Re: [Qemu-devel] [RFC v2 10/22] monitor: allow to use IO thread for parsing

[Peter Xu]

On Thu, Oct 12, 2017 at 01:35:08PM +0100, Stefan Hajnoczi wrote:
> On Fri, Sep 29, 2017 at 11:38:32AM +0800, Peter Xu wrote:
> > For each Monitor, add one field "use_io_thr" to show whether it will be
> > using the dedicated monitor IO thread to handle input/output.  When set,
> > monitor IO parsing work will be offloaded to dedicated monitor IO
> > thread, rather than the original main loop thread.
> 
> Why make use_io_thr optional?  We'd have to maintain and test both code
> paths.

Because of MUXed typed chardevs.  I will add the explanation into this
patch's commit message, and also into the codes where we used it (in a
follow-up patch).

> 
> > @@ -4135,19 +4139,37 @@ void monitor_init(Chardev *chr, int flags)
> >          monitor_read_command(mon, 0);
> >      }
> >  
> > +    if (mon->use_io_thr) {
> > +        /*
> > +         * When use_io_thr is set, we use the global shared dedicated
> > +         * IO thread for this monitor to handle input/output.
> > +         */
> > +        context = monitor_io_context_get();
> > +        /* We should have inited globals before reaching here. */
> > +        assert(context);
> > +    } else {
> > +        /* The default main loop, which is the main thread */
> > +        context = NULL;
> > +    }
> > +
> > +    /*
> > +     * Hang the monitor before running it (which is triggered by
> 
> s/Hang/Add/
> 
> "Hang" isn't used as a verb for adding items to collections.  People
> would probably think about deadlocking the thread instead :-).

Thanks for explaining this!  Fixed.

(I am always happy to improve my English, though in a slow way :-)

> 
> > +     * qemu_chr_fe_set_handlers).  Otherwise one monitor may run while
> > +     * find itself not on the mon_list.
> > +     */
> > +    qemu_mutex_lock(&monitor_lock);
> > +    QTAILQ_INSERT_HEAD(&mon_list, mon, entry);
> > +    qemu_mutex_unlock(&monitor_lock);

-- 
Peter Xu