From: Eric Dumazet <eric.dumazet@gmail.com>
To: David Miller <davem@davemloft.net>
Cc: netdev@vger.kernel.org
Subject: Re: small RPS cache for fragments?
Date: Tue, 17 May 2011 22:14:48 +0200 [thread overview]
Message-ID: <1305663288.2691.2.camel@edumazet-laptop> (raw)
In-Reply-To: <20110517.143342.1566027350038182221.davem@davemloft.net>
Le mardi 17 mai 2011 à 14:33 -0400, David Miller a écrit :
> It seems to me that we can solve the UDP fragmentation problem for
> flow steering very simply by creating a (saddr/daddr/IPID) entry in a
> table that maps to the corresponding RPS flow entry.
>
> When we see the initial frag with the UDP header, we create the
> saddr/daddr/IPID mapping, and we tear it down when we hit the
> saddr/daddr/IPID mapping and the packet has the IP_MF bit clear.
>
> We only inspect the saddr/daddr/IPID cache when iph->frag_off is
> non-zero.
>
> It's best effort and should work quite well.
>
> Even a one-behind cache, per-NAPI instance, would do a lot better than
> what happens at the moment. Especially since the IP fragments mostly
> arrive as one packet train.
> --
OK but do we have workloads actually needing this optimization at all ?
(IP defrag hits a read_lock(&ip4_frags.lock)), so maybe steer all frags
on a given cpu ?)
next prev parent reply other threads:[~2011-05-17 20:14 UTC|newest]
Thread overview: 40+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-05-17 18:33 small RPS cache for fragments? David Miller
2011-05-17 20:02 ` Tom Herbert
2011-05-17 20:17 ` Rick Jones
2011-05-17 20:41 ` Rick Jones
2011-05-17 20:49 ` David Miller
2011-05-17 21:00 ` Eric Dumazet
2011-05-17 21:10 ` David Miller
2011-05-17 21:13 ` Rick Jones
2011-05-17 21:13 ` Ben Hutchings
2011-05-17 21:26 ` David Miller
2011-05-17 21:40 ` Rick Jones
2011-05-17 21:27 ` Eric Dumazet
2011-05-17 21:11 ` Rick Jones
2011-05-17 21:11 ` Ben Hutchings
2011-05-17 21:27 ` Tom Herbert
2011-05-17 21:28 ` David Miller
2011-05-17 23:59 ` Changli Gao
2011-05-18 6:37 ` David Miller
2011-05-17 20:14 ` Eric Dumazet [this message]
2011-05-17 20:47 ` David Miller
2011-05-17 21:44 ` Andi Kleen
2011-05-17 21:52 ` Eric Dumazet
2011-05-17 22:03 ` Andi Kleen
2011-05-17 21:44 ` David Miller
2011-05-17 21:48 ` Andi Kleen
2011-05-17 21:50 ` David Miller
2011-05-17 22:06 ` Andi Kleen
2011-05-17 22:42 ` Rick Jones
2011-05-24 20:01 ` David Miller
2011-05-24 21:38 ` Rick Jones
2011-06-04 20:29 ` David Miller
2011-06-06 17:08 ` Rick Jones
2011-06-06 17:15 ` Eric Dumazet
2011-06-06 18:06 ` Rick Jones
2011-06-06 19:23 ` David Miller
2011-06-06 19:22 ` David Miller
2011-06-06 20:05 ` Rick Jones
2011-06-06 21:06 ` Jay Vosburgh
2011-06-06 21:40 ` David Miller
2011-06-06 22:49 ` Chris Friesen
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1305663288.2691.2.camel@edumazet-laptop \
--to=eric.dumazet@gmail.com \
--cc=davem@davemloft.net \
--cc=netdev@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.