[Courses][Linux comands] customizing a 2.4 kernel

Telsa Gwynne hobbit at aloss.ukuu.org.uk
Thu Mar 25 21:15:51 EST 2004


On Thu, Mar 25, 2004 at 12:19:23PM -0500 or thereabouts, Terri Oda wrote:
> On Mar 24, 2004, at 10:30 PM, Carla Schroder wrote:
> >>My question has always been "Why?"  ie what sort of things require
> >>kernel-haking, or building?  How do we know this its the thing to do?
> >
> >1. To add some functionality it doesn't have
> >2. To clean all the junk out of a stock kernel. The various distros 
> >lard their
> >kernels with all kinds of drivers and features you'll never need
> >3. To optimize it for your hardware. Many distros default to i386, 
> >which is
> >rather silly for Athlon and P4 users. Plus, I've noticed that many 
> >times the
> >architecture is wrong- on both my Red Hat and Debian systems, PIII was
> >selected . Well both systems are Durons.
> >4. To fine-tune which things you want statically compiled, or loadable 
> >modules
> >5. You might want to test some alpha features, such as NTFS write, or 
> >CD
> >packet writing, or some such
> 
> 6. To learn! 
> 7. To use strange binary drivers that don't work with the stock kernel 
> from your distribution. 

What compiling that very first kernel taught me was that
 (a) it's not difficult, just tedious;
 (b) you are quite likely to miss something out;
 (c) it won't matter, because you can do it again; and
 (d) doing it again is _really_ boring.

So I would add

8: To prove to yourself that you can do it, and that it really
isn't that exciting :) 

I think the proving to yourself that you can is well worth
it. I don't think it's worth wasting an evening to prove it
to someone else, though :) 

Telsa



More information about the Courses mailing list