[linux-cifs-client] review 5, was Re: projected date for mount.cifs to support DFS junction points

[Igor Mammedov]

Steve French wrote:
> In testing this (which I plan to push up to cifs-2.6.git today), I
> have been running into problems with the retry that you do.  With Unix
> Extensions enabled Samba does not seem to like the path (looks like a
> bug in the server recognizing dfs paths in POSIX form with / instead
> of \).

it will not work int this case cause samba tell that it is symlink,
and client happily show it as symlink. I dont think that we could fix
anything here on the client side because it must display symlinks with
unix ext. enabled. It can be fixed on the server side by hiding dfs links
from client (bug in samba).

> 
> Disabling Unix Extensions to Samba (to look like Windows), I noticed
> that you get a path_not_found error on the retry (ie
> \\server\share\dfs-ref works but not the retry on \dfs-ref) - this
> happens even if you turn off dfs support in SMB flags2 (Samba seems to
> ignore that flag on QueryPathInfo).
> 
> The mem leak fix helps - but it looks like we have to put in default
> values for the inode (which is soon to be covered anyway) - the one
> question I had was how to generate an inode number when we are using
> the server's inode numbers for the directory (since we can't seem to
> get the server to give us info on the inode without a path not found
> or a path not covered error coming back - since it is a dfs junction).

I've noticed that mountpoint inode_ino is always = 2. it doesn't get inode
number share root form  the server even if we use serverino mount option.

So userspace code will see ino=2 in junction point (mounted dfs refferal)
and will not see actual ino of the inode we are mounted on. 
May be we can use it for thinking up some default ino value for dfs junction
point inode?

Another way is to fix bug in samba so that it could return path info when
path is not in dfs format. (how else we can get server generated ino if 
server refuses to do it)



> 
> On Tue, Mar 11, 2008 at 4:41 AM, Q (Igor Mammedov)
> <qwerty0987654321 at mail.ru> wrote:
>> Mem leak fix in inode.c
>>
> 
> 
> 


-- 

Best regards,

-------------------------
Igor Mammedov,
niallain "at" gmail.com