This is the mail archive of the
libc-help@sourceware.org
mailing list for the glibc project.
Re: Making glibc-2.13 work on other CPUs
Hello,
thanks for replying.
Am Mon, Apr 25, 2011 at 05:59:38AM -0500 schrieb Jonathan Nieder - jrnieder@gmail.com:
>
> Take a look at your favorite .diff.gz file in
> <http://cdn.debian.net/debian/pool/main/e/eglibc/>:
>
> cd $(DEB_BUILDDIR) && \
> CC="$(call xx,CC)" \
> CXX="$(call xx,CXX)" \
> AUTOCONF=false \
> MAKEINFO=: \
> $(CURDIR)/configure \
> --host=$(call xx,configure_target) \
> --build=$$configure_build --prefix=/usr --without-cvs \
> --enable-add-ons=$(standard-add-ons)"$(call xx,add-ons)" \
> --enable-profile \
> --without-selinux \
> --enable-stackguard-randomization \
> --with-pkgversion="Debian EGLIBC $(DEB_VERSION)" \
> --with-bugurl="http://www.debian.org/Bugs/"; \
> $(call xx,with_headers) $(call xx,extra_config_options))
>
> So I think you are looking for the --host option to "configure".
> The INSTALL file has details.
>
This sounded very promising, but it still didn`t work.
I configured glibc thus:
../glibc-2.13/configure --enable-add-ons --with-gd=/overspill/sekundaer --with-headers=/usr/src/linux/usr/include --host=i486-linux-gnu --prefix=/usr
Additionally, I supplied
-O2 -march=i486
as gcc CFLAGS. Build went ok, but make check failed on a number of catgets. I installed anyway, and hell broke loose. Final words:
make[4]: Leaving directory `/mount/glibc-2.13/time'
make[3]: Leaving directory `/mount/glibc-2.13/elf'
/usr/bin/install -c /mount/build_glibc_host/elf/ld.so /lib/ld-2.13.so.new
mv -f /lib/ld-2.13.so.new /lib/ld-2.13.so
/usr/bin/install -c /mount/build_glibc_host/libc.so /lib/libc-2.13.so.new
make[2]: *** [/lib/libc-2.13.so] Segmentation fault
make[2]: Leaving directory `/mount/glibc-2.13/elf'
make[1]: *** [elf/ldso_install] Error 2
make[1]: Leaving directory `/mount/glibc-2.13'
make: *** [install] Error 2
The rescue disk had to step into the limelight.
I'm still all at sea. I should mention that gcc used for building glibc had been rebuilt with glibc-2.13 i686 (i.e. on the Pentium III). Should this be a problem - I thought supplying -march=i486 would make gcc produce i486 code, but smeg knows what's going on internally.
Recap: all I aspire for is a glibc that can deal with both Pentium I and III.
Regards
Andreas