New revisions created in hipl-dev@xxxxxxxxxxxxxxxxxxx/hipl--beet--2.6 2005-11-04 GMT Miika Komu <miika@xxxxxx> patch-57 Summary: Fixed the hipconf add map problem. Revision: hipl--beet--2.6--patch-57 I think the XFRM synchronization is not needed when calling "hipconf add map". On the userspace branch, the synchronization did not result any SA, but rather updated the kernel copy of the hadb. The same applied for the rest of the hip_xfrm_update calls, so I removed them. hip_xfrm_add was not referenced either so I removed it too. Next problem: global SP is not created properly when hipd is started even though hip_setup_sp_prefix_pair() is called from main(). I discovered this by: tools/hipconf add hi default tools/hipconf add map <HIT> <IPv6> nc6 <HIT> 12345 So, (rougly) all of the tasks are the following in cronological order: * Fix global SP HIT prefix setting in hip_setup_sp_prefix_pair() * Triggering of base exchange * I have already registered the daemon to listen for XFRM_NETLINK messages in hipd.c:main() * Implement XFRMGRP_ACQUIRE in netdev.c:hip_netdev_event() * Sending of HIP packets * See preoutput.c:hip_csum_send() * You may want to borrow some functions from the Boeing HIP code for handling the IP header * Receiving of HIP packets * See hipd.c:main() FD_ISSET(hip_raw_sock, &read_fdset) * You may want to borrow some functions from the Boeing HIP code for handling the IP header * See that SA creation works properly during base exchange * Note that we have to create now a "local" SP for each HIT pair, i.e., the global SP HIT prefix is not enough due to the BEET design * See that SA creation works properly for handovers References: http://openhip.sourceforge.net/ http://www.linuxjournal.com/article/7356 modified files: hipd/beet.c hipd/hadb.c hipd/hip.h hipd/hipd.c hipd/input.c hipd/netdev.c hipd/preoutput.c hipd/update.c regards, "tla mail-new-revisions"