[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <1340743747.10893.370.camel@edumazet-glaptop>
Date: Tue, 26 Jun 2012 22:49:07 +0200
From: Eric Dumazet <eric.dumazet@...il.com>
To: Vijay Subramanian <subramanian.vijay@...il.com>
Cc: netdev@...r.kernel.org, davem@...emloft.net, shemminger@...tta.com,
alexander.h.duyck@...el.com
Subject: Re: [PATCH net-next] udp: Add socket early demux support
On Tue, 2012-06-26 at 12:43 -0700, Vijay Subramanian wrote:
> Based on the recent TCP socket early demux code, this patch provides similar
> support for UDP.
>
> Signed-off-by: Vijay Subramanian <subramanian.vijay@...il.com>
> ---
> This has been tested on x86 with UDP iperf flows and seemed to work. If this is
> accepted, I plan to submit one more patch moving common code from TCP and UDP
> early demux code into common helper functions.
> Thanks in advance for feedback.
Hmm... I cant see how it can work.
Have you tested a router can still route DNS packets if you also have a
DNS server on it, listening on 0.0.0.0:53 ?
Most UDP applications are using unconnected sockets.
(In fact few programmers are aware UDP sockets can be connected)
Try to bench how this is going to work with random IP sources, instead
of UDP iperf flows using a single IP ?
And what about multicast ?
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists