From mboxrd@z Thu Jan 1 00:00:00 1970 From: Rusty Russell Subject: Re: [PATCH 1/4] net: skb_orphan on dev_hard_start_xmit Date: Mon, 1 Jun 2009 21:57:28 +0930 Message-ID: <200906012157.29465.rusty@rustcorp.com.au> References: <200905292344.56814.rusty@rustcorp.com.au> <4A1FFB04.30305@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Return-path: In-Reply-To: <4A1FFB04.30305@gmail.com> Content-Disposition: inline List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: virtualization-bounces@lists.linux-foundation.org Errors-To: virtualization-bounces@lists.linux-foundation.org To: Eric Dumazet Cc: libertas-dev@lists.infradead.org, Dan Williams , netdev@vger.kernel.org, virtualization@lists.linux-foundation.org, Roland Dreier , Divy Le Ray , Pavel Emelianov List-Id: virtualization@lists.linuxfoundation.org On Sat, 30 May 2009 12:41:00 am Eric Dumazet wrote: > Rusty Russell a =E9crit : > > 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 an= d an = 100MBit e1000e). Two reasons come to mind: firstly, only the hardware queu= e 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.