On Oct 28, 2016, at 6:57 AM, Ross Finlayson <[email protected]> wrote: > >> Also, whilst I'm not arguing about the validity or otherwise of the cameras >> response (I'm no RTP expert), isn't it reasonable to interpret this as a >> security hole in the live555 library? > > No, because IP source addresses can always be forged - so they should never > be used as a security mechanism.
If that’s what we’re trying to solve — as opposed to the non-ephemeral port number issue — then wouldn’t switching to RTP over TCP work? You can’t usefully forge IPs with TCP because the SYN-ACK can’t go back to the forged IP, so the third handshake packet never happens. And you can’t inject a frame into the stream, because you probably can’t guess sequence numbers and such unless you’re in a MITM position. As for MITM mitigation, that’s the same as always: TLS, VPN, or similar. So, the security bug here is unencrypted UDP, not Live555. _______________________________________________ live-devel mailing list [email protected] http://lists.live555.com/mailman/listinfo/live-devel
