headers_install and ICA

Bruce Dubbs bruce.dubbs at gmail.com
Tue Jan 9 17:11:02 PST 2007


Ken Moffat wrote:
>  After a long absence on other architectures, I'm trying to get back
> into LFS itself.  I've got 2006-12-09 working nicely on two boxes
> (modulo unrelated kernel issues), so now I'm trying to run an
> in-place install so that I can see if farce shows anything useful.
> 
>  All headers in /usr/include are blown away by
> 'make INSTALL_HDR_PATH=/usr headers_install' which has "unfortunate"
> results on glibc (thinks I'm building for i686-pc-linux-gnulibc1 and
> then barfs when testing cpp because limits.h is now missing).  This
> is caused by
> 
> headers_install: include/linux/version.h scripts_basic FORCE
>         @if [ ! -r include/asm-$(ARCH)/Kbuild ]; then \
>           echo '*** Error: Headers not exportable for this architecture ($(ARCH))'; \
>           exit 1 ; fi
>         $(Q)$(MAKE) $(build)=scripts scripts/unifdef
>         $(Q)rm -rf $(INSTALL_HDR_PATH)/include
> 	^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>         $(Q)$(MAKE) -rR -f $(srctree)/scripts/Makefile.headersinst obj=include
> 
>  Still, I guess it probably isn't that important ;-)  OTOH, after
> all the effort people went to last year, we should probably sed it
> out, something like
> 
> sed -i '/^.*Q.rm -rf.*INSTALL_HDR_PATH.*$/d' Makefile
> 
> looks as if it will do the job (it definitely takes the line out,
> but I haven't tested it).  I would appreciate help with explaining
> _why_ the book should include this, when it only benefits people
> running ICA, and potentially assists people stupid enough to update
> the kernel headers in-place on a running system.  Any thoughts ?

I don't think it sounds like it is appropriate for the book.  OTOH, if
the book had wiki entries like BLFS, I think that would be an ideal
place to place the discussion of ICA issues.

  -- Bruce



More information about the lfs-dev mailing list