[Bug 47007] HDMI monitor polling causing 100ms rendering stalls

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=47007

--- Comment #26 from Tvrtko Ursulin <tvrtko.ursulin at onelan.co.uk> 2012-03-15 08:03:32 PDT ---
(In reply to comment #22)
> (In reply to comment #20)
> > Right, thanks for your comments. In this case how about this approach:
> > 
> > 1. Where there is shared DDC, one end being connected will imply the other is
> > disconnected.
> > 
> 
> I'm not sure we necessarily want to do that.  Even though they may have a
> shared ddc line, it would be nice to report the proper connected status.

What do you consider the proper connected status? With DVI-I made out of VGA-1
and DVI-1 in software, only one of them should be connected at a time, no?

> > This will ensure DVI or VGA DDC is not needlessly probed (EDID fetched) for the
> > unconnected end of the pair.
> > 
> > 2. On HPD irq force detect will run on shared DDC connector pairs.
> > 
> > This will ensure correct state after VGA is disconnected and DVI connected.
> > Otherwise VGA remains in its "sticky connect" (can't poll for disconnected).
> > 
> > I can add something to pass a HPD irq flag in radeon_connector which detect
> > functions would consume.
> > 
> > 3. I keep the code which skips expensive probing unless HPD sense has changed.
> > On R600+ and only for unshared connectors.
> > 
> > How does this sound?
> 
> What about the following:
> 
> DVI + VGA with shared ddc line.
> VGA connected and in use.  User connects DVI port, gets hpd irq, detect called.
>  hpd sense returns true, DVI is updated as connected, VGA is marked as
> disconnected.  User then potentially loses the monitor they are currently
> using.

Are you implying DVI + VGA with shared DDC lines come as either one shared
(DVI-I) connector, or two physical ones (DVI-? + VGA)? This is a very deep hole
indeed...

-- 
Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.