All of lore.kernel.org
 help / color / mirror / Atom feed
From: Eric Dumazet <eric.dumazet@gmail.com>
To: xiaosuo@gmail.com
Cc: "David S. Miller" <davem@davemloft.net>,
	Stephen Hemminger <shemminger@vyatta.com>,
	Patrick McHardy <kaber@trash.net>,
	Tom Herbert <therbert@google.com>,
	netdev@vger.kernel.org
Subject: Re: [PATCH] ifb: add multi-queue support
Date: Mon, 16 Nov 2009 09:19:21 +0100	[thread overview]
Message-ID: <4B010B09.9020101@gmail.com> (raw)
In-Reply-To: <4B00FFD8.4040508@gmail.com>

Changli Gao a écrit :
> ifb: add multi-queue support.
> 
> Add multi-queue support, through more than one tasklets. One tasklet per
> queue always steers on the same CPU, and if the number of the
> tasklets(queues) is lager than the number of CPUs, more than one
> tasklets will be assigned to the same CPU.
> 
> Signed-off-by: Changli Gao <xiaosuo@gmail.com>
> ----

> 

I must say I fail to see how multiple tasklets per cpu can be of any use.

And :

+	if (skb_queue_len(&pq->rq) == 1)
+		tasklet_schedule_on(&pq->task, num % num_online_cpus());

Is probably not what you want, because :

A) num_online_cpus() can be expensive to compute,
B) and you can have such configs :

Three online cpus, CPU0 and CPU4, CPU5
-> you call tasklet_schedule_on(... , num = {0|1|2})

To avoid packets reorder, if your hash function selects an offline cpu,
you must forward the packet to a particular cpu, regardless of cpu currently running.
(even if real device is not multiqueue, its RX interrupts can be handled by any online cpu)

You maybe need to maintain a mapping table with cpu hotplug notifiers.

  reply	other threads:[~2009-11-16  8:19 UTC|newest]

Thread overview: 62+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-11-16  7:31 [PATCH] ifb: add multi-queue support Changli Gao
2009-11-16  8:19 ` Eric Dumazet [this message]
2009-11-16  8:43   ` Changli Gao
  -- strict thread matches above, loose matches on Subject: below --
2009-11-13  4:42 Changli Gao
2009-11-13  4:46 ` Changli Gao
2009-11-11  9:51 Changli Gao
2009-11-11  9:56 ` Changli Gao
2009-11-11 10:30 ` Eric Dumazet
2009-11-11 10:57   ` Changli Gao
2009-11-11 15:59 ` Patrick McHardy
2009-11-12  3:12   ` Changli Gao
2009-11-12  8:52     ` Jarek Poplawski
2009-11-12  9:32       ` Changli Gao
2009-11-12 15:10     ` Patrick McHardy
2009-11-13  1:28       ` Changli Gao
2009-11-12  9:44 ` Changli Gao
2009-11-12  9:48   ` Changli Gao
2009-11-12 15:11     ` Patrick McHardy
2009-11-13  1:32       ` Changli Gao
2009-11-13  7:18         ` Patrick McHardy
2009-11-12 12:48   ` Eric Dumazet
2009-11-13  1:26     ` Changli Gao
2009-11-13  5:56       ` Eric Dumazet
2009-11-13  6:16         ` Changli Gao
2009-11-13  7:45           ` Jarek Poplawski
2009-11-13  8:54             ` Changli Gao
2009-11-13  9:18               ` Jarek Poplawski
2009-11-13  9:38                 ` Changli Gao
2009-11-13  9:57                   ` Jarek Poplawski
2009-11-13 11:25                     ` Changli Gao
2009-11-13 12:32                       ` Jarek Poplawski
2009-11-13 13:10                       ` Eric Dumazet
2009-11-13 16:15                   ` Stephen Hemminger
2009-11-13 23:28                     ` Changli Gao
2009-11-13 23:32                       ` Stephen Hemminger
2009-11-13 23:42                         ` Changli Gao
2009-11-14 12:53                           ` Eric Dumazet
2009-11-14 13:30                             ` Changli Gao
2009-11-13 13:55               ` Eric Dumazet
2009-11-13  4:37   ` Changli Gao
2009-11-16 16:39     ` Stephen Hemminger
2009-11-17  3:10       ` David Miller
2009-11-17  5:38         ` Changli Gao
2009-11-17  6:02           ` Stephen Hemminger
2009-11-10  8:30 Changli Gao
2009-11-10  9:07 ` Eric Dumazet
2009-11-10  9:43   ` Changli Gao
2009-11-10 10:57     ` Eric Dumazet
2009-11-10 11:14       ` Changli Gao
2009-11-10 11:41         ` Patrick McHardy
2009-11-10 12:14           ` Changli Gao
2009-11-10 12:19             ` Patrick McHardy
2009-11-10 12:37               ` Changli Gao
2009-11-10 12:45                 ` Patrick McHardy
2009-11-10 13:06                   ` Changli Gao
2009-11-10 13:34                     ` Eric Dumazet
2009-11-10 13:49                       ` Changli Gao
2009-11-10 16:45                         ` Stephen Hemminger
2009-11-11  6:30                           ` Changli Gao
2009-11-10 10:29 ` Patrick McHardy
2009-11-10 10:48   ` Changli Gao
2009-11-10 10:55     ` Eric Dumazet

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=4B010B09.9020101@gmail.com \
    --to=eric.dumazet@gmail.com \
    --cc=davem@davemloft.net \
    --cc=kaber@trash.net \
    --cc=netdev@vger.kernel.org \
    --cc=shemminger@vyatta.com \
    --cc=therbert@google.com \
    --cc=xiaosuo@gmail.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 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.