From: Hannes Frederic Sowa Subject: Re: [PATCH net-next v2] unix: add read side socket memory accounting for dgram sockets Date: Wed, 26 Feb 2014 21:21:54 +0100 Message-ID: <20140226202154.GA9626@order.stressinduktion.org> References: <20140217130353.GA22833@order.stressinduktion.org> <20140219.144927.1558282069067885839.davem@davemloft.net> <20140219200811.GE1179@order.stressinduktion.org> <20140226.145559.1040609936875578841.davem@davemloft.net> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Cc: netdev@vger.kernel.org, yannick@koehler.name, eric.dumazet@gmail.com, xiyou.wangcong@gmail.com, dan@mindstab.net To: David Miller Return-path: Received: from order.stressinduktion.org ([87.106.68.36]:55161 "EHLO order.stressinduktion.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750968AbaBZUVz (ORCPT ); Wed, 26 Feb 2014 15:21:55 -0500 Content-Disposition: inline In-Reply-To: <20140226.145559.1040609936875578841.davem@davemloft.net> Sender: netdev-owner@vger.kernel.org List-ID: On Wed, Feb 26, 2014 at 02:55:59PM -0500, David Miller wrote: > From: Hannes Frederic Sowa > Date: Wed, 19 Feb 2014 21:08:11 +0100 > > > On Wed, Feb 19, 2014 at 02:49:27PM -0500, David Miller wrote: > >> From: Hannes Frederic Sowa > >> Date: Mon, 17 Feb 2014 14:03:53 +0100 > >> > >> > We still allocate dgram packets with sock_alloc_send_pskb, which now > >> > does normally not block if the socket has too many packets in flight. > >> > >> It seems like it does to me, it does sock_wait_for_wmem(), right? > > > > We still do the sock_wait_for_wmem in sock_alloc_send_pskb, correct, > > but we rarly block there, maybe in highly concurrent cases with big > > payloads where one process got interrupted to account the memory to the > > receiving socket. > > > >> Or are you trying to say that usually this is not the point at which > >> we block, but rather it's when we check the peer's receive queue? > > > > Exactly. > > So my only major objection is that we're now going to incur two atomics > on every single datagram UNIX packet sent between two peers. Ok, I see. Daniel Borkmann pointed me to hackbench off-list and I will do some before/after benchmarks. > I think you can safely put the 'other' sock pointer into the control > block. Because when we disassociate from a peer any pending packets > in flight to him will be consumed/dropped, so there can't be any > lingering references, right? I am afraid lingering socket references will happen if the sockets are used in unconnected mode. Greetings, Hannes