Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752817Ab2HNWrb (ORCPT ); Tue, 14 Aug 2012 18:47:31 -0400 Received: from mx1.redhat.com ([209.132.183.28]:41474 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751863Ab2HNWra (ORCPT ); Tue, 14 Aug 2012 18:47:30 -0400 Date: Wed, 15 Aug 2012 01:48:16 +0300 From: "Michael S. Tsirkin" To: Rafael Aquini Cc: linux-mm@kvack.org, linux-kernel@vger.kernel.org, virtualization@lists.linux-foundation.org, Rusty Russell , Rik van Riel , Mel Gorman , Andi Kleen , Andrew Morton , Konrad Rzeszutek Wilk , Minchan Kim Subject: Re: [PATCH v7 2/4] virtio_balloon: introduce migration primitives to balloon pages Message-ID: <20120814224816.GB29180@redhat.com> References: <20120813084123.GF14081@redhat.com> <20120814182244.GB13338@t510.redhat.com> <20120814195139.GA28870@redhat.com> <20120814195916.GC28870@redhat.com> <20120814200830.GD22133@t510.redhat.com> <20120814202401.GB28990@redhat.com> <20120814202949.GF22133@t510.redhat.com> <20120814204906.GD28990@redhat.com> <20120814213412.GG22133@t510.redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20120814213412.GG22133@t510.redhat.com> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 3126 Lines: 77 On Tue, Aug 14, 2012 at 06:34:13PM -0300, Rafael Aquini wrote: > On Tue, Aug 14, 2012 at 11:49:06PM +0300, Michael S. Tsirkin wrote: > > On Tue, Aug 14, 2012 at 05:29:50PM -0300, Rafael Aquini wrote: > > > On Tue, Aug 14, 2012 at 11:24:01PM +0300, Michael S. Tsirkin wrote: > > > > On Tue, Aug 14, 2012 at 05:08:31PM -0300, Rafael Aquini wrote: > > > > > On Tue, Aug 14, 2012 at 10:59:16PM +0300, Michael S. Tsirkin wrote: > > > > > > > > > What if there is more than one balloon device? > > > > > > > > > > > > > > > > Is it possible to load this driver twice, or are you foreseeing a future case > > > > > > > > where this driver will be able to manage several distinct memory balloons for > > > > > > > > the same guest? > > > > > > > > > > > > > > > > > > > > > > Second. > > > > > > > It is easy to create several balloons they are just > > > > > > > pci devices. > > > > > > > > > > > > > > > > > > > > > > > > and it might not be too important to make it work but > > > > > > at least would be nice not to have a crash in this > > > > > > setup. > > > > > > > > > > > Fair enough. For now, as I believe it's safe to assume we are only inflating one > > > > > balloon per guest, I'd like to propose this as a future enhancement. Sounds > > > > > good? > > > > > > > > > > > > > Since guest crashes when it's not the case, no it doesn't, sorry :(. > > > > > > > Ok, but right now this driver only takes care of 1 balloon per guest, > > > > It does? Are you sure? There is no global state as far as I can see. So > > I can create 2 devices and driver will happily create two instances, > > each one can be inflated/deflated independently. > > > > > so how > > > could this approach crash it? > > > > Add device. inflate. Add another device. inflate. deflate. unplug. > > Now you have pointer to freed memory and when mm touches > > page from first device, you ge use after free. > > > > > Your point is a good thing to be on a to-do list for future enhancements, but > > > it's not a dealbreaker for the present balloon driver implementation, IMHO. > > > > > > > Yes it looks like a dealbreaker to me. > > Sorry. You're right, I'm wrong. > > I'll get back to the scracthpad to overcome this constraint. I believe the way > this patch was at its v4 revision (wrt this particular case) could possibly > address this concern of yours. Almost. We still have a global balloon_mapping. The only reason for it to exist seems solely to detect balloon mappings, so it can just be replaced by a flag in the mapping, or in mapping ops, or elsewhere. Also, please add APIs to mm so we can avoid doing internal mm stuff like INIT_RADIX_TREE(&balloon_mapping->page_tree, GFP_ATOMIC | __GFP_NOWARN); in the driver. It should be alloc_address_mapping(&virtio_balloon_aops); free_address_mapping Make page->mapping use rcu, and sync rcu in free_address_mapping. -- MST -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/