On Tue, Apr 24, 2012 at 05:10:23PM +0530, Amit Shah wrote: > > Sure. So fix it for single port now and for multiport in a > > separate patch. Alon's patch is a single line and it made sense, yours > > differs in that you have added more code to implement a bug. > > Disagree; it was not obvious in that patch; a single patch did more > thing than one, commit log didn't note it (it was a side-effect), and > the code was not easy to read. > > With the patch I will propose, there won't be any differentiation for > the multiport and not multiport case. And for virtio-serial, > multiport is the more important use-case (i.e. all recent guests have > multiport support).
OK, so let's see the complete fix. > > > > BTW guest_connected is not cleared on reset either - a bug too? > > > > > > Yes, I think we'll just have to scan the list of things that we want > > > zero'ed at start. > > > > Not at start, at reset. > > Similar case. > > Amit