From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([208.118.235.92]:43570) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1UF6M7-0003If-Hf for qemu-devel@nongnu.org; Mon, 11 Mar 2013 13:17:54 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1UF6M2-00074c-FR for qemu-devel@nongnu.org; Mon, 11 Mar 2013 13:17:47 -0400 Received: from e7.ny.us.ibm.com ([32.97.182.137]:60076) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1UF6M2-00074Q-Bu for qemu-devel@nongnu.org; Mon, 11 Mar 2013 13:17:42 -0400 Received: from /spool/local by e7.ny.us.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Mon, 11 Mar 2013 13:17:41 -0400 Received: from d01relay02.pok.ibm.com (d01relay02.pok.ibm.com [9.56.227.234]) by d01dlp01.pok.ibm.com (Postfix) with ESMTP id B265538C8096 for ; Mon, 11 Mar 2013 13:17:28 -0400 (EDT) Received: from d01av02.pok.ibm.com (d01av02.pok.ibm.com [9.56.224.216]) by d01relay02.pok.ibm.com (8.13.8/8.13.8/NCO v10.0) with ESMTP id r2BHHRYq265940 for ; Mon, 11 Mar 2013 13:17:27 -0400 Received: from d01av02.pok.ibm.com (loopback [127.0.0.1]) by d01av02.pok.ibm.com (8.14.4/8.13.1/NCO v10.0 AVout) with ESMTP id r2BHHNkr017541 for ; Mon, 11 Mar 2013 14:17:23 -0300 Message-ID: <513E119F.1010203@linux.vnet.ibm.com> Date: Mon, 11 Mar 2013 13:17:19 -0400 From: "Michael R. Hines" MIME-Version: 1.0 References: <1362976414-21396-1-git-send-email-mrhines@us.ibm.com> <1362976414-21396-4-git-send-email-mrhines@us.ibm.com> <20130311115140.GC23359@redhat.com> <513E0555.1060205@linux.vnet.ibm.com> <20130311170515.GB28930@redhat.com> In-Reply-To: <20130311170515.GB28930@redhat.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [RFC PATCH RDMA support v3: 03/10] documentation of RDMA protocol in docs/rdma.txt List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: "Michael S. Tsirkin" Cc: aliguori@us.ibm.com, michael.r.hines.mrhines@linux.vnet.ibm.com, qemu-devel@nongnu.org, owasserm@redhat.com, abali@us.ibm.com, mrhines@us.ibm.com, gokul@us.ibm.com, pbonzini@redhat.com On 03/11/2013 01:05 PM, Michael S. Tsirkin wrote: > Well that's exactly the question. As far as I remember the RDMA memory > model, you need to know a key and address to execute RDMA writes. > Remote memory also needs to be locked, so you need some mechanism to > lock chunks of memory, do RDMA write and unlock when done. Yes, memory is registered before the write occurs - that's is all taken care of in rdma.c (patch #04/10) Same answer for the SEND messages: memory for each send must be registered before you send. The pinning (mlock()) is already handled by libibverbs (by calling the function ibv_reg_mr()) - standard infiniband protocol initialization. - Michael