[linux-cifs-client] Re: dfs path construction fix - send dfs paths
on all path based operations on share in dfs
Steve French
smfrench at gmail.com
Thu May 15 06:22:28 GMT 2008
just checked in some of the GetDFSRefer cleanup into cifs-2.6. (with
minor style changes)
Missing still:
1) faking up inode after initial PATH_NOT_COVERED
2) cleanup of the code which parses the SMB response in GetDFSRefer
function in fs/cifs/cifssmb.c
On Wed, May 14, 2008 at 8:53 PM, Steve French <smfrench at gmail.com> wrote:
> I found the december version - and checked in part. The next two
> checkins will be:
>
> 1) the part of GetDFSRefer before the smb send which should work as is
> 2) the post-processing (looking at the SMB response buffer and taking
> the dfs structs out of it) which Christoph says was too big and needs
> to be put in a different function - so needs more coding
>
> On Wed, May 14, 2008 at 1:50 PM, Steve French <smfrench at gmail.com> wrote:
>> Do you have a newer version of the GetDFSRefer patch - or point me to
>> which you think is most current?
>>
>> On Wed, May 14, 2008 at 6:50 AM, Igor Mammedov <niallain at gmail.com> wrote:
>>> Steve French wrote:
>>>> Samba was not handling paths without the \\server\share prefix (which
>>>> our current code sends on QueryPathInfo) when "SHARE_IN_DFS" on
>>>> operations such as rmdir and delete.
>>>>
>>>> This patch fixes that:
>>>
>>>
>>> Attempted to test patch and
>>>
>>> -
>>> at the first attempt I've got oops caused by a missing CIFSGetDFSRefer patch,
>>> after digging it up and applying it, oops is disappeared. But it doesn't
>>> help much because your patch broke traversal over a dfs link, at the moment
>>> dfs lookup relies on the second lookup attempt with a short path name (without
>>> tree name).
>>> BTW, you said that you almost made patch for a fake dfs inode, it could fix
>>> things up.
>>>
>>> +
>>> mkdir and rmdir on a dfs enabled share work (at least with samba), can't check
>>> it on MS however.
>>>
>>> PS:
>>> commiting CIFSGetDFSRefer patch will help to avoid oops-es and more people
>>> would be able to test the DFS related patches.
>>>
>>> --
>>>
>>> Best regards,
>>>
>>> -------------------------
>>> Igor Mammedov,
>>> niallain "at" gmail.com
>>>
>>>
>>>
>>>
>>>
>>
>>
>>
>> --
>> Thanks,
>>
>> Steve
>>
>
>
>
> --
> Thanks,
>
> Steve
>
--
Thanks,
Steve
More information about the linux-cifs-client
mailing list