On 04/10/2013 06:07 PM, Joke de Buhr wrote: > yes i'm using the non-blocking mode. for client and server sesssion. > > gnutls_init(&session_, GNUTLS_DATAGRAM | GNUTLS_NONBLOCK | ...); > > > here is a stack trace where the pull function is called: > > #0 _gnutls_io_check_recv() at /lib/gnutls_buffers.c:677 > #1 _dtls_wait_and_retransmit() at /lib/gnutls_dtls.c:414 > #2 _gnutls_recv_handshake_final() at /lib/gnutls_handshake.c:2883 > #3 _gnutls_handshake_client() at /lib/gnutls_handshake.c:2730 > #4 gnutls_handshake() at /lib/gnutls_handshake.c:2450 > #5 session<server::gateway<server::server<boost::asio::ip::udp> > >> ::handshake() at /home/joke/Projects/vpn/src/session.t++:175 > > the pull timeout function returns GNUTLS_E_AGAIN. > but _dtls_retransmit is called only if the return value was GNUTLS_E_TIMEDOUT.
That's pretty correct. _gnutls_io_check_recv() returns GNUTLS_E_TIMEDOUT, zero or fatal error values. Your pull timeout function should return 0 on timeout, a positive number if data can be received and -1 on error. I'll make the documentation more clear on that, but in your case you should have returned zero instead of EAGAIN. regards, Nikos _______________________________________________ Gnutls-help mailing list [email protected] http://lists.gnupg.org/mailman/listinfo/gnutls-help
