2 Aug
2004
2 Aug
'04
11:03 a.m.
hi,
As i understand it would be virtually impossible to replay with proper
timestamp for certain request and with required ip, but nevetheless it
would be possible to fake required source_ip and in that case my patch
(posted earlier on this list) wouldn't help too. Adding source ip to nonce
would remove additional administrative burden, isn't it? Probably the only
advanatage of authorization by entering IP adresses separately would be
independance from any other encryption or authorization mechanisms which
could be used with SIP.
Antanas
On Sun, 1 Aug 2004, Jan Janak wrote:
No, but I have this on my todo list. Currently it is
possible to re-use
credentials generated by user agent A also by user agent B provided that
it can sniff the SIP messages and it is fast enough to send another
message including the sniffed credentials. The credentials have limited
lifetime (1 minute by default) so after 1 minute they cannot be re-used
in other SIP messages.
If you have two user agents connected to the same hub (so that they can
see SIP messages of each other) then you can modify one of them to steal
the calls to the other user agent using sniffed digest credentials.
I am thinking about including the source IP address of the SIP message
and some other header fields (Contact) into nonce to eliminate this
weakness.
Jan.
On 28-07 15:00, zolia(a)z1sys.com wrote:
hello,
is it possible to do source ip authentication besides normal
www_authorize() for every user account?. This, as i understand, should
prevent from intercepting credentials and later faking sip message to
bypass www_authorization ? Or maybe there are some other counter measures
against such fraud?
Does src_ip comes directly from ip layer? If so, i could probably use this
to check with some external database (ie. ser subscriber)?
Antanas
NTT
_______________________________________________
Serusers mailing list
serusers(a)lists.iptel.org
http://lists.iptel.org/mailman/listinfo/serusers