mån 2012-02-27 klockan 21:17 -0700 skrev Steven Dake: > I have found with my tracing patch I previously posted that the cause of > the problem of the heavy leak was related to using the following api: > > libssh2_keepalive_config
Gah, keepalive again.. > If the code passes 1 in the want_reply parameter, the ssh server (fedora > 16 opensshd) responds with a REQUEST_FAILURE (82) for every > keepalive_send the code calls. I am not sure where the error is there, It's the expected response. keep-alive is done using an unhandled MSG_GLOBAL_REQUEST message. > but I'd expect a REQUEST_FAILURE would result in no packet allocation in > packet add. I haven't provided a patch as I'm not sure exactly what the > solution should be. Errors also need processing, but no one is eating the keepalive error response. Regards Henrik _______________________________________________ libssh2-devel http://cool.haxx.se/cgi-bin/mailman/listinfo/libssh2-devel
