Bug#348873: regression: r128 on laptop ignores supplied modeline for external display, picking LCD panel pixel dimensions

[Tim Connors]

On Sat, 23 Jun 2007, Tim Connors wrote:

> On Sat, 23 Jun 2007, Brice Goglin wrote:
>
> > Tim Connors wrote:
> > > and then Xorg.0.log goes on to claim it knows nothing about the IBM
> > > modelines:
> > > (II) R128(0): Not using mode "IBM_mode_1" (no mode of this name)
> > > (II) R128(0): Not using mode "IBM_mode_2" (no mode of this name)
> > > (II) R128(0): Not using mode "IBM_mode_3" (no mode of this name)
> > >
> > > Then I end up with cruddy 1024x768:
> > > (**) R128(0): *Mode "1024x768 at 75": 65.0 MHz (scaled from 83.9 MHz), 60.1 kHz, 74.4 Hz
> > > (which I find odd, because if it truly was using the flat panel's
> > > settings, should it not have picked up the 60Hz rate the flat panel
> > > requires?  It works on both displays when I cycle through the
> > > laptop's external/interal settings)
> > >
> >
> > Could you try with xserver-xorg-video-ati 1:6.6.192-1 currently in
> > experimental in case it helps?
>
> Same result with 1:6.6.192-1.
>
> > Or you feel more adventurous, you could try the randr-1.2 branch of the
> > upstream git repository since it should be better at discovering
> > modes/resolutions/... automatically without the need for anything in
> > xorg.conf. I have some packages of this somewhere if you want to try it.
>
> The latter certainly will be a problem -- I have good reason for wanting
> to supply my own modelines -- I have a fixed frequency external monitor.
> I don't need better discovery -- this all used to work, and I've already
> worked out which modelines will work on my monitor, and X will not do a
> better job than me at detecting necessary modelines from a monitor that
> doesn't do DDC.  This used to work prior to 6.9 -- I haven't changed
> anything to do with my carefully crafted modelines, so those should
> continue to be working.

A mandrake user has discovered a workaround, and I have just submitted
more info upstream to https://bugs.freedesktop.org/show_bug.cgi?id=5832

Bad assumptions seem to be made about the presense of flat panel registers
implying that a flat panel is being used.  I have proposed a more robust
workaround rather than the complete disabling of the
valid-modelines-for-a-flat-panel check, but wouldn't know how to actually
put this into action.

Hopefully someone knows enough about the driver to be able to check this.

-- 
TimC
"I give up," said Pierre de Fermat's friend. "How DO you keep a
mathematician busy for 350 years?"