Unnable to do ICA/farce comparative builds

M.Canales.es macana at macana-es.com
Wed Dec 6 03:33:42 PST 2006

El Miércoles, 6 de Diciembre de 2006 12:10, Greg Schafer escribió:

> IMHO there is little point in reinstalling the kernel headers during
> subsequent ICA iterations. I've never done it. After all, they are not
> binary, they are just ascii text.

Are the headers files already sanitized inside the kernel tree or are they 
generated on-the-fly by "make headers-install" command ?

If the last, the tools used to generate the headers files (the ones in /tools 
for the first build, but the ones in /{bin,usr} in iterative builds) could 
have some impact on how they are generated and what contains each one.

> That said, it does seem strange that current LFS uses different methods
> for kernel headers installation in each phase ie: they are cp'd into place
> in the the temp system, but "make INSTALL_HDR_PATH=/usr headers_install"
> is used directly in the real system. It's unfortunate that
> "headers_install" wants to wipe out /usr/include in that scenario.

In the chapter05 phase when the linux kernel headers are installed 
the /tools/include directory already contains headers from binutils-pass1 
build, thus the "cp" method is used to not delete them.

> If you insist upon reinstalling the kernel headers during subsequent ICA
> iterations, the chroot phase kernel headers cmds should be switched to the
> cp method. Otherwise, just skip the reinstall like I do.

To switch to the "cp" method in chapter06 would be the most failsafe solution, 
but for jhalfs to work that change should be done in the book. If not, the 
most simple is to skip the headers reinstallation, if we are sure that that 
will have no impact on the comparative analysis.

Manuel Canales Esparcia
Usuario de LFS nº2886:       http://www.linuxfromscratch.org
LFS en castellano: http://www.escomposlinux.org/lfs-es http://www.lfs-es.info
TLDP-ES:                           http://es.tldp.org

More information about the lfs-dev mailing list