Radeon 3650HD laptop LVDS lid open/closed detection problem

[Pasi Kärkkäinen]

On Mon, Jul 26, 2010 at 05:13:30PM -0400, Alex Deucher wrote:
> On Mon, Jul 26, 2010 at 3:42 PM, Pasi Kärkkäinen <pasik at iki.fi> wrote:
> > On Mon, Jul 12, 2010 at 01:37:28PM -0400, Alex Deucher wrote:
> >> >> >>
> >> >> >
> >> >> > The other patch (0001-drm-radeon-kms-fix-shared-ddc-handling.patch) applied
> >> >> > my system mostly works now, but here's a summary about the problem I still have
> >> >> > with the lid detection:
> >> >> >
> >> >> > - I boot up the laptop with lid closed (LVDS inactive) so there's only external
> >> >> > DVI display connected. Kernel boot messages show up on the external DVI display,
> >> >> > and GDM login prompt appears on the external DVI display. All fine so far.
> >> >> >
> >> >> > - The actual problem: when X starts gnome panel etc show up on the internal LVDS
> >> >> > display, which I can't see at all since the lid is closed! So those should go to the
> >> >> > external DVI display only.. LVDS should be disconnected or inactive or something..
> >> >> >
> >> >> > Any pointers appreciated where to look at in the source.. I can do some debugging.
> >> >>
> >> >> Your desktop session manager should check the lid status when it loads
> >> >> and attempt to do the right thing if there is an external monitor
> >> >> detected.
> >> >>
> >> >
> >> > Ok.
> >> > So you think it's not a bug in the lid detection?
> >>
> >> Not sure.  That's handled by apci, not the video driver.  You can
> >> check it the lid is producing proper events by running:
> >> cat /proc/acpi/button/lid/LID/state
> >> with the lid open and closed.  The desktop manager decides what the
> >> policy is for the lid (blank display, suspend, turn off the connector,
> >> etc.).  It should also take into account other connected outputs, but
> >> I don't think it handles that too well at the moment.
> >>
> >
> > Yes, the lid acpi stuff seems to work:
> >
> > lid closed:
> > $ cat /proc/acpi/button/lid/LID/state
> > state:      closed
> >
> > lid open:
> > $ cat /proc/acpi/button/lid/LID/state
> > state:      open
> >
> > I also verified that the initial lid state is "closed" when
> > the lid has been closed all the time during system startup
> > and only external DVI display is in use.
> >
> > (I modified /etc/rc5.d/S01sysstat to sleep+print+sleep
> > so I can check it during system startup before X starts).
> >
> > When the lid is closed xrandr says "LVDS connected", is that correct?
> 
> Yes.  The LVDS is connected, even if you don't necessarily want to use it.
> 

That's what I was thinking of. But good to get confirmation :)

> >
> > I think LVDS actually is ON when lid is closed, since I can immediately
> > see everything when I open the lid.. correct colors etc.
> >
> > So what's the component I should start looking at.. gnome-power-manager?
> > or something else?
> >
> > Actually.. I just noticed that already in GDM prompt the internal LVDS
> > gets enabled/turned on, even when the lid is closed.. I think.
> 
> Yes, it's up to to gdm, gnome-power-manager, etc. to decide the
> display policy based on the lid state.
> 

Ok. Is there a way to monitor the status of drm from /proc or /sys or from somewhere? 

I guess I'll have to start reading GDM code to check what it does..

-- Pasi