Radeon 3650HD laptop LVDS lid open/closed detection problem

Alex Deucher alexdeucher at gmail.com
Mon Jun 21 08:12:51 PDT 2010


On Mon, Jun 21, 2010 at 10:53 AM, Francisco Jerez <currojerez at riseup.net> wrote:
> Jerome Glisse <glisse at freedesktop.org> writes:
>
>> On Mon, Jun 21, 2010 at 03:31:19PM +0300, Pasi Kärkkäinen wrote:
>>> Hello,
>>>
>>> After fixing the dvi/hdmi detection problem I now have another problem
>>> with the HP EliteBook 8530p, which has Radeon 3650HD adapter.
>>>
>>> Here's a summary of the environment:
>>>
>>>      - laptop connected to a docking station.
>>>      - external display in use, connected with DVI to the dock.
>>>      - laptop lid closed, so internal LVDS display is not used.
>>>
>>> Now, when I start the laptop, I can see the BIOS and grub on the external DVI display.
>>> All fine so far. I select the Fedora 13 kernel, and Linux starts. I see the Fedora
>>> graphical boot on the external DVI display, just like it should be. GDM login prompt
>>> appears on the external DVI display, still everything fine.
>>>
>>> And then it goes wrong. After I login to X, the external display only shows the background
>>> picture.. it turns out the desktop stuff has been started to the internal LVDS display,
>>> which shouldn't be used at all since the laptop lid is closed!!
>>>
>>> When the laptop lid is closed, and external display is connected, I want to use only the external display..
>>>
>>> Any ideas how to troubleshoot this one?

Does it work ok if you boot up without the external display connected
and then connect and enable the secondary display after you've logged
in?  I have a similar issue here;  I think it's an issue with rhgb and
starting X.  If I boot with an external display, the entire plymouth
load sequence works fine, but then when X starts the external display
goes blank and the internal display hangs on the plymouth screen.  On
a related note, if I close the lid and turn off the LVDS output, gpm
never blanks the external monitor.  I have to manually force dpms with
xset.


>>>
>>> -- Pasi
>>>
>>
>> It's better to open bug when you face issue rather than mail, as it's
>> harder to track information in mail thread than in a bug. Your issue
>> is not easily fixed because there is many laptop with broken acpi which
>> report wrong lid status (some of them always report lid closed what ever
>> is the lid status, other always report lid open, ... i am not expert on
>> how broken this is but from what i have been told i should rather consider
>> drinking than trying to look into it and then go to the drinking step).
>>
>> Bottom line is that lid detection is unreliable thus so far we ignore
>> it silently. I think the plan is to monitor lid status change and if
>> we detect change from either open to close or close to open then we
>> can start assuming that acpi lid status is reliable and act accordingly.
>>
> In Nouveau we report connector_status_unknown for closed lids (On the
> kernel side unknown outputs are left disabled unless there's nothing
> else definitely connected: if lid detection doesn't work at all the
> system will still be usable). This would solve your problem if we made
> the X server set the first output known to be connected as RandR
> primary.
>
> In short, I see two different "bugs" here:
>  * radeon reports connector_status_connected when the lid is closed.

Do we really want to report disconnected when the lid is closed?  The
monitor is still connected.  Considering how unreliable lid events are
I don't think that makes sense.  We probably actually want dpms off,
but connected which is more of a policy thing and should be handled by
userspace.

Alex

>  * the X server doesn't select a primary output among the definitely
>   connected ones.
>
>> Cheers,
>> Jerome
>> _______________________________________________
>> dri-devel mailing list
>> dri-devel at lists.freedesktop.org
>> http://lists.freedesktop.org/mailman/listinfo/dri-devel
>
> _______________________________________________
> dri-devel mailing list
> dri-devel at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/dri-devel
>
>


More information about the dri-devel mailing list