[cifs-protocol] [REG:111081664438980] RE: Level 257 FindFirst rejected by some Windows servers even though NTLM
George K Colley
gcolley at apple.com
Wed Aug 24 16:00:11 MDT 2011
On Aug 24, 2011, at 2:23 PM, Steve French wrote:
> On Tue, Aug 23, 2011 at 10:43 PM, Hongwei Sun <hongweis at microsoft.com> wrote:
>> Steve,
>>
>> Windows CE is not within the scope of protocol documentation, such as MS-SMB and MS-CIFS. Therefore it is understandable that it doesn't behave as specified in the protocol documents.
>
> This is more about what Windows clients do not about the WindowsCE server.
>
> Clearly windows clients works (to WindowsCE server) - and it appears
> it is because they choose levels carefully to avoid the WindowsCE
> server problems
>
> instead of
> FIND_FILE_DIRECTORY_INFO (level 257)
> nor
> FIND_FILE_FULL_DIRECTORY_INFO (258)
> nor
> FIND_FILE_ID_FULL_DIRECTORY_INFO (262)
>
> Windows clients (XP, Vista, etc) know enough to send
> SMB_FIND_FILE_BOTH_DIRECTORY_INFO (260)
>
> which doesn't make sense since FIND_FILE_BOTH_DIRECTORY_INFO requires
> the server to return the short name (which no longer seems relevant to
> windows clients but they are requesting it - even of WindowsCE).
I do not believe it requires short names. See {MS-SMB].pdf 2.2.4.7 SMB_COM_TREE_CONNECT_ANDX (0x75)
When a server returns extended information, the response takes the following format. Aside from the WordCount, MaximalShareAccessRights, and GuestMaximalShareAccessRights fields, and the new OptionalSupport flags, all other fields are defined as specified in [MS-CIFS] section
SMB_UNIQUE_FILE_NAME
If set, then the server is using long file names only and does not support short file names. If set, then the server allows the client to assume that there is no name aliasing for this share (in other words, a single file cannot have two different names). If set, then the server permits the client to cache directory enumerations and file metadata based on the pathname.The client MAY<48> choose to satisfy file attribute queries from its cache and thus could present a slightly stale view of files on the share. The client MUST NOT cache remote file system information for more than 60 seconds.
>
> Windows is NOT returning operation not supported (or the eqiuvalent)
> to the application, rather it is selectively choosing to use level 260
> (rather than the 3 other more logical find levels)
>
> So the question is - how does Windows (clients) determine which level
> to request on FindFirst - in particular when not to use 257, 258 or
> 262 and fall back to 260?
>
>> As far as Windows systems, as per 2.2.2.3.1 MS-CIFS, for Windows NT and earlier, the Find information levels supported are clearly specified
>>
>> SMB_INFO_STANDARD 0x0001 (LANMAN2.0)
>> SMB_INFO_QUERY_EA_SIZE 0x0002 (LANMAN2.0)
>> SMB_INFO_QUERY_EAS_FROM_LIST 0x0003 (LANMAN2.0)
>> SMB_FIND_FILE_DIRECTORY_INFO 0x0101 (NT LANMAN)
>> SMB_FIND_FILE_FULL_DIRECTORY_INFO 0x0102 (NT LANMAN)
>> SMB_FIND_FILE_NAMES_INFO 0x0103 (NT LANMAN)
>> SMB_FIND_FILE_BOTH_DIRECTORY_INFO 0x0104 (NT LANMAN)
>>
>> For Windows 2000 and later , in addition to the levels above , the following levels are added as per MS-SMB 2.2.6.1.1
>>
>> SMB_FIND_FILE_ID_FULL_DIRECTORY_INFO 0x0105 (NT LANMAN)
>> SMB_FIND_FILE_ID_BOTH_DIRECTORY_INFO 0x0106 (NT LANMAN)
>>
>> As per 2.2.8 MS-CIFS, The client MUST map the application-provided [MS-FSCC] information levels to SMB information Levels. For all other [MS-FSCC] information levels, the client MUST fail the request with STATUS_NOT_SUPPORTED. In some case, the client MUST send a fixed level. For example, a client that has not negotiated long names support MUST request only SMB_INFO_STANDARD.
>>
>> Please let us know if you have more questions.
>>
>> Thanks!
>>
>> Hongwei
>>
>>
>>
>>
>> -----Original Message-----
>> From: cifs-protocol-bounces at cifs.org [mailto:cifs-protocol-bounces at cifs.org] On Behalf Of Steve French
>> Sent: Thursday, August 18, 2011 4:54 PM
>> To: Edgar Olougouna
>> Cc: pfif at tridgell.net; cifs-protocol at samba.org
>> Subject: Re: [cifs-protocol] Level 257 FindFirst rejected by some Windows servers even though NTLM
>>
>> It looks like Windows CE takes (only?) level 260 but I can't easily prove it without access to a test system (I just have some customer traces) - so how does Windows clients (Windows XP/Vista/7 etc.) determine which FindFirst level to send to these given that the Microsoft server in this case is reporting NT Find and NT SMB support but in practice not supporting most FindFirst levels.
>>
>> On Tue, Aug 16, 2011 at 12:56 PM, Edgar Olougouna <edgaro at microsoft.com> wrote:
>>> [Dochelp to bcc]
>>>
>>> Steve,
>>>
>>> One of our engineers will follow-up soon on this inquiry. The case number is 111081664438980.
>>>
>>> Regards,
>>> Edgar
>>>
>>> -----Original Message-----
>>> From: Steve French [mailto:smfrench at gmail.com]
>>> Sent: Tuesday, August 16, 2011 12:35 PM
>>> To: Interoperability Documentation Help
>>> Cc: cifs-protocol at samba.org; pfif at tridgell.net
>>> Subject: Level 257 FindFirst rejected by some Windows servers even
>>> though NTLM
>>>
>>> A user sent me a trace of FindFirst level 257 (0x101 ) failing to
>>> Windows CE with NT Status: STATUS_INVALID_LEVEL (0xc0000148)
>>>
>>> even though dialect negotiated was NT LM 012 and that dialect is the only prereq listed in MS-SMB for the level (see page 64).
>>>
>>> How can the client determine under what condition that the server does
>>> not support that level - - and what level to fall back (or move up to higher level)? Level 257 is pretty basic.
>>>
>>>
>>> --
>>> Thanks,
>>>
>>> Steve
>>>
>>>
>>
>>
>>
>> --
>> Thanks,
>>
>> Steve
>> _______________________________________________
>> cifs-protocol mailing list
>> cifs-protocol at cifs.org
>> https://lists.samba.org/mailman/listinfo/cifs-protocol
>>
>>
>
>
>
> --
> Thanks,
>
> Steve
> _______________________________________________
> cifs-protocol mailing list
> cifs-protocol at cifs.org
> https://lists.samba.org/mailman/listinfo/cifs-protocol
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.samba.org/pipermail/cifs-protocol/attachments/20110824/e1fbc662/attachment-0001.html>
More information about the cifs-protocol
mailing list