Re: [Freeipmi-devel] Re: GE rsh problem

[Anand Babu]

,----[ Ian Zimmerman <address@hidden> ]
| ab> Please run gdb (or at least strace) to figure out who (xinetd or
| ab> in.rshd) opens these new free ports per connection and why. Look
| ab> for open ports both on server and client.
| 
| Albert> I believe both in.rshd (through rresvport()) and rsh (through
| Albert> rcmd(), which subsequently calls rresvport()) use sockets with
| Albert> privileged ports.
| 
| The mystery part is solved: rsh protocol uses _two_ connections, the
| additional one (for standard error) from the server to the client,
| with a source port allocated by rresvport.
`----
Good trace. Thanks Al and Ian.

,----[ Ian Zimmerman <address@hidden> ]
| Now, why isn't the blocking daemon working to make the two ports
| EADDRINUSE?
`----
Yes, that is what the daemon is trying to achieve indirectly. By
opening the reserved ports in advance, Kernel will automagically
return EADDRINUSE to the rest of the world. This was the simplest fix
I can think about in user space without requiring a patch to the
kernel or rsh. Any better ideas are most welcome.


,----[ Ian Zimmerman <address@hidden> ]
| I think the ball is in Mr. Manian's court.
`----
Yes, We are waiting to hear from Anand Manian. Please ask Pradhap to
give him a call.

-- 
Anand Babu
Free as in Freedom <www.gnu.org>