On Sun, May 21, 2006 at 02:16:42PM +0200, [EMAIL PROTECTED] wrote:
> i don`t now wath is it or im to stupid. since 5 week, i try to build a
> clfs-multilib but every time a error cames up. the last 6 trys ending at
> the last step in "constructing-cross-compoile-tools"
> 5.10. GCC-4.1.0 - Cross Compiler Final
> ..
> ..
> checking for ld that supports -Wl,-z,relro... yes
> checking for main in -lm... configure: error: Link tests are not allowed
> after GCC_NO_EXECUTABLES.
> make[1]: *** [configure-target-libstdc++-v3] Error 1
> make[1]: Leaving directory `/media/hda6/sources/gcc-build'
> make: *** [all] Error 2
>
> i don't now why! )-:
> i attached my "little" build script.
> can everybody help me?
I've only taken a quick look at this, it seems to have the
instructions from the book, and a delay for each of the 'Important'
warnings. My immediate thought is that if the patches fail to apply
correctly (that is, the patch files exist, but parts do not apply),
the only indication will be rejection messages(s) on the screen (for
any of the patches, not just these). But, that normally only
happens if a patch gets damaged, e.g. downloading it to a windows
machine will add ^M at the end of each line.
Similarly, you will never get an error from sed if it fails to
change anything, but only if it detects a syntax error. Again, I
don't think problems with the seds are very likely.
So, we need to find out some more information - what is in
$MACHTYPE ? What sort of machine, and OS or distro, are you
building on ? If you are building on an athlon64 running linux, is
your current kernel 32 or 64 bit ('uname -m'), and is the userspace
32 or 64 ?
Also, please look at config.log in the libstdc++ directory to find
the *last* error message in it, perhaps it will be more informative
(e.g. you might be missing a library on the host, or more likely
something like crt1.o in the cross toolchain did not get installed)
Because this error is happening on the first package after you
installed a 64-bit libc, I suspect something there is missing (or
else something from the 32-bit libc).
In an ideal world, it often helps to log all the output (stdout and
stderr in the same file) so that errors which do not halt the build
or install can be recorded. But, mindlessly logging can bring its
own problems (like catching the output from sed in the log instead
of the desired file), so I can't automatically recommend it. What
might be useful, if we can't get anywhere, is to add something like
make 2>&1 >$S/glibc-32-buildlog
make install 2>&1 >$s/glibc-32-installlog
for the 32-bit glibc 'make' and 'make install', and something
similar (different names, of course) for the 64-bit glibc, so that
you can then look to see what happened.
Ken
--
das eine Mal als Tragödie, das andere Mal als Farce
_______________________________________________
Clfs-support mailing list
[email protected]
http://ninja.linux-phreak.biz/mailman/listinfo/clfs-support
