[PlanetCCRMA] MadWIFI and edge kernels

Fernando Lopez-Lezcano nando@ccrma.Stanford.EDU
Sat May 14 11:16:02 2005

On Sat, 2005-05-14 at 10:48, Frode Petersen wrote:
> I wrote (some time ago, but have not got any response, so I try again):
> > I guess this would seem a trivial thing to ask, but could the madWifi 
> > kernel driver for any of the stock FC3 kernels be used with the edge 
> > kernels? If so, how do I determine the correct one to try? If not, are 
> > there any issues with compiling them?
> The issue I'm trying to get an answer to is general, in that I'm curious 
> about the relationship between kernel modules compiled for an official 
> FC kernel and the modified kernels from PlanetCCRMA, especially the edge 
> kernels. Is the base kernel that you modify from Fedora, or from Linus' 
> tree? 

Hmmm, mostly Linus plus the realtime preempt patch from Ingo. But I
migrate _some_ of the patches from the Fedora kernelsas well. I have not
done that in quite a while (it is very time comsuming) so I don't know
which ones may be missing. Presumably it is not that important as in
general the edge kernels are based on a much newer version of the base
kernel, one would imagine many of the bugs are already fixed :-)

> You create new alsa modules for all(?) PlanetCCRMA kernels; is 
> this a requirement for all kernel modules? 

A kernel module must match the kernel it was built for (including the
compiler that was used to build it). So, yes, when I build - for example
- the additional, more up to date, alsa driver kernel modules I do that
for every kernel I want to support for that release. A module built for
kernel "a" will not load under kernel "b". 

> Is there any difference 
> between compiling a module towards the FC kernels and towards the Planet 
> ones?

No, it should be the same deal. Provided, of course, that what the
kernel module needs is supplied by the kernel itself. Some module may
not be happy with the configuration options I used for building the
kernel although I try to stay as close to the Fedora kernels as

> I'll be thankful for any answer, and I won't be resentful for getting 
> none: I just hope... ;-)

Sorry for the delay...
-- Fernando