> In 2.0.x we had some horribly gross (yet effective) hacks in the
> packet queueing layer that would go right into the devices queues and
> take back + requeue packets if the device queue overflowed, then try
> again later. It was gross, but it worked and solved most of these
> serial link problems.
I did not understand...
The only difference from 2.0 is that we clone skbs and, hence, have no
device lock, but it did not help to PREVENT drops in any case
because affects only retransmissions. If 2.0 really behaves better,
it requires explanation.
BTW it is easy to make static clamping ssthresh and snd_cwnd,
based on routing tables. Probably, it will really help folks with slow
links.
Alexey
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/