From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: [RFC 0/4] Using a generic bus_type for virtio Date: Sun, 08 Jul 2007 12:45:34 +0300 Message-ID: <4690B23E.8030009@qumranet.com> References: <20070706124200.988637662@arndb.de> <1183860930.6005.212.camel@localhost.localdomain> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <1183860930.6005.212.camel@localhost.localdomain> 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: Rusty Russell Cc: arnd@arndb.de, virtualization@lists.linux-foundation.org List-Id: virtualization@lists.linuxfoundation.org Rusty Russell wrote: > On Fri, 2007-07-06 at 14:42 +0200, arnd@arndb.de wrote: > >> This is a subject that came up in the virtio BOF session >> at OLS. I decided to go forward and implement something >> that I like, based on the latest virtio proposal at the >> time, which was draft III. >> >> It's not a drop-in replacement, because it's missing a >> host implementation. I first started my own, which is >> not done yet, but wanted to do one for lguest and one >> for emulated PCI next. It's also entirely untested. >> > > Hi Arnd, > > I think it will come down to how neat PCI<->virtio is. Can we push > further towards PCI without screwing non-PCI? eg. can we use > pci_device_id? struct pci_driver? (Might be pushing it, but should > probably be considered: it'd be neat if some platforms could #define > virtio_driver_register pci_driver_register). > This is liable to cause breakage when something changes. And if I weren't such a sweet person I'd say it's also horribly ugly. > Standardizing how to pack the info for each device into the config > space would be especially useful. Our drivers are going to get more > featureful, and we're going to need a versioning/compatibility scheme > too. > > Basically, I'd like to see someone start with work from the PCI side, > then make sure non-PCI isn't overly burdened. > We can certainly make non-bus specific code, like the feature bitmaps, shared. I do agree that it makes sense to start from the PCI side as virtbus doesn't have any constraints. -- error compiling committee.c: too many arguments to function