Re: Nvidia compile error [SOLVED]

[Elijah <desiderata@softhome.net>]

> I'm glad you got it working!
> 
> The above doesn't make sense to me.  If you compiled your kernel you have
> your kernel headers.  /usr/share/doc/nvidia-kernel-src/README.Debian.gz
> describes this.
> 
> If you look at the Nvidia tarball file NVIDIA_kernel-1.0-4191/Makefile it
> uses
> 
>   KERNDIR=/lib/modules/$(shell uname -r)
> 
>   ifeq ($(shell if test -d $(KERNDIR)/build; then echo yes; fi),yes) 
>     KERNINC=$(KERNDIR)/build/include
>   else
>     KERNINC=/usr/src/linux/include
>   endif
> 
> And on my machine:
> 
> $ ls -l /lib/modules/`uname -r`/build
> lrwxr-xr-x    1 root     root           29 Aug  7 22:07
> /lib/modules/2.4.18-xfs-laptop/build -> /usr/src/kernel-source-2.4.18
> 
> points back to the kernel source directory.
> 
> I've never used a kernel-headers package.  I assume it creates a similar
> symlink.
> 
> -- 
> Bill Moseley moseley@hank.org
> 

As I remember from unpacking the kernel-headers it unloads a directory
kernel-header>include> it seems to only have the 'include' directory
(I'm not sure though). Then I do: (from the nvidia-kernel)

make SYSLINUX=/usr/src/kernel-header/include/

it compiles after that (no need to compile the headers, as I didn't but
still worked). 

After all that excitement I think I also remember symlinking the
headers:
ln -s /usr/src/kernel-headers<version> /usr/src/linux

then compile the nvdia-kernel

(I can't remember much of which method I used but I assume both worked
since they seem all the same to me as it points to my kernel-headers)