pimsm-v1-current.tar.gz - current snapshot pimsm-v1-ss*.tar.gz - snapshots. Out of date: .obsolete/pimd-helmy.tar.gz - original PIM-SM v2 daemon by Ahmed Helmy .obsolete/pimd-cap.tar.gz - downgrade of Helmy's pimd to PIM-SM v1 and .obsolete/pimk_2.1.55.dif.gz - corresponding kernel patch by Carlos Picoto Changes: * accept_register_v1, accept_register_stop_v1, accept_assert_v1 are implemented Now we can be RP. * aging out of RPs by rp-reachability messages is not implemented, but skeleton of accept_rpreach_v1 is OK now, so that sending and redistribution works. * Cisco autorp protocol is implemented. We can opearate as RP-DISC client and send RP-ANNOUNCE, if we are RP. We still are not able to be RP-mapping agent, but it is not necessary. * The first attempt to play with aliases. Unfortunately, Cisco does not understand them, but if LAN is served only by Linuxes, it should work. * PIMv1 - PIMv2 duality. It should work in mixed environment. * Dense mode started, but still does not work. * A lot of big-little endian conflicts are resolved. Apparently, original pimd was developed on big-endian platform, so that it absolutely ignored byte ordering. * PIMv2 is upgraded to current pimsmv2 RFC2362, which is incompatible with older ones. Not tested, it is very unlikely, that it is bug free :-) [990327] * holdtime for PIMv2 correctly formatted * autorp announce threshold and announce filters. [990330] * well, a little bit of PIMv2 debugging. At least, registers are sent and reg-stops are received now. 8) Not so much, but it is the first step. * some strange bug(?) in update_kernel_leaves(). Check it. [990530] * calculate checksums right. Seems, it is standard BSD bug. * when we need to send something to RP (f.e. join/prune for (S,G)RP), it is possible that (*,G) does not exist, so that we should find interface directly. NOTES: * Dense mode is not supported yet. So that, if other routers work in pure sparse mode, you should supply rp for autorp. rp_address 224.0.1.40 rp_address 224.0.1.39 If they use "sparse-dense" mode (preferred setup), you should write rp_address 224.0.1.40 0.0.0.0 rp_address 224.0.1.39 0.0.0.0 to avoid confusing Ciscos by your invalid registers. Note, that autorp will work if and only if multicast route points to the direction of mapping agent. Ideally, you could write rp_address 224.0.1.40 dense rp_address 224.0.1.39 dense but it DOES NOT WORK at the moment. ---------------------------------------------------------------- Please, report bugs and just examples of strange behaviour to me (kuznet@ms2.inr.ac.ru). Alexey Kuznetsov