A RPC patch for glibc

[H.J. Lu]

> 
> >This patch tries to fix it. When there is a timeout in RPC, it checks
> >if there is any network interface up. If not, it returns immediately.
> 
> Ugh.  Is there not some way we can check for failure to transmit the RPC 
> request instead?  If no network is up we should get "network is unreachable" 
> or some similar error return, and the code ought to notice that no request was 
> actually sent and not hang around waiting for a reply.
> 

The problem is with UDP and the loopback address. The kernel will
return ok for sendto () for the loopback address and UDP even if
the loopback interface is down.

-- 
H.J. Lu (hjl@gnu.org)