From: Rusty Russell <rusty@rustcorp•com.au>
To: Eric Dumazet <eric.dumazet@gmail•com>
Cc: libertas-dev@lists•infradead.org, Dan Williams <dcbw@redhat•com>,
netdev@vger•kernel.org,
virtualization@lists•linux-foundation.org,
Roland Dreier <rolandd@cisco•com>, Divy Le Ray <divy@chelsio•com>,
Pavel Emelianov <xemul@openvz•org>
Subject: Re: [PATCH 1/4] net: skb_orphan on dev_hard_start_xmit
Date: Mon, 1 Jun 2009 21:57:28 +0930 [thread overview]
Message-ID: <200906012157.29465.rusty@rustcorp.com.au> (raw)
In-Reply-To: <4A1FFB04.30305@gmail.com>
On Sat, 30 May 2009 12:41:00 am Eric Dumazet wrote:
> Rusty Russell a écrit :
> > DaveM points out that there are advantages to doing it generally (it's
> > more likely to be on same CPU than after xmit), and I couldn't find
> > any new starvation issues in simple benchmarking here.
>
> If really no starvations are possible at all, I really wonder why some
> guys added memory accounting to UDP flows. Maybe they dont run "simple
> benchmarks" but real apps ? :)
Well, without any accounting at all you could use quite a lot of memory as
there are many places packets can be queued.
> For TCP, I agree your patch is a huge benefit, since its paced by remote
> ACKS and window control
I doubt that. There'll be some cache friendliness, but I'm not sure it'll be
measurable, let alone "huge". It's the win to drivers which don't have a
timely and batching tx free mechanism which I aim for.
> , but an UDP sender will likely be able to saturate
> a link.
I couldn't see any difference in saturation here (with default scheduler and an
100MBit e1000e). Two reasons come to mind: firstly, only the hardware queue is
unregulated: the tx queue is still accounted. And when you add scheduling to
the mix, I can't in practice cause starvation of other senders.
Hope that clarifies,
Rusty.
next prev parent reply other threads:[~2009-06-01 12:27 UTC|newest]
Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-05-29 14:14 [PATCH 1/4] net: skb_orphan on dev_hard_start_xmit Rusty Russell
2009-05-29 15:11 ` Eric Dumazet
2009-06-01 12:27 ` Rusty Russell [this message]
2009-06-03 21:02 ` Eric Dumazet
2009-06-04 3:54 ` Rusty Russell
2009-06-04 4:00 ` David Miller
2009-06-04 4:54 ` Eric Dumazet
2009-06-04 4:56 ` David Miller
2009-06-04 9:18 ` [PATCH] net: No more expensive sock_hold()/sock_put() on each tx Eric Dumazet
2009-06-04 9:26 ` David Miller
2009-06-10 8:17 ` David Miller
2009-06-10 8:30 ` Eric Dumazet
2009-06-11 9:56 ` David Miller
2009-06-01 19:47 ` [PATCH 1/4] net: skb_orphan on dev_hard_start_xmit Patrick Ohly
2009-06-02 7:25 ` David Miller
2009-06-02 14:08 ` Rusty Russell
2009-06-03 0:14 ` David Miller
2009-07-03 7:55 ` Herbert Xu
2009-07-04 3:02 ` David Miller
2009-07-04 3:08 ` Herbert Xu
2009-07-04 3:13 ` David Miller
2009-07-04 7:42 ` Herbert Xu
2009-07-04 9:09 ` Herbert Xu
2009-07-05 3:26 ` Herbert Xu
2009-07-05 3:34 ` Herbert Xu
2009-08-18 1:47 ` David Miller
2009-08-19 3:19 ` Herbert Xu
2009-08-19 3:34 ` David Miller
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=200906012157.29465.rusty@rustcorp.com.au \
--to=rusty@rustcorp$(echo .)com.au \
--cc=dcbw@redhat$(echo .)com \
--cc=divy@chelsio$(echo .)com \
--cc=eric.dumazet@gmail$(echo .)com \
--cc=libertas-dev@lists$(echo .)infradead.org \
--cc=netdev@vger$(echo .)kernel.org \
--cc=rolandd@cisco$(echo .)com \
--cc=virtualization@lists$(echo .)linux-foundation.org \
--cc=xemul@openvz$(echo .)org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
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