DeforaNetworks/siproxd
DeforaNetworks/siproxd
3
0
Fork 0
Code Releases Activity

feature: siproxd "in front of" a NAT router should work

Browse Source
This commit is contained in:
Thomas Ries 2005-03-27 16:05:16 +00:00
parent 533a1f4c56
commit 8f211d48af
3 changed files with 16 additions and 25 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
ChangeLog
View File

@ -1,5 +1,6 @@
0.5.11
======
27-Mar-2005: - feature: siproxd "in front of" a NAT router should work
20-Mar-2005: - when allocating local ports for RTP relaying, do this
in a round-robin way over the available port range.
1-Mar-2005: - working on running siproxd in front of a routing device

38
src/proxy.c
View File

@ -671,8 +671,8 @@ int proxy_response (sip_ticket_t *ticket) {
(e.g 1393xxx@proxy01.sipphone.com for calls made sipphone -> FWD)
How can we deal with this? Should I take into consideration the 'Via'
headers? This is the only clue I have, pointing to the *real* UA.
Maybe I should put in a 'siproxd' ftag value to recognize it a header
put in by myself
Maybe I should put in a 'siproxd' ftag value to recognize it as a header
inserted by myself
*/
if ((type == 0) && (!osip_list_eol(response->vias, 0))) {
osip_via_t *via;
@ -1084,7 +1084,6 @@ if (configuration.debuglevel)
* Rewrite
* an IP address of 0.0.0.0 means *MUTE*, don't rewrite such
*/
/*&&&& should use gethostbyname here */
if (strcmp(sdp->c_connection->c_addr, "0.0.0.0") != 0) {
osip_free(sdp->c_connection->c_addr);
sdp->c_connection->c_addr=osip_malloc(HOSTNAME_SIZE);
@ -1129,7 +1128,6 @@ if (configuration.debuglevel)
have_c_media=0;
sdp_conn=sdp_message_connection_get(sdp, media_stream_no, 0);
if (sdp_conn && sdp_conn->c_addr) {
/*&&&& should use gethostbyname here as well */
if (strcmp(sdp_conn->c_addr, "0.0.0.0") != 0) {
sts = get_ip_by_host(sdp_conn->c_addr, &addr_media);
have_c_media=1;
@ -1173,26 +1171,18 @@ if (configuration.debuglevel)
memcpy(&addr_media, &addr_sess, sizeof(addr_sess));
}
/*&&&& If I'm sitting BEFORE the actual masquerading router:
ok: I _must_ _not_ _try_ _to_ _use_ _my_ _outbound_ _address_ (map_addr)
but my inbound address instead. Also make sure that the port
seems ok: assignment does not get fucked up.
todo: RPORT option must be removed.
seens ok: Possibly a new RTP_DIRECTION value is needed ("no double assignments
of same port on different IPs" or something similar)??
*/
/*
* I am running in front of the routing device. I cannot use the
* external IP to bind a listen socket to, so force the use of
* my inbound IP for listening
*/
if ((rtp_direction == DIR_INCOMING) &&
(configuration.outbound_host) &&
(strcmp(configuration.outbound_host, "")!=0)) {
/*&&&&*/
INFO("**** Front-Routing Hack ****");
memcpy(&map_addr, &inside_addr, sizeof (map_addr));
}
/*
* Am I running in front of the routing device? Then I cannot
* use the external IP to bind a listen socket to, so force
* the use of my inbound IP for listening.
*/
if ((rtp_direction == DIR_INCOMING) &&
(configuration.outbound_host) &&
(strcmp(configuration.outbound_host, "")!=0)) {
DEBUGC(DBCLASS_PROXY, "proxy_rewrite_invitation_body: "
"in-front-of-NAT-Router");
memcpy(&map_addr, &inside_addr, sizeof (map_addr));
}
sts = rtp_start_fwd(osip_message_get_call_id(mymsg),
client_id,

2
src/rtpproxy_relay.c
View File

@ -219,7 +219,7 @@ static void *rtpproxy_main(void *arg) {
* and hope that next time we pass by it will be ok again.
*/
if (errno == EAGAIN) {
/*&&&& I may want to remove this WARNing */
/* I may want to remove this WARNing */
WARN("read() [fd=%i, %s:%i] would block, but select() "
"claimed to be readable!",
rtp_proxytable[i].rtp_rx_sock,