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 256D91B28F for ; Thu, 25 Jan 2018 20:15:56 +0100 (CET) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from fmsmga008.fm.intel.com ([10.253.24.58]) by fmsmga102.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 25 Jan 2018 11:15:56 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.46,413,1511856000"; d="scan'208";a="13165471" Received: from tanjianf-mobl.ccr.corp.intel.com (HELO [10.255.27.46]) ([10.255.27.46]) by fmsmga008.fm.intel.com with ESMTP; 25 Jan 2018 11:15:54 -0800 To: dev@dpdk.org References: <1512067450-59203-1-git-send-email-jianfeng.tan@intel.com> <1516907684-118975-1-git-send-email-jianfeng.tan@intel.com> Cc: anatoly.burakov@intel.com, bruce.richardson@intel.com, konstantin.ananyev@intel.com, thomas@monjalon.net From: "Tan, Jianfeng" Message-ID: <811eed1f-e6ef-0738-0f2d-e4289a9e12ab@intel.com> Date: Fri, 26 Jan 2018 03:15:53 +0800 User-Agent: Mozilla/5.0 (Windows NT 6.3; WOW64; rv:45.0) Gecko/20100101 Thunderbird/45.8.0 MIME-Version: 1.0 In-Reply-To: <1516907684-118975-1-git-send-email-jianfeng.tan@intel.com> Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit Subject: Re: [dpdk-dev] [PATCH v4 0/2] generic channel for multi-process communication X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 25 Jan 2018 19:15:57 -0000 Apology, please ignore this version which is not correct. Will send out a new version. On 1/26/2018 3:14 AM, Jianfeng Tan wrote: > v3->v4: > - Drop the patch 3 on vfio communication (postponed). > - Change names from rte_eal_mp_* -> rte_mp_* as suggested by Thomas. > - Add nb_sent and nb_received in struct rte_mp_reply. > - Standardize the return val of sendmsg, request, reply: 0 on sucess, > (-1) on failure. > - If we found an peer error when we send msg in primary, we try to > remove the secondary socket; as there is no sync mechanism there > (cannot do flock like regular file for socket file), we use a more > complex socket name (with tsc in it). > - Some other small changes. > > v2->v3: > - Add pre-check for each APIs. > - Remove the limitation of 8 secondary processes by: discard original > register/unregister mechanism of secondary process, instead, primary > discoveries secondary processes by looking up the folder for regex match. > - Previous implementation use two sockets for msg and request, this version > just uses one socket. And receive all kinds of messages in mp thread. > > v1->v2: (Address comments from Anatoly and Konstantin) > - Use datagram unix socket to supersede stream unix socket + epoll. > - Change the secondary add/del mechanism as now we use connection-less channel. > - Add mp_mutex_action to sync action register/unregister/reference. > - Limit max length of action name to 64B. > - New APIs for synchronous communication: rte_eal_mp_request/rte_eal_mp_reply. > - Formalize the errno handle. > - Some other small issues. > > This patchset adds a generic channel for multi-process (primary/secondary) > communication. > > Patch 1: addess the purpose and howto; > Patch 2: add a syncrhonous way for the requests which need a immediate response. > > Jianfeng Tan (2): > eal: add synchronous multi-process communication > vfio: use the generic multi-process channel > > doc/guides/rel_notes/release_18_02.rst | 2 + > lib/librte_eal/common/eal_common_proc.c | 254 +++++++++++++-- > lib/librte_eal/common/include/rte_eal.h | 58 +++- > lib/librte_eal/linuxapp/eal/eal.c | 14 +- > lib/librte_eal/linuxapp/eal/eal_vfio.c | 169 ++++------ > lib/librte_eal/linuxapp/eal/eal_vfio.h | 15 +- > lib/librte_eal/linuxapp/eal/eal_vfio_mp_sync.c | 410 ++++--------------------- > lib/librte_eal/rte_eal_version.map | 2 + > 8 files changed, 429 insertions(+), 495 deletions(-) >