mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Robert Olsson <Robert.Olsson@data.slu.se>
To: <mingo@elte.hu>
Cc: jamal <hadi@cyberus.ca>, <linux-kernel@vger.kernel.org>,
	Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>,
	Robert Olsson <Robert.Olsson@data.slu.se>,
	Benjamin LaHaise <bcrl@redhat.com>, <netdev@oss.sgi.com>,
	Linus Torvalds <torvalds@transmeta.com>,
	Alan Cox <alan@lxorguk.ukuu.org.uk>
Subject: Re: [announce] [patch] limiting IRQ load, irq-rewrite-2.4.11-B5
Date: Wed, 3 Oct 2001 23:08:07 +0200	[thread overview]
Message-ID: <15291.32311.499838.886628@robur.slu.se> (raw)
In-Reply-To: <Pine.LNX.4.33.0110031828060.8633-100000@localhost.localdomain>
In-Reply-To: <Pine.GSO.4.30.0110031138150.4833-100000@shell.cyberus.ca> <Pine.LNX.4.33.0110031828060.8633-100000@localhost.localdomain>


Ingo Molnar writes:

 > (i did not criticize the list_add/list_del in any way, it's obviously
 > correct to cycle the polled devices. I highlited that code only to show
 > that the current patch as-is polls too agressively for generic server
 > load.)

 Yes I think we need some data here... 

 > can you really make it 100% successful for rx? Ie. do you only ever call
 > the ->poll() function if there is a new packet waiting? How do you know
 > with a 100% probability that someone on the network just sent a new packet
 > waiting? (without receiving an interrupt to begin with that is.)

 Well we need RX-interrupts not to spin away the CPU or exhaust the the PCI-
 bus. The NAPI scheme is simple, turn off RX-interrupts when the first packet
 comes and have the kernel to pull packets from the RX-ring. 

 I tried have pure polling... it easy do just have your driver return
 "not_done" all the time. Not a good idea. :-) Maybe as sofirq test.
 
 If the device has more packets to deliver than is "allowed" we put this
 back on list and the polling process can give the next device its share
 and so on. So we handle screaming network devices and packet flooding 
 nicely a deliver a decent performance even under those circumstances.

 As you seen from some code fragments we have played with some mechanisms 
 to delay the transition from polling to irq-enable. I think I accepted
 a not_done polls for jiffies in some of the tests. Agree other variants 
 are possible and hopefully better.

 SMP is another area, robustness and performance of course but in case
 of SMP we also have to deal with packet reordering which is something
 we really like to minimize. Even here I think the NAPI polling scheme 
 is interesting. During consecutive polls the device is bound to the same 
 CPU and no packet reordering should occur. 

 And from data we have now we can see packet load is even distributed
 among different CPU's and should follow the smp_affinity.

 Cheers.

						--ro


 

  parent reply	other threads:[~2001-10-03 21:06 UTC|newest]

Thread overview: 151+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2001-10-02  0:41 jamal
2001-10-02  1:04 ` Benjamin LaHaise
2001-10-02  1:54   ` jamal
2001-10-02  5:13     ` Benjamin LaHaise
2001-10-02  5:55       ` Ben Greear
2001-10-03  9:22         ` Ingo Molnar
2001-10-03 14:06           ` David Brownell
2001-10-02 12:10       ` jamal
2001-10-02 22:00         ` jamal
2001-10-03  8:34           ` Ingo Molnar
2001-10-03  9:29             ` Helge Hafting
2001-10-03 12:49             ` jamal
2001-10-03 14:51               ` Ingo Molnar
2001-10-03 15:14                 ` jamal
2001-10-03 17:28                   ` Ingo Molnar
2001-10-04  0:53                     ` jamal
2001-10-04  6:28                       ` Ingo Molnar
2001-10-04 11:34                         ` jamal
2001-10-04 17:40                           ` Andreas Dilger
2001-10-04 18:33                             ` jamal
2001-10-04  6:50                       ` Ben Greear
2001-10-04  6:52                         ` Ingo Molnar
2001-10-04 11:50                           ` jamal
2001-10-04  6:55                         ` Jeff Garzik
2001-10-04  6:56                           ` Ingo Molnar
2001-10-04 21:28                 ` Alex Bligh - linux-kernel
2001-10-04 21:49                   ` Benjamin LaHaise
2001-10-04 23:20                     ` Alex Bligh - linux-kernel
2001-10-04 23:26                       ` Benjamin LaHaise
2001-10-04 23:47                       ` Robert Love
2001-10-04 23:51                         ` Linus Torvalds
2001-10-05  0:00                           ` Ben Greear
2001-10-05  0:18                             ` Davide Libenzi
2001-10-05  2:01                             ` jamal
2001-10-04 22:01                   ` Simon Kirby
2001-10-04 23:25                     ` Alex Bligh - linux-kernel
2001-10-04 23:34                       ` Simon Kirby
2001-10-04 22:10                   ` Alan Cox
2001-10-04 23:28                     ` Alex Bligh - linux-kernel
2001-10-05 15:22                   ` Robert Olsson
2001-10-03  9:38           ` Ingo Molnar
2001-10-03 13:03             ` jamal
2001-10-03 13:25               ` jamal
2001-10-03 15:28               ` Ingo Molnar
2001-10-03 15:56                 ` jamal
2001-10-03 16:51                   ` Ingo Molnar
2001-10-04  0:46                     ` jamal
2001-10-08  0:31                     ` Andrea Arcangeli
2001-10-08  4:58                       ` Bernd Eckenfels
2001-10-08 15:00                       ` Alan Cox
2001-10-08 15:03                         ` Jeff Garzik
2001-10-08 15:12                           ` Alan Cox
2001-10-08 15:09                             ` jamal
2001-10-08 15:22                               ` Alan Cox
2001-10-08 15:20                                 ` jamal
2001-10-08 15:35                                   ` Alan Cox
2001-10-08 15:57                                     ` jamal
2001-10-08 16:11                                       ` Alan Cox
2001-10-08 16:11                                         ` jamal
2001-10-10 16:26                                         ` Pavel Machek
2001-10-10 16:25                                     ` Pavel Machek
2001-10-08 15:24                             ` Andrea Arcangeli
2001-10-08 15:35                               ` Alan Cox
2001-10-08 15:19                         ` Andrea Arcangeli
2001-10-08 15:10                       ` bill davidsen
2001-10-03 21:08                   ` Robert Olsson [this message]
2001-10-03 22:22                     ` Andreas Dilger
2001-10-04 17:32                       ` Davide Libenzi
2001-10-05 14:52                     ` Robert Olsson
2001-10-05 18:48                       ` Andreas Dilger
2001-10-05 19:07                         ` Davide Libenzi
2001-10-05 19:17                         ` kuznet
2001-10-08 13:58                           ` jamal
2001-10-08 17:42                           ` Robert Olsson
2001-10-08 17:39                             ` jamal
2001-10-07  6:11                         ` Robert Olsson
2001-10-03 16:53                 ` kuznet
2001-10-03 17:06                   ` Ingo Molnar
2001-10-04  0:44                     ` jamal
2001-10-04  6:35                       ` Ingo Molnar
2001-10-04 11:41                         ` jamal
2001-10-05 16:42                         ` kuznet
2001-10-04 13:05                       ` Robert Olsson
2001-10-03 19:03                   ` Benjamin LaHaise
2001-10-04  1:10                     ` jamal
2001-10-04  1:30                       ` Benjamin LaHaise
2001-10-03 22:31                         ` Rob Landley
2001-10-04  1:39                         ` jamal
2001-10-03 15:42               ` Ben Greear
2001-10-03 15:58                 ` jamal
2001-10-03 16:09                   ` Ben Greear
2001-10-03 16:14                     ` Ingo Molnar
2001-10-03 16:20                     ` Jeff Garzik
2001-10-03 16:33                 ` Linus Torvalds
2001-10-03 17:25                   ` Ingo Molnar
2001-10-03 18:11                     ` Linus Torvalds
2001-10-03 20:41                       ` Jeremy Hansen
2001-10-03 20:02                   ` Simon Kirby
2001-10-04  1:04                     ` jamal
2001-10-04  6:47                       ` Ben Greear
2001-10-04  7:41                         ` Henning P. Schmiedehausen
2001-10-04 16:09                           ` Ben Greear
2001-10-04 17:32                             ` Henning P. Schmiedehausen
2001-10-04 18:03                               ` Ben Greear
2001-10-04 18:30                           ` Christopher E. Brown
2001-10-04 11:47                         ` jamal
2001-10-04 15:56                           ` Ben Greear
2001-10-04 18:23                             ` jamal
2001-10-04  6:50                       ` Ingo Molnar
2001-10-04 11:49                         ` jamal
2001-10-04  8:45                       ` Simon Kirby
2001-10-04 11:54                         ` jamal
2001-10-04 15:03                           ` Tim Hockin
2001-10-04 18:55                           ` Ion Badulescu
2001-10-04 19:00                             ` jamal
2001-10-04 21:16                               ` Ion Badulescu
2001-10-04  4:12                   ` bill davidsen
2001-10-04 18:16                   ` Alan Cox
2001-10-03 13:38             ` Robert Olsson
2001-10-04 21:22               ` Alex Bligh - linux-kernel
2001-10-05 14:32               ` Robert Olsson
2001-10-03  8:38         ` Ingo Molnar
2001-10-04  3:50           ` bill davidsen
2001-10-02 17:03       ` Robert Olsson
2001-10-02 17:37         ` jamal
2001-10-02 19:46         ` Andreas Dilger
  -- strict thread matches above, loose matches on Subject: below --
2001-10-08 14:45 jamal
2001-10-09  0:36 ` Scott Laird
2001-10-09  3:17   ` jamal
2001-10-09  4:04 ` Werner Almesberger
2001-10-04  8:25 Magnus Redin
2001-10-04 11:39 ` Trever L. Adams
     [not found] <200110031811.f93IBoN10026@penguin.transmeta.com>
2001-10-03 18:23 ` Ingo Molnar
2001-10-04  9:19   ` BALBIR SINGH
2001-10-04  9:22     ` Ingo Molnar
2001-10-04  9:49       ` BALBIR SINGH
2001-10-04 10:25         ` Ingo Molnar
2001-10-07 20:37           ` Andrea Arcangeli
2001-10-03 14:15 Manfred Spraul
2001-10-03 15:09 ` jamal
2001-10-03 18:37   ` Davide Libenzi
2001-10-01 22:16 Ingo Molnar
2001-10-01 22:26 ` Tim Hockin
2001-10-01 22:50   ` Ingo Molnar
2001-10-01 22:36 ` Andreas Dilger
2001-10-01 22:50 ` Ben Greear
2001-10-02 14:30   ` Alan Cox
2001-10-02 20:51     ` Ingo Molnar
2001-10-01 23:03 ` Linus Torvalds
2001-10-02  6:50 ` Marcus Sundberg
2001-10-03  8:47   ` Ingo Molnar

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=15291.32311.499838.886628@robur.slu.se \
    --to=robert.olsson@data.slu.se \
    --cc=alan@lxorguk.ukuu.org.uk \
    --cc=bcrl@redhat.com \
    --cc=hadi@cyberus.ca \
    --cc=kuznet@ms2.inr.ac.ru \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=netdev@oss.sgi.com \
    --cc=torvalds@transmeta.com \
    /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

all inboxes | Powered by JetHome®