hybrid-graphics-linux team mailing list archive

[Moshe Nissim <mnissim@xxxxxxxxx>]

On 14/6/11 4:20 PM, Albert Vilella wrote:
>
> Right now there is no way of ignoring the intel graphics driver other
> than directly from the BIOS, and only in certain models, but with the
> optimus bumblebee you *are* using nvidia hardware *and* nvidia drivers
> for all the applications you call with optirun. There is a penalty in
> using VirtualGL routing, but other than that, it does use the nvidia
> drivers, that are on-par with the Windows drivers in performance.
>
> It would be useful if you could run this on Windows and Linux
> bumblebee to check the performance difference:
I tried to make this work, but I'm encountering a problem unrelated to
bumblebee.
I know this may be the wrong thread to ask, but maybe people already
solved it.

First step is to make the Intel driver work with Xorg and have a working
LCD display.
(without yest starting the second Xorg for the nVidia card).
This goes wrong. The LCD gets corrupted during boot, even before Xorg is
started.
(In fact, I disabled Xorg for the moment, by going to runlevel 3 instead
of 5).

It seems to happen late in the boot, I think at the moment i915 kicks in
and does the modesetting.
The LCD goes into a strange state, three-quarters of it is black, and
the right-hand quarter is
alternating white and black lines. The image is steady.
"underneath" everything works, the boot completes, and I can even start
Xorg  with the intel driver.
Its just that the actual displayed image remains bad. It does not even
'blink' when Xorg starts.
If I disable modesetting altogether (kernel command-line arg nomodeset)
the corruption is gone,
but the intel xorg driver refuses to work. I think I read somewhere that
on later versions (which I must use,
having a Sandybrdige graphics hardware), kernel-mode-setting is a
requirement for it to work.
Again, this is a Dell Latitude E6420.
Anyone encountering a similar problem?