Re: [PATCH v2 16/21] virtio-net: Use replay_schedule_bh_event for bhs that affect machine state

[Nicholas Piggin]

On Fri Aug 16, 2024 at 12:28 AM AEST, Michael S. Tsirkin wrote:
> On Thu, Aug 15, 2024 at 05:12:32PM +1000, Nicholas Piggin wrote:
> > Could be a good idea. Although I'm not sure what to do with
> > all types, maybe we can restrict what is supported.
> > 
> > > Is this wider re-factoring something that can wait for the next
> > > developer cycle?
> > 
> > I would say so. It's not quite trivial to do nicely since
> > things are a bit tangled between util/async and replay.
> > 
> > > >> I had started on a conversion once but not completed it.
> > > >> I could resurrect if there is agreement on the API?
> > >
> > > I would certainly welcome it being cleaned up. The supported replay
> > > devices are very piecemeal at the moment.
> > 
> > I'll tidy up and post an RFC for how the new API might look.
> > 
> > Thanks,
> > Nick
>
> Fundamentally it's virtio net, up to Jason. I don't like messy
> APIs and people tend to get distracted and not fix them up
> if one does not make this a blocker.

Happy for objections, but FYI Michael did find the proposed API
tweak nicer, so shall we get these minimal fixes in for 9.1 then
switch them for next release?

Thanks,
Nick