Re: Device drivers != Modules

Clifford Wolf (clifford@clifford.at)
Sat, 13 Jun 1998 23:02:28 +0200 (MEST)


On Sat, 13 Jun 1998, Fabricio Chalub wrote:

> One of the good and bad points of Linux is that all device drivers are so
> tightly integrated with the kernel. This is good because, you have a really
> tight code and really fast performance (I think). But this is extremely bad
> because you can't dissociate device drivers with the main kernel. Are there
> any projects of making it possible for adding some kind of support for
> "hooks" in the kernel, where you could add a device driver for some piece of
> equipment without even messing with the kernel? This is not the same as
> modules, though, since the kernel have to know which module is which, in
> *compile time*. I'm talking about adding drivers to the kernel in
> *run-time*.

The kernel does not need to know about modules at compile time! It looks
like becouse you have the same configuration tool for compileing something
staticaly into the kernel or making it a module - but in fact you can
compile a module compleatly independed from compileing the kernel and then
just insmod it. I did this allready a lot of times with the modules i
wrote. The only exceptions are modules they are not compleatly written as
modules - so you need to compile the non-module part of it into the kernel.

But again: normaly you can just write a module and call then functions
like register_filesystem(), register_netdev() or register_blkdev() in
the init function of the module (this functions are the hooks you would
like to have - and there are much more of them than this 3).

- clifford

-- -- -- -- -- -- -- -- -- -- -- -- --
Clifford Wolf
magnet - Internet at Work IRC: efnet / clifford
Director of System Development http://www.clifford.at/
e-mail: c.wolf@magnet.at email: god@clifford.at

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu