Comment # 4 on bug 1231067 from pallas wept

Hi Stefan!

We had seen this document from nvidia and figured out that the driver was the
root cause. But the real issue here, the reason why I have logged this bug, is
that the same driver packages worked for some users, and not for other users.
There was some indication that the kernel module parameters are not being
handled as expected.

All users involved, whether working or not, had the modprobe conf file which
set both of these parameters. For the users who had a failure occur, we have
all been able to work around it by adding (not removing, strange) the two
module parameters (modeset=1 fbdev=1) to the boot config.

I'm sure there has been one misunderstanding - since we saw success by adding
these params, we assumed that the modprobe conf files were not taking effect
previously. What you have explained shows that in fact, they were taking
effect, and that is what broke the system.

Still, this leaves me wondering why things start to work when we add the
params. Are we effectively 'breaking' fbdev, thus working around the bug? Are
we accidentally 'fixing' fbdev, so things work now? Something else? It seems
weird.

My only concern here is that I have stumbled over a small bug and I might leave
it behind for someone else to stumble on later. If there's no future problem
for future people, I'm not concerned now :) I know the driver itself is fixed,
so if the package's config files are working as expected, there's nothing to do
here.

Cheers for your help, especially lately I know there has been a lot of pressure
on you, especially because of all the beta drivers that people think have been
released. I think you're doing great.