Re: [Qemu-devel] [PATCH 2/3] migration: Wake rate limiting for urgent requests

["Dr. David Alan Gilbert" <dgilbert@redhat.com>]

* Peter Xu (peterx@redhat.com) wrote:
> On Wed, Jun 13, 2018 at 11:26:41AM +0100, Dr. David Alan Gilbert (git) wrote:
> 
> [...]
> 
> > @@ -2932,10 +2943,24 @@ static void *migration_thread(void *opaque)
> >  
> >          migration_update_counters(s, current_time);
> >  
> > +        urgent = false;
> >          if (qemu_file_rate_limit(s->to_dst_file)) {
> > -            /* usleep expects microseconds */
> > -            g_usleep((s->iteration_start_time + BUFFER_DELAY -
> > -                      current_time) * 1000);
> > +            /* Wait for a delay to do rate limiting OR
> > +             * something urgent to post the semaphore.
> > +             */
> > +            int ms = s->iteration_start_time + BUFFER_DELAY - current_time;
> > +            trace_migration_thread_ratelimit_pre(ms);
> > +            if (qemu_sem_timedwait(&s->rate_limit_sem, ms) == 0) {
> > +                /* We were worken by one or more urgent things but
> > +                 * the timedwait will have consumed one of them.
> > +                 * The service routine for the urgent wake will dec
> > +                 * the semaphore itself for each item it consumes,
> > +                 * so add this one we just eat back.
> > +                 */
> > +                qemu_sem_post(&s->rate_limit_sem);
> 
> Is it possible that we just avoid eating that in the next patch?  Then
> we only provide a helper to "trigger an urgent request" but we only
> consume the point here?

I think it's harder;
This code is generic in migration.c where as the next patch is all
specific in ram.c; so we'd have to push a flag all the way down.
Also, the code later is very simple - every request it adds/removes
it posts/waits the semaphore - it's nice to keep that simple.

Dave

> -- 
> Peter Xu
--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK