From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Morton Subject: Re: [PATCH v4 2/3] virtio_balloon: introduce migration primitives to balloon pages Date: Wed, 18 Jul 2012 15:49:08 -0700 Message-ID: <20120718154908.14704344.akpm@linux-foundation.org> References: <050e06731e0489867ed804387509e36d072507ec.1342485774.git.aquini@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <050e06731e0489867ed804387509e36d072507ec.1342485774.git.aquini@redhat.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: virtualization-bounces@lists.linux-foundation.org Errors-To: virtualization-bounces@lists.linux-foundation.org To: Rafael Aquini Cc: Rik van Riel , Konrad Rzeszutek Wilk , "Michael S. Tsirkin" , linux-kernel@vger.kernel.org, virtualization@lists.linux-foundation.org, linux-mm@kvack.org, Andi Kleen , Minchan Kim , Rafael Aquini List-Id: virtualization@lists.linuxfoundation.org On Tue, 17 Jul 2012 13:50:42 -0300 Rafael Aquini wrote: > Besides making balloon pages movable at allocation time and introducing > the necessary primitives to perform balloon page migration/compaction, > this patch also introduces the following locking scheme to provide the > proper synchronization and protection for struct virtio_balloon elements > against concurrent accesses due to parallel operations introduced by > memory compaction / page migration. > - balloon_lock (mutex) : synchronizes the access demand to elements of > struct virtio_balloon and its queue operations; > - pages_lock (spinlock): special protection to balloon pages list against > concurrent list handling operations; > > ... > > + balloon_mapping->a_ops = &virtio_balloon_aops; > + balloon_mapping->backing_dev_info = (void *)vb; hoo boy. We're making page->mapping->backing_dev_info point at a struct which does not have type `struct backing_dev_info'. And then we are exposing that page to core MM functions. So we're hoping that core MM will never walk down page->mapping->backing_dev_info and explode. That's nasty, hacky and fragile.