useful LKM?

Raul Miller (rdm@test.legislate.com)
Wed, 1 Oct 1997 15:09:30 -0400


For linux kernel modules to be useful, you'd have to be able to build
modules in a fashion which is independent of the current kernel
configuration. Is anyone working on this?

As it stands, modules are a neat hack, and they let kernel distributors
ship for a variety of architectures without having an outrageously large
kernel image, but they're pretty useless once you start playing
with different configurations under the same kernel version.

I can think of a variety of hacks to work around incompatibilities
(e.g. an abstract binding method that lets you insert conversion
code between the module and the kernel when the interface changes),
but none of this means anything if drivers really have to be
kernel configuration dependent.

I suppose it's meaningful to allow driver configuration to be
orthogonal to kernel configuration?

[I'm trying to get 2.0.31pre10 working, and it's not recognizing
my scsi drives, it turns out to be a simple configuration problem,
but some module issues got in the way for a bit.]

-- 
Raul