From mboxrd@z Thu Jan 1 00:00:00 1970 From: Elena Ufimtseva Subject: Re: [PATCH RFC v2 0/2] linux/vnuma: vNUMA PV guest support introduction Date: Wed, 18 Sep 2013 12:20:15 -0400 Message-ID: References: <1379406841-7441-1-git-send-email-ufimtseva@gmail.com> <1379520998.18543.183.camel@Abyss> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <1379520998.18543.183.camel@Abyss> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Dario Faggioli Cc: Li Yechen , George Dunlap , "xen-devel@lists.xen.org" , David Vrabel , Matt Wilson , Boris Ostrovsky List-Id: xen-devel@lists.xenproject.org On Wed, Sep 18, 2013 at 12:16 PM, Dario Faggioli wrote: > On mar, 2013-09-17 at 04:33 -0400, Elena Ufimtseva wrote: >> This patchset introduces vNUMA for PV domU guest. >> > I'm picky and I know it, but I think domU and guest are synonyms... > Having just one of them should be enough. > > Also, the subject line "linux/vnuma: vNUMA...". It is indeed a good > practice to indicate to what component and subsystem the patches applies > to. However, in this case, I don't think havin "linux/" there adds much, > since you'll be sending these mainly in LKML (although, yes, they'll go > on xen-devel too, but I honestly think we can manage). > > Regarding the "/vnuma" part, well, vnuma isn't really a subsystem. > Actually, it does not even exist before this series, so again, I won't > put it there. > > Actually, from a Linux developer/maintainer point of view, these patches > are about Xen, so something like "xen:..." or "x86/xen:" is probably > better > >> Enables PV guest to discover NUMA topology provided by Xen >> and initializes NUMA topology on boot. XENMEM subop hypercall >> is used to retreive information from Xen. >> > The fact that this happens during the regular x86 NUMA initialization > phase (i.e., in x86_numa_init()) is worth mentioning here. > >> Xen provides number >> of NUMA nodes, memory regions (start and end pfn) constructed >> based on e820 domU map, distance table and cpu to node map. i >> xen_numa_init is called to setup NUMA related structures. >> To enable this mechanism, kernel should be compiled as PV guest >> with CONFIG_NUMA=y and Xen should support vNUMA functionality >> (patchset http://lists.xenproject.org/archives/html/xen-devel/2013-09/msg01337.html ). >> > This is fine. Perhaps I'd add something about future plans, which are to > extend this to work for Dom0 too. > Sure, will change these. > Dario > > -- > <> (Raistlin Majere) > ----------------------------------------------------------------- > Dario Faggioli, Ph.D, http://about.me/dario.faggioli > Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK) > -- Elena