Spake Eben King on Tuesday, April 17, 2007 at 03:44PM -0400:
> Fine. Thousands of people _don't_ do it that way and it works. Heck, I
> used to do it the standard Nvidia way, and the only thing different between
> then and now is the GPU (still supported) and the kernel version (up
> 0.0.5). It's even the same version of the driver. My kernel is down to
> 1.1M, simply by taking out unneeded drivers in which one day a bug may have
> been tickled. So what does the Nvidia driver actually require?
>
To use a custom compiled kernel, you just have to run 'make-kpkg kernel_image' instead of 'make' in the kernel source dir.
This should be enough for module-assistant to tell the nvidia driver
where the kernel headers are.
I don't know what modules nvidia needs, but module-assistant
auto-install will complain when they're not there.
It is perilous to go outside of one's package management system.
On another note: I used to remove all "unneeded" drivers from my kernels
(and I still used module-assistant for nvidia and ndiswrapper).
Eventually, I got tired of getting new hardware devices and having them
not work. It's also very nice to buy an entirely new computer, swap
harddrives, and have everything just boot and work.
(Except for perhaps modprobing an ethernet driver.)
I've never successfully installed nvidia drivers using the
non-package-managed way. I've tried dozens of times, but it always
breaks... On the other hand, it's one or two commands to install it via
the package systems of Arch Linux, Gentoo, and Debian.
-----------------------------------------------------------------------
This list is provided as an unmoderated internet service by Networked
Knowledge Systems (NKS). Views and opinions expressed in messages
posted are those of the author and do not necessarily reflect the
official policy or position of NKS or any of its employees.
This archive was generated by hypermail 2.1.3 : Fri Aug 01 2014 - 16:13:35 EDT