Re: Notes from the Real World

jmjonesat_private

On Tue, 1 May 2001, Crispin Cowan wrote:

> Greg KH wrote:
> 
> > On Tue, May 01, 2001 at 08:03:48PM -0400, jmjonesat_private wrote:
> > > Okay, I stand corrected, but a loadable module DOES provide a means of
> > > extending the kernel to provide other functionality without actually
> > > touching the Kernel proper.  Not So?
> >
> > Yes, that is true.  I guess it depends on where you see the "kernel
> > proper" ending.  Remember Linux is a monolithic kernel, it's all "proper" :)
> 
> What loadable kernel modules do is separate the the kernel address space domain
> (what code can read/write directly to kernel memory) from the kernel source
> code domain (who can add code to the kernel, and whether they have to
> re-compile the whole kernel to do it):
> 
>    * Without loadable modules:  if you wanna run some code inside the kernel's
>      address space, you hafta patch the source and re-compile the kernel.  A
>      big hassle for all, and beyond the capabilities of many.
>    * With loadable modules:  someone can prepare a module that works with the
>      standard kernel, and give it to their friends/customers.  With no special
>      knowledge and no re-compiling, that code runs inside the kernel address
>      space.
> 
> However, Greg's point stands:  building kernel modules is not for amateurs.
> Someone who wants to build an LSM module had better be comfortable with kernel
> hacking.  The LSM is about making it easy distribute & install such
> enhancements, not about making it user-friendly to write them.
> 

Um, define "amateurs"?  Was Linus a "professional" when he built the
kernel all those years ago?

I suspect, your definition will include most of the linux world.
How about we change that to "inexperienced"?  In that case, there are
a LOT of us out here who could legitimately "try their hand" at an LSM on
their own systems, with a little "enlightenment".  A coherent, well
designed set of patches would make it easier for "those of us not blessed"
to add to the total linux picture.

> Naturally, we should make it as easy as possible to write LSM modules, but not
> at the expense of kernel mainline conventions.  And there is no escaping the
> hard fact that you are writing kernel code when you write LSM module code.

True enough.  Describe the "necessities" in plain language, you have
something very valuable here.

> 
> Crispin
> 
> --
> Crispin Cowan, Ph.D.
> Chief Scientist, WireX Communications, Inc. http://wirex.com
> Security Hardened Linux Distribution:       http://immunix.org
> 
> 
> _______________________________________________
> linux-security-module mailing list
> linux-security-moduleat_private
> http://mail.wirex.com/mailman/listinfo/linux-security-module
> 

|>------------------------------------------------------
||  J. MELVIN JONES            jmjonesat_private 
|>------------------------------------------------------
||  Microcomputer Systems Consultant  
||  Software Developer
||  Web Site Design, Hosting, and Administration
||  Network and Systems Administration
|>------------------------------------------------------
||  http://www.jmjones.com/
|>------------------------------------------------------

_______________________________________________
linux-security-module mailing list
linux-security-moduleat_private
http://mail.wirex.com/mailman/listinfo/linux-security-module