Netdev List
 help / color / mirror / Atom feed
From: Eric Dumazet <eric.dumazet@gmail.com>
To: "Ilpo Järvinen" <ilpo.jarvinen@helsinki.fi>
Cc: David Miller <davem@davemloft.net>,
	rick.jones2@hp.com, Netdev <netdev@vger.kernel.org>,
	therbert@google.com, ncardwell@google.com, maze@google.com,
	Yuchung Cheng <ycheng@google.com>
Subject: [RFC] allow skb->head to point/alias to first skb frag
Date: Thu, 26 Apr 2012 10:10:54 +0200	[thread overview]
Message-ID: <1335427854.2775.15.camel@edumazet-glaptop> (raw)
In-Reply-To: <1335257803.5205.133.camel@edumazet-glaptop>

On Tue, 2012-04-24 at 10:56 +0200, Eric Dumazet wrote:

> Really I have many doubts about GRO today.

Particularly if a NIC driver provides linear skbs :
Resulting gro skb is a list of skbs, no memory savings, and many cache
line misses when copying to userland.

Maybe we could have a new kind of skb head allocation/setup, pointing to
a frag of 2048 bytes instead of a kmalloc() blob.

Right now, a fragged skb used 3 blocks of memory :

1) struct sk_buff
2) a bloc of 512 or 1024 or 2048 bytes of memory (skb->head)
3) a frag of 2048 (or PAGE_SIZE/2 or PAGE_SIZE)

While a linear skb has :

1) struct sk_buff
2) a bloc of 512 or 1024 or 2048 bytes of memory (skb->head) from
kmalloc()


Idea would have :

1) struct sk_buff
2) skb->head points to frag (aliasing, no memory allocation)
3) frag of 2048 (or PAGE_SIZE/2 or PAGE_SIZE)

Or the reverse (no frag so that skb is considered as linea), but special
code to 'allow' this skb head be considered as a frag when needed
(splice() code, or GRO merge, or TCP coalescing)


That would make GRO (and TCP coalescing) much more efficient, since the
resulting aggregated skb would be :
1) struct sk_buff
2) skb->head points to 1st frag (aliasing, no memory allocation)
3) array of [1..16] frags

  reply	other threads:[~2012-04-26  8:11 UTC|newest]

Thread overview: 37+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-04-23  9:38 [PATCH 2/2 net-next] tcp: sk_add_backlog() is too agressive for TCP Eric Dumazet
2012-04-23 17:14 ` Rick Jones
2012-04-23 17:23   ` Eric Dumazet
2012-04-23 20:01     ` David Miller
2012-04-23 20:37       ` Eric Dumazet
2012-04-23 20:57         ` Rick Jones
2012-04-23 21:30           ` Eric Dumazet
2012-04-23 21:51             ` Rick Jones
2012-04-23 21:56               ` Rick Jones
2012-04-23 22:05               ` Eric Dumazet
2012-04-23 22:16                 ` Rick Jones
2012-04-24 15:25                   ` Christoph Lameter
2012-04-23 21:01         ` David Miller
2012-04-23 21:38           ` Eric Dumazet
2012-04-24  2:20         ` Eric Dumazet
2012-04-24  2:27           ` David Miller
2012-04-24  8:01           ` Ilpo Järvinen
2012-04-24  8:10             ` David Miller
2012-04-24  8:21               ` Ilpo Järvinen
2012-04-24  8:25                 ` David Miller
2012-04-24  8:40                   ` Ilpo Järvinen
2012-04-24  8:48                     ` David Miller
2012-04-24 10:32                       ` Ilpo Järvinen
2012-04-24  8:56                     ` Eric Dumazet
2012-04-26  8:10                       ` Eric Dumazet [this message]
2012-04-26  8:36                         ` [RFC] allow skb->head to point/alias to first skb frag David Miller
2012-04-26  9:10                           ` Eric Dumazet
2012-04-26  9:18                             ` David Miller
2012-04-26  9:22                               ` Eric Dumazet
2012-04-26 10:09                                 ` Maciej Żenczykowski
2012-04-26 12:32                                   ` Eric Dumazet
2012-04-26 13:50                                     ` Eric Dumazet
2012-04-26 20:12                                       ` David Miller
2012-04-26 20:18                                         ` Eric Dumazet
2012-04-24  8:49             ` [PATCH 2/2 net-next] tcp: sk_add_backlog() is too agressive for TCP Eric Dumazet
2012-04-24  8:44         ` David Laight
2012-04-24  8:53           ` 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=1335427854.2775.15.camel@edumazet-glaptop \
    --to=eric.dumazet@gmail.com \
    --cc=davem@davemloft.net \
    --cc=ilpo.jarvinen@helsinki.fi \
    --cc=maze@google.com \
    --cc=ncardwell@google.com \
    --cc=netdev@vger.kernel.org \
    --cc=rick.jones2@hp.com \
    --cc=therbert@google.com \
    --cc=ycheng@google.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