From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mga11.intel.com (mga11.intel.com [192.55.52.93]) by dpdk.org (Postfix) with ESMTP id 9B930DE6 for ; Wed, 27 May 2015 11:00:08 +0200 (CEST) Received: from fmsmga003.fm.intel.com ([10.253.24.29]) by fmsmga102.fm.intel.com with ESMTP; 27 May 2015 02:00:07 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.13,503,1427785200"; d="scan'208";a="498801439" Received: from shwdeisgchi017.ccr.corp.intel.com (HELO [10.239.66.47]) ([10.239.66.47]) by FMSMGA003.fm.intel.com with ESMTP; 27 May 2015 02:00:06 -0700 Message-ID: <55658797.5000809@intel.com> Date: Wed, 27 May 2015 17:00:07 +0800 From: "Liang, Cunming" User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) Gecko/20100101 Thunderbird/31.6.0 MIME-Version: 1.0 To: Stephen Hemminger References: <1430804386-28949-1-git-send-email-cunming.liang@intel.com> <1432198563-16334-1-git-send-email-cunming.liang@intel.com> <1432198563-16334-6-git-send-email-cunming.liang@intel.com> <20150522132100.3f0829e9@urahara> In-Reply-To: <20150522132100.3f0829e9@urahara> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Cc: dev@dpdk.org, liang-min.wang@intel.com Subject: Re: [dpdk-dev] [PATCH v8 05/11] eal/linux: add interrupt vectors handling on VFIO X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: patches and discussions about DPDK List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 27 May 2015 09:00:09 -0000 On 5/23/2015 4:21 AM, Stephen Hemminger wrote: > On Thu, 21 May 2015 16:55:57 +0800 > Cunming Liang wrote: > >> This patch does below: >> - Create VFIO eventfds for each interrupt vector (move to next) >> - Assign per interrupt vector's eventfd to VFIO by ioctl >> >> Signed-off-by: Danny Zhou >> Signed-off-by: Cunming Liang > One non-trivial performance related change here would be to set SMP > affinity of the receive IRQ to the CPU that is handling that receive queue. > Not sure the full API to do this, but ideally you should not have the > receive interrupt occurring on one CPU then having to cause scheduler > to wakeup receive thread on another CPU. > That's a good point. The previous thought was to configure irq affinity by script from outside. I haven't found some API to do that well, the well known way is by sysfs.