On Tue, Jun 15, 2004 at 02:05:49PM +0200, Arjan van de Ven wrote:
> > for some value of work. I can guarantee you that this will
break within the
> > next half year or so.
>
> Can you explain that?
Not using the headers that *exactly match* the kernel binary is broken.
Rebuilding a lot of stuff from /usr/src/<anything> is not going to
guarantee that it's an EXACT match.
> choice please don't break it. You would be removing appreciated added
> value and would piss off all kernel-level ISVs and repos including
> fedora.us.
It's broken infrastructure and is using implementation details instead
of properly provided infrastructure. Those implementation details are
subject to change. I mean that. Let me repeat it one more time:
If you don't use the headers that come with the kernel binary, things
will break, now or in the future.
The fact that you could get away with it in the past does not mean that
it will remain working in the future.
Is there something wrong with using the headers from
make mrproper
modify EXTRAVERSION in Makefile to reflect target arch/flavour
cp -a configs/<my flavour and arch>.config .config
make oldconfig_nonint x2 & make dep or make nonint_oldconfig & make prepare
make -s include/linux/version.h
These _are_ the headers matching the targeted arch/flavour combination
w/o the need to build or even install/run the kernel itself. And it is
not violationg rules or breaking anything. These are the usual steps
required and recommended for building external kernel modules adapted
to Red Hat/Fedora's kernel source mods (nonint and configs/ stuff).
Would something break the above? It should not. At least there is no
sign upstream considers changing kbuild again in this respect, and the
rpms are under your control.
--
Axel.Thimm at
ATrpms.net