[Spice-devel] [PATCH spice-gtk] Notify about existence of monitor for all display channels

[Jonathon Jongsma]

----- Original Message -----
> From: "Pavel Grunt" <pgrunt at redhat.com>
> To: "Marc-André Lureau" <mlureau at redhat.com>
> Cc: spice-devel at lists.freedesktop.org
> Sent: Thursday, July 30, 2015 11:20:27 AM
> Subject: Re: [Spice-devel] [PATCH spice-gtk] Notify about existence of monitor	for all display channels
> 
> Hi
> 
> On Thu, 2015-07-30 at 10:35 -0400, Marc-André Lureau wrote:
> > Hi
> > 
> > ----- Original Message -----
> > > Windows guest can have disabled display on the display chanel #0, in
> > > that case it will not be listed in virt-viewer's "View->Displays" menu
> > > 
> > > Resolves:
> > > https://bugs.freedesktop.org/show_bug.cgi?id=91489
> > > ---
> > > I think notify::monitors should stay in create_canvas(), because in that
> > > function the monitor
> > > configuration changes (if the channel does not have
> > > SPICE_DISPLAY_CAP_MONITORS_CONFIG),
> > > so the notify should be emitted. The notify is also emitted when
> > > SPICE_MSG_DISPLAY_MONITORS_CONFIG
> > > is recieved, Windows guests are getting this message only for enabled
> > > displays.
> > > ---
> > >  src/channel-display.c | 3 +--
> > >  1 file changed, 1 insertion(+), 2 deletions(-)
> > > 
> > > diff --git a/src/channel-display.c b/src/channel-display.c
> > > index e3f03fc..8837bd0 100644
> > > --- a/src/channel-display.c
> > > +++ b/src/channel-display.c
> > > @@ -811,8 +811,7 @@ static void spice_display_channel_up(SpiceChannel
> > > *channel)
> > >  
> > >      /* if we are not using monitors config, notify of existence of
> > >         this monitor */
> > > -    if (channel->priv->channel_id != 0)
> > > -        g_coroutine_object_notify(G_OBJECT(channel), "monitors");
> > > +    g_coroutine_object_notify(G_OBJECT(channel), "monitors");
> > >  
> > 
> > And for guests that do send MonitorConfig messages on channel 0, this is
> > not
> > going to race?
> 
> There should not be the race, this notify happens in _channel_up() and always
> has monitors->len = 0, monitors_max  = 1. virt-viewer then creates
> monitors_max
> spice display widgets, but does not enable any of them because
> monitors->len=0.
> Later the SPICE_MSG_DISPLAY_MONITORS_CONFIG is received and the displays are
> enabled in virt-viewer.


In addition to what Pavel said, if there was a potential for a race on channel 
0, that possibility would also exist for channels 1, 2, and 3 (since they can 
already be enabled when virt-viewer connects). So I don't see any problem with 
treating all channels the same.

I'm comfortable ACKing this patch, but I'd like to have the comment above 
this line of code changed or removed. It's a bit confusing.

Jonathon