Actually, it probably did work, as it seemed like the commands I typed into the blank screened console were accepted, so X probably didn't load, but I still could not see what I was typing without using ssh.

And thank you for explaining how slim got removed from the rcS.conf file. I had not mentioned it because I wasn't sure I did not remove it myself, but I didn't think I had.

I must say that I think the changing of configuration files based on the kernel boot parameters seems like a really bad idea, especially since the changes aren't logged anywhere. It is one thing to skip loading X because of a boot parameter and quite another to change a config file that effects every subsequent boot.

mojo- is there any way to warn unsuspecting users to prevent this problem? Either a new package or at least remove the broken one?

The version of the nvidia driver is to old to work with the version of xorg in cooking.The issue with nouveau is more complicated. I believe the newer versions of the nvidia-installer from nvidia's web site check for the presence of nouveau on your system. If detected a warning is given to remove nouveau.

If the user has not setup an ssh server on the box as I had before installing the nvidia driver repairing this problem could prove challenging, especially for an inexperienced user.

To boot into console to do a repair add screen=text to the bootcode of grub.
When this is parsed by /etc/init.d/bootopts.sh it removes slim from /etc/rcS.conf RUN_DAEMONS=
After you fix the problem you have to add slim back to /etc/rcS.conf RUN_DAEMONS="dbus hald slim to auto start slim/Xorg on boot.

Running tazx and choosing the nv driver does most of what you suggested I think, but that wasn't all the shrapnel left behind by the nvidia bomb. Here's what else I found:

1.) The nouveau kernel driver kept loading so I had to add it to the blacklist, even though all packages with nouveau in the name are removed. Finally I found that the kernel drivers had been added to /etc/rcS.conf, on the "LOAD_MODULES" line, and I deleted nouveau and nvidia-xxxxxxxxxx from that line.

2.) I noticed that at least two additional files were created in /etc/X11/xorg.conf.d, 55-Device-something and 90-Dri-something (sorry, I already deleted them so I don't have the exact names anymore). I deleted them.

After these changes the system seems to be back to normal.

So, if anyone else has this problem, edit/delete those files in addition to following mojo's directions above (or run tazx, then edit/delete files.)

mojo- is there any way to warn unsuspecting users to prevent this problem? Either a new package or at least remove the broken one? If the user has not setup an ssh server on the box as I had before installing the nvidia driver repairing this problem could prove challenging, especially for an inexperienced user.

Today I installed from the 5/4/11 rolling release. Installed and configured with nv driver, everything worked fine. Tried to install the nvidia driver using controlbox/hardware. The driver seemed to install fine, checked slim.conf, driver not listed, edited file to load driver. Rebooted, got blank screen. Could not change to a console that I could see.

logged in via ssh, ran tazx, tried to change back to nv driver. Rebooted. Same problem. Ssh again, lsmod shows nouveau driver still loading even though tazpkg says it is not installed. Add nouveau to /etc/modprobe.d/blacklist. Reboot. System boots, but will not autologin anymore. I have to login and run startx, slim doesn't run at all.

How do I restore my system to the condition it was in prior to loading the nvidia drivers? How do I remove the nouveau driver completely and prevent it from loading without using the blacklist, as it was before I tried installing nvidia? What other changes did the failed attempt to install the nvidia driver make and why doesn't reinstalling the correct driver work anymore?