Re: [patch] ns83820 0.17

pjd@fred001.dynip.com
Tue, 12 Mar 2002 14:52:59 -0500 (EST)


David Miller wrote:
>
> I said we don't need NAPI for just bandwidth streams, you mention
> routing which is specifically the case I mention that NAPI is good for
> (high packet rates).

In particular, if you have a small number of high-speed streams the
TCP window mechanism will protect against receive livelock. (actually
a medium number of streams would still be protected - it's not until
the total offered window size in packets exceeds the input packet
queue size that you would become vulnerable to livelock)

Routing, on the other hand, can be driven into a state where you spend
all your CPU processing receive interrupts, and no CPU actually
forwarding the packets, for a net throughput approaching zero.

Peter Desnoyers
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/