From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Michael S. Tsirkin" Subject: Re: [PATCH] rdma: don't make pages writeable if not requiested Date: Thu, 21 Mar 2013 10:51:07 +0200 Message-ID: <20130321085107.GE28328@redhat.com> References: <20130321061838.GA28319@redhat.com> <20130321070357.GD28328@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: Sender: linux-rdma-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Roland Dreier Cc: "Michael R. Hines" , Sean Hefty , Hal Rosenstock , Yishai Hadas , Christoph Lameter , "linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org" , LKML , qemu-devel-qX2TKyscuCcdnm+yROfE0A@public.gmane.org List-Id: linux-rdma@vger.kernel.org On Thu, Mar 21, 2013 at 12:15:33AM -0700, Roland Dreier wrote: > >> I think this change will break the case where userspace tries to > >> register an MR with read-only permission, but intends locally through > >> the CPU to write to the memory. > > > Shouldn't it set LOCAL_WRITE then? > > We're talking about the permissions for the register MR operation, > right? (That's what the kernel RDMA driver code that does > get_user_pages() sees) > > In that case, no, I don't see any reason for LOCAL_WRITE, since the > only RDMA operations that will access this memory are remote reads. What is the meaning of LOCAL_WRITE then? There are no local RDMA writes as far as I can see. > The writing (that triggers COW) is coming from normal process access > triggering a page fault, etc. This is a pretty standard way of using > RDMA... For example, I allocate some memory and register it for RDMA > read (and pass the R_Key to the remote system) with only REMOTE_READ > permission. Then I fill in the memory with the results of some > computation and the remote system does an RDMA read to get those > results. > > - R. OK then what we need is a new flag saying "I really do not intend to write into this memory please do not break COW or do anything else just in case I do". -- MST -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S933374Ab3CUIuu (ORCPT ); Thu, 21 Mar 2013 04:50:50 -0400 Received: from mx1.redhat.com ([209.132.183.28]:43887 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932625Ab3CUIur (ORCPT ); Thu, 21 Mar 2013 04:50:47 -0400 Date: Thu, 21 Mar 2013 10:51:07 +0200 From: "Michael S. Tsirkin" To: Roland Dreier Cc: "Michael R. Hines" , Sean Hefty , Hal Rosenstock , Yishai Hadas , Christoph Lameter , "linux-rdma@vger.kernel.org" , LKML , qemu-devel@nongnu.org Subject: Re: [PATCH] rdma: don't make pages writeable if not requiested Message-ID: <20130321085107.GE28328@redhat.com> References: <20130321061838.GA28319@redhat.com> <20130321070357.GD28328@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, Mar 21, 2013 at 12:15:33AM -0700, Roland Dreier wrote: > >> I think this change will break the case where userspace tries to > >> register an MR with read-only permission, but intends locally through > >> the CPU to write to the memory. > > > Shouldn't it set LOCAL_WRITE then? > > We're talking about the permissions for the register MR operation, > right? (That's what the kernel RDMA driver code that does > get_user_pages() sees) > > In that case, no, I don't see any reason for LOCAL_WRITE, since the > only RDMA operations that will access this memory are remote reads. What is the meaning of LOCAL_WRITE then? There are no local RDMA writes as far as I can see. > The writing (that triggers COW) is coming from normal process access > triggering a page fault, etc. This is a pretty standard way of using > RDMA... For example, I allocate some memory and register it for RDMA > read (and pass the R_Key to the remote system) with only REMOTE_READ > permission. Then I fill in the memory with the results of some > computation and the remote system does an RDMA read to get those > results. > > - R. OK then what we need is a new flag saying "I really do not intend to write into this memory please do not break COW or do anything else just in case I do". -- MST From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([208.118.235.92]:56149) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1UIbCz-0005kC-Ix for qemu-devel@nongnu.org; Thu, 21 Mar 2013 04:50:52 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1UIbCx-0008A1-34 for qemu-devel@nongnu.org; Thu, 21 Mar 2013 04:50:49 -0400 Received: from mx1.redhat.com ([209.132.183.28]:30578) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1UIbCw-00089q-Nt for qemu-devel@nongnu.org; Thu, 21 Mar 2013 04:50:46 -0400 Date: Thu, 21 Mar 2013 10:51:07 +0200 From: "Michael S. Tsirkin" Message-ID: <20130321085107.GE28328@redhat.com> References: <20130321061838.GA28319@redhat.com> <20130321070357.GD28328@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: Subject: Re: [Qemu-devel] [PATCH] rdma: don't make pages writeable if not requiested List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Roland Dreier Cc: qemu-devel@nongnu.org, "linux-rdma@vger.kernel.org" , Yishai Hadas , LKML , "Michael R. Hines" , Hal Rosenstock , Sean Hefty , Christoph Lameter On Thu, Mar 21, 2013 at 12:15:33AM -0700, Roland Dreier wrote: > >> I think this change will break the case where userspace tries to > >> register an MR with read-only permission, but intends locally through > >> the CPU to write to the memory. > > > Shouldn't it set LOCAL_WRITE then? > > We're talking about the permissions for the register MR operation, > right? (That's what the kernel RDMA driver code that does > get_user_pages() sees) > > In that case, no, I don't see any reason for LOCAL_WRITE, since the > only RDMA operations that will access this memory are remote reads. What is the meaning of LOCAL_WRITE then? There are no local RDMA writes as far as I can see. > The writing (that triggers COW) is coming from normal process access > triggering a page fault, etc. This is a pretty standard way of using > RDMA... For example, I allocate some memory and register it for RDMA > read (and pass the R_Key to the remote system) with only REMOTE_READ > permission. Then I fill in the memory with the results of some > computation and the remote system does an RDMA read to get those > results. > > - R. OK then what we need is a new flag saying "I really do not intend to write into this memory please do not break COW or do anything else just in case I do". -- MST