	192.168.69.115 SMB Tree Connect AndX Response
> 192.168.69.115	192.168.69.249 SMB Trans2 Request, GET_DFS_REFERRAL,
> File: \sma-server1\pub-data
> 192.168.69.249	192.168.69.115 SMB Trans2 Response,
> GET_DFS_REFERRAL[Malformed Packet]
> 192.168.69.115	192.168.69.249 SMB Logoff AndX Request
> 
>   However, I do not really have a clue what to infer from the above.
> There appears to be a level of error checking in the newer version of
> SMB that did not exist in samba v3.6.3.

this has nothing to do with the supplied samba version. The mount command
uses the *kernel* supplied cifs.ko module!

In addition please post the following infos:
   uname -a
   modinfo cifs
   the *exact* mount command you are using

Btw - sometimes it's easier to join the #samba IRC channel on irc.freenode.net
and ask questions there. :-)
Then questions can be handled in "somewhat real time" ....

Cheers, Günter

ps - i'm running a 64Bit kernel here:
Linux li4771-131 3.11.10-11-desktop #1 SMP PREEMPT Mon May 12 13:37:06 UTC 2014 (3d22b5f) x86_64 x86_64 x86_64 GNU/Linux

-- 



More information about the samba mailing list