Hello everyone. It's been a while since I have posted here....truthfully I was tired of giving progress reports wherein I said "ssh server coming soon" :(....there's good news and bad news.....the bad news - "ssh server still coming soon"....the good it's a lot stronger, lot more robust and hopefully it will come in real soon :).....and it's now multithreaded too :)
By this I mean that a single instance can now handle multiple connections from ssh clients, therefore the eBrainPool client will in theory be able to start a SSH server and serve requests to different clients connecting in and X forward applications.
Unfortunately there are a few bugs that still need squashing. The bugs are most probably in my code however the libssh guys are kind enough to help me debug and even look into their own code.
You can read about some of the issues we've been looking into at the following libssh mail archives.
http://www.libssh.org/archive/libssh/2011-06/0000003.html
...I think I have this mostly sorted and an unitialized buffer and reading right to the edge of buffer size (possibly allowing for an overrun) were the culprits here.
http://www.libssh.org/archive/libssh/2011-06/0000005.html
..this is what we are currently looking at.
The good thing is the problem is consistently reproducible. If I launch two applications i.e. make two ssh client connections to my server and have these X forwarded,say xeyes and xclock, and then I try to keep moving one of the windows around vigorously, sooner or later one of the windows (or both) loses the ability to refresh itself.
The links above give more debugging info....any ideas would be appreciated...and I shall update you on the progress...till then keep reading and keep encouraging us pls :)