BUGS: ----- * fix "to " bug on FreeBSD 2.2.8 fastroute works =============================================================================== GENERAL: -------- * use fr_tcpstate() with NAT code for increased NAT usage security or even fr_checkstate() - suspect this is not possible. time permitting: * load balancing across interfaces * record buffering for TCP/UDP * modular application proxying available * allow multiple ip addresses in a source route list for ipsend * complete Linux port to implement all the IP Filter features return-rst done, to/dup-to/fastroute remain - ip_forward() problems :-( on hold until rewrite * add a flag to automate src spoofing done * ipfsync() should change IP#'s in current mappings as well as what's in rules. done * document bimap * document NAT rule order processing * add more docs in progress 3.4: XDDD. I agree. Bandwidth Shapping and QoS (Quality of Service, AKA traffic priorization) should be *TOP* in the TO DO list. * irc proxy for dcc * Bandwidth limiting!!! * More examples * More documentation * And did I mention bandwidth limiting??? * Load balancing features added to the NAT code, so that I can have something coming in for 20.20.20.20:80 and it gets shuffled around between internal addresses 10.10.10.1:8000 and 10.10.10.2:8000. or whatever. - done, stage 1 (round robin/split) The one thing that Cisco's PIX has on IPF that I can see is that rewrites the sequence numbers with semi-random ones. I would also love to see a more extensive NAT. It can choose to do rdr and map based on saddr, daddr, sport and dport. (Does the kernel module already have functionality for that and it just needs support in the userland ipnat?) * intrusion detection detection of port scans detection of multiple connection attempts * support for multiple log files i.e. all connections to ftp and telnet logged to a seperate log file * multiple levels of log severity with E-mail notification of intrusion alerts or other high priority errors * poison pill facility after detection of a port scan, start sending back large packets of garbage or other packets to otherwise confuse the intruder (ping of death?) * I ran into your solaris streams stuff and noticed you are playing with mblk's in an unsafe way. You seem to be modifying the underlying datab without checking db_ref. If db_ref is greater than one, you'll need to copy the mblk, - fixed * fix up where manual pages go for Solaris2 IPv6: ----- * NAT is yet not available, either as a null proxy or address translation BSD: * "to " and "to :" are not supported, but "fastroute" is. Solaris: * "to :" is not supported, but "fastroute" is and "to " are.