uclibc vs glibc

Ryan.Oliver at pha.com.au Ryan.Oliver at pha.com.au
Wed Nov 3 14:22:31 PST 2004








-----hlfs-dev-bounces at linuxfromscratch.org wrote: -----

>To: Hardened LFS Development List
>From: Robert Connolly
>Sent by: hlfs-dev-bounces at linuxfromscratch.org
>Date: 11/03/2004 06:23PM
>Subject: Re: uclibc vs glibc
>
>I still have no luck getting the stage 1 toolchain up. The scripted
>toolchains
>for Uclibc work fine for me, I logged their commands but so far I'm
>unable to
>reproduce it by hand.
>
>I unpack Uclibc and Linux, point uclibc's config to the linux
>sources, do
>`make headers`, and copy Uclibc/include/* to /tools/include. I notice
>uclibc
>uses symlinks for include/asm and include/linux which is a bad habit,
>but I
>leave the linux sources alone so the symlinks stay working. I build
>binutils
>(with uclibc's patches) with --build=i386-pc-linux-gnu
>--host=i386-pc-linux-gnu --target=i386-linux-uclibc. Then I get stuck
>on gcc
>trying to build the same cross target (with uclibc's patches) with
>--enable-languages=c. I get:
>
>../../gcc-3.4.2/gcc/config/i386/linux.h:227:20: signal.h: No such
>file or
>directory
>../../gcc-3.4.2/gcc/config/i386/linux.h:228:26: sys/ucontext.h: No
>such file
>ordirectory
>make[2]: *** [libgcc/./_muldi3.o] Error 1
>
>Even though /tools/include/{signal.h,sys/ucontext.h} both exist. If I
>
>experiment more I continue getting lots of errors about missing
>headers that
>are not missing. :(
>
>Robert

Try hacking CROSS_SYSTEM_HEADER_DIR in gcc/Makfile.in to point
at /tools/include.
This will set cpp's search path to look there w a cross-gcc (and set the
correct path to be used by fixincludes).

 HTH
[R]




More information about the hlfs-dev mailing list