From: "David S. Miller" <davem@davemloft.net>
To: hadi@cyberus.ca
Cc: shemminger@osdl.org, rhee@eos.ncsu.edu, jheffner@psc.edu,
Yee-Ting.Li@nuim.ie, baruch@ev-en.org, netdev@oss.sgi.com
Subject: Re: netif_rx packet dumping
Date: Thu, 3 Mar 2005 13:32:37 -0800 [thread overview]
Message-ID: <20050303133237.5d64578f.davem@davemloft.net> (raw)
In-Reply-To: <1109885065.1098.285.camel@jzny.localdomain>
On 03 Mar 2005 16:24:25 -0500
jamal <hadi@cyberus.ca> wrote:
> Ok, this does sound more reasonable. Out of curiosity, are packets being
> dropped at the socket queue? Why is "dump till empty" behaviour screwing
> over TCP.
Because it does the same thing tail-drop in routers do.
It makes everything back off a lot and go into slow start.
If we'd just drop 1 packet per flow or something like that
(so it could be fixed with a quick fast retransmit), TCP
would avoid regressing into slow start.
You say "use a NAPI driver", but netif_rx() _IS_ a NAPI driver.
process_backlog() adheres to quotas and every other stabilizing
effect NAPI drivers use, the only missing part is the RX interrupt
disabling.
We should eliminate the max backlog thing completely. There is
no need for it.
next prev parent reply other threads:[~2005-03-03 21:32 UTC|newest]
Thread overview: 51+ messages / expand[flat|nested] mbox.gz Atom feed top
2005-03-03 20:38 netif_rx packet dumping Stephen Hemminger
2005-03-03 20:55 ` David S. Miller
2005-03-03 21:01 ` Stephen Hemminger
2005-03-03 21:18 ` jamal
2005-03-03 21:21 ` Stephen Hemminger
2005-03-03 21:24 ` jamal
2005-03-03 21:32 ` David S. Miller [this message]
2005-03-03 21:54 ` Stephen Hemminger
2005-03-03 22:02 ` John Heffner
2005-03-03 22:26 ` jamal
2005-03-03 23:16 ` Stephen Hemminger
2005-03-03 23:40 ` jamal
2005-03-03 23:48 ` Baruch Even
2005-03-04 3:45 ` jamal
2005-03-04 8:47 ` Baruch Even
2005-03-07 13:55 ` jamal
2005-03-08 15:56 ` Baruch Even
2005-03-08 22:02 ` jamal
2005-03-22 21:55 ` cliff white
2005-03-03 23:48 ` John Heffner
2005-03-04 1:42 ` Lennert Buytenhek
2005-03-04 3:10 ` John Heffner
2005-03-04 3:31 ` Lennert Buytenhek
2005-03-04 19:52 ` Edgar E Iglesias
2005-03-04 19:54 ` Stephen Hemminger
2005-03-04 21:41 ` Edgar E Iglesias
2005-03-04 19:49 ` Jason Lunz
2005-03-03 22:01 ` jamal
2005-03-03 21:26 ` Baruch Even
2005-03-03 21:36 ` David S. Miller
2005-03-03 21:44 ` Baruch Even
2005-03-03 21:54 ` Andi Kleen
2005-03-03 22:04 ` David S. Miller
2005-03-03 21:57 ` David S. Miller
2005-03-03 22:14 ` Baruch Even
2005-03-08 15:42 ` Baruch Even
2005-03-08 17:00 ` Andi Kleen
2005-03-08 18:01 ` Baruch Even
2005-03-08 18:09 ` David S. Miller
2005-03-08 18:18 ` Andi Kleen
2005-03-08 18:37 ` Thomas Graf
2005-03-08 18:51 ` Arnaldo Carvalho de Melo
2005-03-08 22:16 ` Andi Kleen
2005-03-08 18:27 ` Ben Greear
2005-03-09 23:57 ` Thomas Graf
2005-03-10 0:03 ` Stephen Hemminger
2005-03-10 8:33 ` Andi Kleen
2005-03-10 14:08 ` Thomas Graf
2005-03-31 16:33 ` Baruch Even
2005-03-03 22:03 ` jamal
2005-03-03 22:31 ` Baruch Even
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=20050303133237.5d64578f.davem@davemloft.net \
--to=davem@davemloft.net \
--cc=Yee-Ting.Li@nuim.ie \
--cc=baruch@ev-en.org \
--cc=hadi@cyberus.ca \
--cc=jheffner@psc.edu \
--cc=netdev@oss.sgi.com \
--cc=rhee@eos.ncsu.edu \
--cc=shemminger@osdl.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).