Hi,

we have a question which is not directly related to SIP but SRTP, but we hope 
someone can still clarify the scenario.

We are encountering an issue where the authentication tag validation for a 
sequence of SRTP packets (SSRC 0xCB23BAA3) starts failing after processing 
several packets. This failure occurs after the sender delivers around 15 G.722 
packets with a different SSRC (0xB12E5B92). Source and destination address are 
the same for both streams. The reason for these G.722 packets being sent is 
unclear, especially since the call was negotiated to use G.711. We would 
consider this the first issue.

Here is a sample of the packet sequence we received:

SSRC        SEQ      Codec      Timestamp
0xCB23BAA3  1363     G.711      597285443
0xCB23BAA3  1364     G.711      597285603
0xCB23BAA3  1365     G.711      597285763
...
0xCB23BAA3  1379     G.711      597288003
0xB12E5B92  58505    G.722      2973790942
0xB12E5B92  58506    G.722      2973791102
0xCB23BAA3  1380     G.711      597288163
...

The first packet failing authentication is 0xCB23BAA3/1380. All subsequent 
packets from this stream also fail authentication.

After capturing the scenario and debugging it, we found that authentication 
would succeed if the sequence from 0xB12E5B92/58506 to 0xCB23BAA3/1380 was 
considered a rollover of sequence numbers, thereby increasing the "v" according 
to RFC 3711 when calculating the authentication tag.

According to our understanding of RFC 3711, "s_l" (the highest received 
sequence number) is part of the cryptographic context, and each SSRC has its 
own cryptographic context:
> A cryptographic context SHALL be uniquely identified by the triplet context 
> identifier:
> context id = <SSRC, destination network address, destination transport port 
> number>

It seems to us that the sender is incorrectly authenticating subsequent packets 
of stream 0xCB23BAA3, possibly not considering 0xCB23BAA3 and 0xB12E5B92 as 
different cryptographic contexts.

Thanks for any feedback.

Regards,
Schraffl Peter
_______________________________________________
Sip-implementors mailing list
Sip-implementors@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/sip-implementors

Reply via email to