I used your script (with IPs modified for my setup), removed the “server” line at the end, and used it to set up a path without a default server. I was not able to reproduce the segfault. That setup worked for me just fine, which means I can’t help you debug this directly, but also means that tsocks is supposed to handle this use case. It is likely that our setups still differ significantly (I’m using SOCKS 5, for example) that I’m not triggering the code path that has the bug in it.

If you know the C language, you can try to instruct MacPorts to compile with the -g flag. Then, you can run tsocks under GDB, which will give you the exact backtrace of which function failed at what line of source code. At that point, we can try to see if we can fix that bug ourselves. I’m very curious as to why it doesn’t seem to work for you when it works on my setup — there must be a relatively simple fix.

So using the config below, I can ‘tsocks ssh user@150.60.198.12‘ without any problem only if a default server is specified correctly. The path stanza just seems to be ignored. And if I leave out the default server entry all together, tsocks seg faults.

local = 192.168.1.0/255.255.255.0
local = 129.39.109.0/255.255.255.0
path {
reaches = 150.60.198.0/255.255.255.0
server = 129.39.109.202
server_type = 4
server_port = 1080
}
server = 129.39.109.202

Is your end host’s DNS name resolvable only through the “gateway” server, or should it be globally resolvable? If it’s globally resolvable, it should try to use your usual ISP DNS server (or whatever DNS you have configured in Network preference pane). If it’s a private hostname that only the gateway knows, then you’ll probably have to find some way to get DNS resolution working with tsocks.

OSX 10.5
tsocks from port

tsocks.conf:
local = 127.0.0.1/255.255.255.255
server = 127.0.0.1
server_port = 1080
server_type = 5
tordns_enable = false

ssh -D 1080 user@server

I’m actually trying to use this with git-svn so I type something like:

git svn rebase

It chugs away (debug output seems to provide nothing too useful, lots of “Call to close(n)” then it says “Host not found”. Similarly if I say tsocks ping remote-server same thing. However, if I ssh user@server I can ping the server that I’m trying to reach so tsocks should be able to get there. Any ideas?

I do need a bit more context, but I’ll make some assumptions here. If I understand your context correctly, you’re trying to SSH through the gateway server G to a second, target host H, using SOCKS server S. If you’re using ssh -D, I’m also assuming that you’re trying to use SSH as a SOCKS server (meaning that your SOCKS server S actually sits locally on your localhost 127.0.0.1, you’re using the network through G to go to H, and that you can reach G without using any proxies). If these assumptions are not true, please let me know.

Here’s a checklist of configuration-related things (pardon me if some of these are obvious — just trying to cover the context):
- when you ran ssh -D, you connected to G and the SOCKS server is up. You said Firefox worked — this should mean that you used Firefox’s SOCKS5 feature to connect to your local SOCKS server at 127.0.0.1, and appear to other hosts on the Web as G)
- tsocks.conf: server=127.0.0.1
- tsocks.conf: port= (whatever number you set after ssh -D)
- tsocks.conf: local=127.0.0.1/255.255.255.255
- that you’re using tsocks when making the second SSH connection – tsocks ssh H.

If all of these are right, you might try something simple, like telnet. If you are able to tsocks telnet your way out, then the problem might lie in SSH. Also, make sure that you do have DNS resolution, either via your ordinary DNS server, or your SOCKS server.

If none of these help, you might try asking someone who works on tsocks or (more likely) a tsocks-spinoff project. Sorry I can’t be of more help.