From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Yang, Zhiyong" Subject: Re: [PATCH v3 3/4] net/virtio-user: add support for server mode Date: Fri, 30 Mar 2018 02:08:13 +0000 Message-ID: References: <20180214145330.4679-1-zhiyong.yang@intel.com> <20180321030343.64399-1-zhiyong.yang@intel.com> <20180321030343.64399-4-zhiyong.yang@intel.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Cc: "Wang, Zhihong" , "maxime.coquelin@redhat.com" , "thomas@monjalon.net" , "Wang, Dong1" , "Bie, Tiwei" To: "Tan, Jianfeng" , "dev@dpdk.org" Return-path: Received: from mga11.intel.com (mga11.intel.com [192.55.52.93]) by dpdk.org (Postfix) with ESMTP id BC1B65681 for ; Fri, 30 Mar 2018 04:08:18 +0200 (CEST) In-Reply-To: Content-Language: en-US List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" Hi Jianfeng, > -----Original Message----- > From: Tan, Jianfeng > Sent: Wednesday, March 28, 2018 11:15 PM > To: Yang, Zhiyong ; dev@dpdk.org > Cc: Wang, Zhihong ; > maxime.coquelin@redhat.com; thomas@monjalon.net; Wang, Dong1 > ; Bie, Tiwei > Subject: Re: [PATCH v3 3/4] net/virtio-user: add support for server mode >=20 > Hi Zhiyong, >=20 > Triggered by the community discussion of thread creation in a library > (http://dpdk.org/dev/patchwork/patch/36579/), we can think about if > possible to avoid the pthread creation here. >=20 > Just a quick think, it could be feasible. The key idea is to properly set= up the > LSC interrupt for a server mode virtio-user. > - If the virtio-user device is just probed, we can setup LSC interrupt by > registering the listen fd to interrupt thread. Whenever a vhost-user > connection is coming, we will get a chance to read status, at that time, = we > could accept the connection, and report the status is up. > - And after the connection is setup, we setup LSC interrupt by registerin= g the > connection fd to interrupt thread; if the connection is broken, we also g= et a > change to read status, in which we can switch to the above setting. >=20 > How do you think? >=20 Very nice idea, which looks reasonable. Let me try it.=20 Thanks Zhiyong > Thanks, > Jianfeng >=20