After installing slackware64 13.37 and going into runlevel 3, I downloaded the NVIDIA-Linux-x86_64-285.05.09.run file from the nvidia website. I ran it with
as root. Seems that slackware64 13.37 comes with the kernel headers already installed because the nvidia install had no problems. I see that it built the nvidia module as well as compatible 32 bit opengl libs. I do see also that the nvidia script added
to a file in /etc/modprobe.d

Now for my question. How and where exactly does this module get loaded at boot? I do see it's located at
yet there's no mention of it in /etc/modprobe.conf(seems like this file is not even used anymore in slackware64 13.37). The module does not load in /etc/rc.d/rc.modules-2.6.37.6 either. Is it just because the module is behind rc.modules-2.6.37.6, then it automatically gets loaded at boot?

If that were true, then all the other modules in
would get loaded as well. What makes nvidia.ko special? Which file actually modprobes the module?

Trying to understand exactly how it works.
In dmesg I see this pertinent output.
Which script exactly loads the nvidia.ko module?

My GUESS is this:

All the available .ko modules are not linked at boot unless hardware is found that matches them. Even though the module is linked it's still not loaded until X or some other application is called. X binary claims the "available" nvidia driver on execution as identified in the /etc/xorg.conf config file.

There maybe a script that modprobes it, but I couldn't find it either.

Good question, I'd love to hear someone elses take.

smoooth103 is essentially correct, Xorg can load the kernel module if it's not already loaded when X starts. This is also how the open source DRM drivers would often get loaded before they implemented KMS. Now they are usually either loaded into the initrd or loaded by udev. I would guess that udev could also be configured to load the nvidia kernel module, but I don't know how common that is.

Adam

Have a look in /usr/share/X11/xorg.conf.d/

I have rc.hald and rc.udev set to run. My default runlevel is 3. When I boot and login to console, lsmod shows nvidia as being loaded. So it should be safe to assume that Xorg did not load it. Now I do see udevd --daemon running, which is listening for kernel uevents. Where else should I check.

Essentialy what you are asking is why the system is decide to load a spcific module and not the other (all of the in the /lib/module dir).
And is true in all of the hardware you have.
I assume that there is just one module (plus a generic one) that is good for your hardware and it is load based on the kernel probe, hence you see the other module in a blacklist, in order for it not to be loaded before the nvidia one, i assume that all the other modules that are in your video/ dir aren't accepted by you system (not include the generic one usually vesa).

Yea that makes sense. I see all those video modules under /etc/modprobe.d/blacklist.conf. Ok great thanks.

It might be the kernel hotplug scripts are loading matching available driver/hardware configs in the /lib/modules/ dir, like hen770 is suggesting. I think the answer to your question though is: it's mostly in the kernel.