From: Santiago Gala (sgala_at_hisitech.com)
Date: 2002-05-01 13:38:08 UTC
Actually, I found that it is quite simple to get rid of the problem, by
just chaging the definition of
KERNELRELEASE := $(shell if [ -r $(VERFILE) ]; \
then grep UTS_RELEASE $(VERFILE) | cut -d" " -f3 | xargs echo; \ else uname -r; fi)
to just
KERNELRELEASE := $(shell uname -r;)
I wonder if a test which skipped the RedHat and Mandrake kernels to the else would be simple to do (or if the complex if is really needed).
The issue of compiling against a kernel other than the one running should be a configuration input, else this results in very confusing bugs.
In my case, I have sometimes three or four kernel sources, and the one running is sometimes different than the one I'm compiling for.
I think something along the lines of
export USE_KERNEL=/usr/src/linux-2.4.18-36mypatchedsmp make pci
together with some simple guessing (just the running one?) if the variable is not defined would actually simplfy the process. Now I have to either patch the Makefile, and loose the patches with every new release, since I cannot find a cvs repository.
Regards,
Santiago Gala
P.S.) Really nice and cool work. The fact that people doesn't speak means that it works very well, as you know ;) I can tell you from my host_ap_pccard powered laptop speaking with my host_ap_pci powered SMP box up in the office. Thanks for it!