behind the same siproxd to have conversation together
UA1 -->--\ /-->--\
siproxd Registrar
UA2 --<--/ \--<--/
- Redone code for evaluation if a received packet
if coming from inbound or outbound interface
- RTP stream are now identified by call_id AND
USERNAME of the contact header. This provides
support for RTP proxying between 2 UAs sitting on the
inbound network. -> Calls between local UAs going via
siproxd should now work.
UA1 -->--\
siproxd
UA2 --<--/
- Rewriting of SUBSCRIBE should now work.
- Removed obsolete prototypes from rtpproxy.h
- If the RTP stream in one direction is found to be
stopped (sendto()) also stop the opposite direction
found by dmalloc memory poisoning
- fix: new branch id hash calculation was not working properly
(dont check return status in osip_xxx_to_str() functions -
the docu does not say anything about them (looks like 0 == success)
but better check the returnet string pointer value -> NULL = failed)
This resulted in a memory leak an imporper calculated MD5 hash.
- RTPPROXY correction: match RTP ports crosswise -
use one single port (and socket) on each side (inbound/
outbound) to send and receive RTP traffic for every
active stream (patch by Christof Meerwald).
a completely statically linked executable
- REGISTER takes honors the expires parameter
of the contact header
- Contact header of REGISTER response must be
rewritten back to the local (true) URL
his work on this). Up to now, only the RTP *Relay*
has been tested (works with KPhone, BudgeTone)
- fix: SIP phones that allocate a random port for
incomming SIP traffic should now work (like BudgeTone)
- fix: some SIP phones do change the RTP port number
during a session (like KPhone during HOLD/unHOLD)
uses threads it probably is a very good idea (some funny things
seen with Linux 2.4.18 kernel & RTP proxy - gethostbyname()
did return a NULL result but has h_errno set to 0 "every fine, thanks")
Future will show it this helps.
- Had overlooked one inet_aton in register.c - replaced to utils_inet_aton
- Fixed a compiler warning in log.c for Solaris. Siproxd now at least
builds on Solaris (tested on: SunOS 5.9 sun4u sparc SUNW,Ultra-60)
* have siproxd compile on Solaris and BSD/OS (more to come)
* ./configure option --with-libosip-prefix
* properly handle getopt_long()/getopt()
- First attempt of iptables support. Routines to add/remove
DNAT entries must still be done.
