On Wed, Jun 24, 2015 at 07:57:45AM +0200, Michael S. Tsirkin wrote: > On Wed, Jun 24, 2015 at 02:46:30PM +0900, Tetsuya Mukawa wrote: > > On 2015/06/23 18:41, Michael S. Tsirkin wrote: > > > On Tue, Jun 23, 2015 at 05:31:06PM +0900, Tetsuya Mukawa wrote: > > >> On 2015/06/22 17:14, Michael S. Tsirkin wrote: > > >>> On Mon, Jun 22, 2015 at 12:50:43PM +0900, Tetsuya Mukawa wrote: > > >>>> Hi guys, > > >>>> > > >>>> Here are patches to add feature to start QEMU without vhost-user > > >>>> backend. > > >>>> Currently, if we want to use vhost-user backend, the backend must > > >>>> start before > > >>>> QEMU. > > >>> Aren't you looking for something like xinetd? > > >> It will help for restricting starting order, but not help for > > >> reconnection of vhost-user backend. > > > At this point, I suggest that we always connect at vm start. > > > With that, you can add an option to reset the VM > > > on backend disconnect. > > > So > > > - detect backend disconnect > > > - reset and stop (notifying management) > > > - reconnect or detect backend reconnect > > > - proceed with boot > > > > > > As I tried to explain below, getting the full functionality > > > will require guest driver changes. They aren't hard to get > > > done, patches welcome. > > > > > > > Could you please let me know your thinking about using > > DEVICE_NEEDS_RESET for vhost-user reconnection? > > If it's works, I will try to submit it. > > DEVICE_NEEDS_RESET is hard to handle correctly in guest: > you need to reconfigure a bunch of state, > so far no one wrote the necessary support.
But yes, if you write the guest code, I think it can work. We probably want a feature bit so host can know guest can handle reconnections, but that's a minor detail, it would be easy to add afterwards. -- MST
