On Tue, 2010-07-06 at 17:35 -0400, Mauriat Miranda wrote:
> On Tue, Jul 6, 2010 at 1:36 PM, Gerhard Magnus <email@example.com> wrote:
> Sadly the proprietary nvidia driver is a mess.
> The best answer to all your questions is: Try and see. If you don't
> have problems then you're fine.
> > QUESTION: Google tells me that specifying "nomodeset" basically tells
> > the system to use an older set of software/drivers which does work but
> > may be slower or have some other issues that people would like to
> > eliminate. Is the "nomodeset" parameter necessary in this case?
> Some people claim the driver didn't work without it, while others
> (myself included) have found it made no noticeable difference(?). It
> only takes 1 extra reboot to test this.
> > The rpmfusion.org/Howto/nVidia instructions also say that SElinux
> > protection should be lowered:
> > setsebool -P allow_execstack on
> > QUESTION: Is this really necessary? Does it risk compromising the
> > security of my system?
> I would guess no (not any more). I didn't need it when I tested it on
> a few different systems. I think it was needed once upon a time.
> Again, test your installation without running that command (more
> secure). If you find SELinux errors in your log files or that the
> driver won't load due to SElinux then execute that command.
> > MOST IMPORTANT QUESTION: Suppose this doesn't work and the reboot hangs
> > (presumably with the X startup). Will I be able to use a previous kernel
> > (on the grub menu) or will yum have automatically associated
> > kmod-nvidia-PAE with all the retained kernels? Or is the back-out
> > procedure (from run level 3) to (1) yum remove kmod-nvidia-PAE (2) take
> > out the extra parameters on the kernel line in grub.conf, and (3)
> > reboot?
> Typically when you 'yum' install the Nvidia driver like this, it is
> built for a specific kernel. The yum command may also install a new
> kernel in order to match the nvidia driver. Going back the previous
> kernel should work (I think). Try and see, no harm done.
> In the case of a locked X-server, you can as you state go back to runlevel 3.
> To force this at grub, hit a key to see the kernels menu and select
> one but don't hit enter.
> Hit 'E' (to edit) and scroll to the end of the 'kernel' line and add '3'.
> Then hit enter and 'B' to boot.
> Then you can do the steps you stated to remove.
I've been able to replace the nouveau driver with an nvidia one by (1)
adding rdblacklist=nouveau to the active kernel entry in grub.conf and
(2) yum install kmod-nvidia-PAE. The last little "Gotcha!" for his
problem was the need to run nvidia-settings as root (rather than from
the Applications --> System Tools --> nvidia Display Settings menu):
Otherwise I was unable to permanently activate the second monitor by
saving the X configuration file.
Thanks for the help with this!
users mailing list
To unsubscribe or change subscription options: