On 10 October 2014 12:19, Amit Shah <amit.s...@redhat.com> wrote: > On (Thu) 09 Oct 2014 [14:04:53], Peter Maydell wrote: >> What I would like to see in this patch is a comment giving >> much clearer definition of the semantics of the guest_writable >> call: for instance, is it always called when the guest is >> writable, or is it only guaranteed to be called if the >> QEMU backend has previously tried to do virtio_serial_write >> and got back a return code indicating an incomplete write? > > Right now it's called always; if it's necessary to only call after an > incomplete / failed write, we just need to implement that (and also > add that state to the migration data). > > I have no preference; just depends on the users of the API.
Called-always makes sense; I'd just like to see the semantics documented so you don't have to read the whole implementation of virtio to figure out what they are :-) -- PMM