diff options
author | Othmar Gsenger <otti@anytun.org> | 2008-04-12 11:38:42 +0000 |
---|---|---|
committer | Othmar Gsenger <otti@anytun.org> | 2008-04-12 11:38:42 +0000 |
commit | fffd213c8cba2135afda493d797c41c10354770e (patch) | |
tree | bb5eea1b12871d8c3fed0e687d83be3e504d11b2 /src/openvpn/PORTS | |
parent | svn cleanup (diff) |
big svn cleanup
Diffstat (limited to 'src/openvpn/PORTS')
-rw-r--r-- | src/openvpn/PORTS | 94 |
1 files changed, 94 insertions, 0 deletions
diff --git a/src/openvpn/PORTS b/src/openvpn/PORTS new file mode 100644 index 0000000..f0967a7 --- /dev/null +++ b/src/openvpn/PORTS @@ -0,0 +1,94 @@ +OpenVPN +Copyright (C) 2002-2005 OpenVPN Solutions LLC <info@openvpn.net> + + OpenVPN has been written to try to avoid features + that are not standardized well across different + OSes, so porting OpenVPN itself will probably be + straightforward if a tun or tap driver already exists. + + Where special OS features are used, they are usually + bracketed with #ifdef HAVE_SOME_FUNCTION. + +PLATFORM STATUS: + + * Linux 2.2+ (supported) + * Solaris (supported) + * OpenBSD 3.0 (supported but pthreads are broken) + * Max OS X Darwin + * FreeBSD + * NetBSD + * Windows + * 64 bit platforms -- I have heard reports that + OpenVPN runs on Alpha Linux and FreeBSD. + * ARM -- I have heard of at least one case + where OpenVPN was successfully built and + run on the ARM architecture. + +PORTING NOTES: + + * Make sure that OpenSSL will build on your + platform. + * Make sure that a tun or tap virtual device + driver exists for your platform. See + http://vtun.sourceforge.net/tun/ for examples + of tun and tap drivers that have been written + for Linux, Solaris, and FreeBSD. + * Make sure you have autoconf 2.50+ and + automake 1.6+. + * Edit configure.ac, adding platform specific + config code, and a TARGET_YOUROS define. + * Add platform-specific includes to syshead.h. + * Add an #ifdef TARGET_YOUROS to the do_ifconfig() + function in tun.c to generate a correct "ifconfig" + command for your platform. Note that OpenVPN + determines the ifconfig path at ./configure time. + * Add an ifconfig_order() variant for your OS so + openvpn knows whether to call ifconfig before + or after tun/tap dev open. + * Add an #ifdef TARGET_YOUROS block in tun.c and define + the open_tun, close_tun, read_tun, and write_tun + functions. If your tun/tap virtual device is + sufficiently generic, you may be able to use the + default case. + * Add appropriate code to route.c to handle + the route command on your platform. This + is necessary for the --route option to + work correctly. + * After you successfully build OpenVPN, run + the loopback tests as described in INSTALL. + * For the next test, confirm that the UDP socket + functionality is working independently of the + tun device, by doing something like: + ./openvpn --remote localhost --verb 9 --ping 1 --dev null + * Now try with --remote [a real host] + * Now try with a real tun/tap device, you will + need to figure out the appropriate ifconfig + command to use once openvpn has opened the tun/tap + device. + * Once you have simple tests working on the tun device, + try more complex tests such as using TLS mode. + * Stress test the link by doing ping -f across it. + * Make sure that packet fragmenting is happening + correctly by doing a ping -s 2000 or higher. + * Ensure that OpenVPN on your platform will talk + to OpenVPN on other platforms such as Linux. + Some tun/tap driver implementations will prepend + unnecessary stuff onto the datagram that must be + disabled with an explicit ioctl call if cross-platform + compatibility is to be preserved. You can see some + examples of this in tun.c. + * If your system supports pthreads, try building + with ./configure --enable-pthread and do a stress + test in TLS mode. + * Try the ultimate stress test which is --gremlin + --reneg-sec 10 in TLS mode (preferably with pthreads + enabled), then do a flood ping across the tunnel + (ping -f remote-endpoint) in both directions and let + it run overnight. --gremlin will induce massive + corruption and packet loss, but you win if you + wake up the next morning and both peers are still + running and occasionally even succeeding in their + attempted once-per-10-seconds TLS handshake. + * When it's working, submit your patch to + <openvpn-devel@lists.sourceforge.net> + and rejoice :) |