From mboxrd@z Thu Jan 1 00:00:00 1970 From: Keir Fraser Subject: Re: [PATCH v2] add locking around certain calls to map_pages_to_xen() Date: Fri, 12 Jul 2013 15:01:45 +0100 Message-ID: References: <51E023BB02000078000E47E1@nat28.tlf.novell.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <51E023BB02000078000E47E1@nat28.tlf.novell.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Jan Beulich Cc: Andrew Cooper , xen-devel List-Id: xen-devel@lists.xenproject.org On 12/07/2013 14:41, "Jan Beulich" wrote: >> Is it unsafe to just stick a lock around the guts of map_pages_to_xen(), or >> at least the parts that add new page tables? > > I'm not certain about the safety of this, but clearly two CPUs > changing entirely different parts of the address space don't need > to lock out one another, so I rather view adding a global lock here > as being (potentially) harmful in terms of performance (and hence > the thought of locking at page table entry granularity instead). Ah, I see. Well, locking only on changes to page-directory entries wouldn't be too bad, even if it were a single global lock? That would be a rare occurrence. It's reasonable to assume that callers will not conflict on the page-aligned regions they modify, so this would suffice? -- Keir