RFA: patch to remote.c for larger download packet support (part 1)

J.T. Conklin jtc@redback.com
Tue Oct 5 15:53:00 GMT 1999


>>>>> "Frank" == Frank Ch Eigler <fche@cygnus.com> writes:
>> How much faster do you expect this to be?  [...] Is it just the
>> packet latency caused by a stop-and-wait protocol?  
>> Assuming that the delay is in the transport layer itself, [...]

Frank> Yes, it's the latency that's a killer, for some transports, on
Frank> some hosts.  My main focus is a cygwin host and a TCP-connected
Frank> target.  This change, plus a friend in generic_load(), gives 1+
Frank> order of magnitude improvement in throughput.

Ouch!  

Fortunately memory writes/downloads are infrequent operations in real
world environments, but this type of latency must make debugging in
general fairly painful. Inserting and removing multiple breakpoints
while stepping through code is one case that would be particularly
effected.

This would argue that we look at replacing the remote protocol that we
look at windowed framing schemes so we can have multiple commands out-
standing.

        --jtc

-- 
J.T. Conklin
RedBack Networks


More information about the Gdb-patches mailing list