"Michael J. McGillick" <[EMAIL PROTECTED]> wrote;
>Afternoon Everyone:
>
>It seems I very easily get myself into trouble trying to recompile the
>Linux Kernel. Some attempts in the past have worked just fine, others
>tend to blow things up, and I usually end up having to reinstall. So far,
>I understand the first few steps of what needs to be done to recompile the
>kernel.
There are a few subtleties involved in recompiling a kernel if you
want it to coexist with the stock kernel or other custom kernels
in a RedHat world. I wrote a document describing the steps for doing
this in RH5.2, and I recently updated it for RH6.x. You can grab
it at
ftp.math.clemson.edu/users/mjs/Linux/Redhat-6.1-custom-kernel.txt
>[...]
>When I did this, I got the help screen for make which tells me that this
>is not a valid argument. A little guesswork told me to use:
>
>make oldconfig
Good point. I didn't include this step, but I will in a future version.
>Ok. Rolling along now. A small shell script appeared to run, and then I
>typed:
>
>make menuconfig
>[...]
>Ok, it's at this point that I'm completely baffled as to what to do
>next. In my root directory, I now have the two files vmlinuz, and
>System.map. I know that both of these files need to be linked up to the
>lilo configuration, so the system can boot to them. When I look in the
>/boot directory, I see a symlink called System.map pointing to the
>System.map for the current kernel. I also see ones for vmlinuz and
>modules. Now, I've been warned in the past that you don;t just blindly
>replace stuff, as the new kernel may not boot.
Right.
>What is the exact sequence of steps to set up this new kernel in a test
>position, so that I don;t disturb the original kernel, and can get back
>to it? What files, symlinks, etc. need to be changed? Red Hat apparently
>has their reasons for setting up the system the way that they do. When I
>first started with linux, and recompiling a kernel, there were never any
>symlinks to deal with. You simply added an entry into lilo.conf, and told
>it where to find the new kernel.
This is what I tried to make sure I covered carefully. Let me know
if it helps or if you have suggestions for improvement.
Matthew Saltzman
Clemson University Math Sciences
[EMAIL PROTECTED]
http://www.math.clemson.edu/~mjs
--
To unsubscribe: mail [EMAIL PROTECTED] with "unsubscribe"
as the Subject.