From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tariq Toukan Subject: Re: [PATCH net-next 9/9] mlx4: add page recycling in receive path Date: Tue, 7 Feb 2017 18:20:12 +0200 Message-ID: <92a6e32b-6f0a-88a6-52d3-6ca352c26b22@gmail.com> References: <20170207030240.31357-1-edumazet@google.com> <20170207030240.31357-10-edumazet@google.com> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit Cc: netdev , Tariq Toukan , Martin KaFai Lau , Willem de Bruijn , Jesper Dangaard Brouer , Brenden Blanco , Alexei Starovoitov , Eric Dumazet To: Eric Dumazet , "David S . Miller" Return-path: Received: from mail-wr0-f196.google.com ([209.85.128.196]:33163 "EHLO mail-wr0-f196.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753894AbdBGQUP (ORCPT ); Tue, 7 Feb 2017 11:20:15 -0500 Received: by mail-wr0-f196.google.com with SMTP id i10so5993566wrb.0 for ; Tue, 07 Feb 2017 08:20:15 -0800 (PST) In-Reply-To: <20170207030240.31357-10-edumazet@google.com> Sender: netdev-owner@vger.kernel.org List-ID: On 07/02/2017 5:02 AM, Eric Dumazet wrote: > Same technique than some Intel drivers, for arches where PAGE_SIZE = 4096 > > In most cases, pages are reused because they were consumed > before we could loop around the RX ring. This might not be true when multiple streams are handled in the same ring. As the process time in stack will be longer. We will test this scenario as well. > > This brings back performance, and is even better, > a single TCP flow reaches 30Gbit on my hosts. > > Signed-off-by: Eric Dumazet > --- > drivers/net/ethernet/mellanox/mlx4/en_rx.c | 238 ++++++++------------------- > drivers/net/ethernet/mellanox/mlx4/mlx4_en.h | 1 - > 2 files changed, 68 insertions(+), 171 deletions(-) > >