The kernel doesn’t automagically find any devices. They must all be specified either on the command line, or as what typically happens by loading specific modules. They could be compiled into the kernel. But that should not be the case. You should be using standard “stock” kernels. Now how does the kernel know what to load? It looks at the file:
This file lists modules which are loaded at boot time.
hendry@bilbo:~$ cat /etc/modules
# /etc/modules: kernel modules to load at boot time.
# This file should contain the names of kernel modules that are
# to be loaded at boot time, one per line. Comments begin with
# a #, and everything on the line after them are ignored.
These get setup by base-config when installing Debian(possibly? installer?). Anyway, if you copying a Debian partition onto another machine, then this probably needs checking as the other machine will undoubtedly have different hardware. Now:
Is a definition for what modules that get loaded by an alias trigger or something. I am not sure about this myself, so I need to update this. This file is pretty important. You should manage this file with update-modules and some other tools. Looking here solved a problem for a friend of mine(Jamie), who had a problem with his network cards:
I edited /etc/modules.conf from
alias eth0 8139too
alias eth1 8139too
alias eth2 8139too
alias eth0 eepro100
alias eth1 eepro100
RH specific (as this file does not exist on my system):
(I got the "eepro100" from the eth driver info from /etc/sysconfig/hwconf)
Jamie Kitson (who solved this particular problem by himself. I didn’t. He did. He asked for me to make that clear) adds:
I think you should mention that the problem only arose because I had told
kudzu to ignore the changes, and it did detect the h/w changes, and would
have updated modules.conf for me if I had let it.