null user patch
smfltc at us.ibm.com
Thu Nov 2 16:01:54 GMT 2006
Shirish S Pargaonkar wrote:
> If the user is not specified, should it be treated as anonymous instead of
> picking currently logged user?
> Also, I think cifs should not send password if user is anonymous.
If the user is not specified, I believe we still should choose the
"default" user (who is logged on), as
the anonymous user (null user) case is a special case, needed less often
(and forbidden by
configuration of the windows registry on some servers now in any
case). The open questions
seem to be:
1) should you be allowed to specify a password when "sec=none" ie null
I lean toward yes but it complicates the logic if the answer to the
below is also yes. I don't think we should prompt for the password
when "sec=none" but
still should allow "password=" on mount command
2) should for the default case (sec=none, no password specified on mount
for the null user be sent as zero length (instead of encrypting a
password of a single byte of null)?
3) should mount.cifs automatically retry with null user if a mount fails
with certain error codes
(access denied e.g.) as smbfs does. I suggest that we retry only if
"sec=" is not specified on
the command line and only on a small number of return codes on mount.
In addition I suggest
that we notify (printf to the command prompt) when we retry with null
user (as we do when
we retry with uppercased share name).
More information about the samba-technical