Building und/libiberty for powerpc-eabi cross compiler fails due to linker error

[Kai Ruottu]

On 3 Feb 2004 at 13:41, Hermann-Simon Lichte wrote:

> I am trying to build a powerpc-eabi cross compiler on an i686-pc-cygwin 
> host. I followed Bill Gatliff's tutorial, however, building the complete 
> cross compiler fails.
> 
> I am using binutils-2.14, gcc-3.3.2 and newlib-1.11.0. Compiling 
> binutils, the bootstrap gcc, and newlib isn't a problem.
> 
> When compiling the complete gcc, 'make all' fails because the compiler 
> does not know what target it should generate executables for.

 Is somewhere any evidence that the usual way of building newlib-
based GCCs doesn't produce working libraries with the generic
newlib headers?  Ie. when one copies the '.../newlib/libc/include'
into one's chosen $prefix/$target, makes the necessary symlinks
to be seen in the $prefix/$target/sys-include, configures GCC
using the '--with-newlib', builds (almost) everything coming with
the GCC sources and installs the stuff. Then builds and installs
newlib.

 The PowerPC target(s) replace some 'generic' headers with the
PPC-dependent ones with AltiVec-function protos and such, but
do these really have any influence to libgcc, libiberty and
libstdc++ ?  Most targets use the generic headers as they are,
so building all the newlib-based GCCs in the same way could be
the rule.

 Someone with a quick PC could experiment and see whether
the generic headers could produce different results than when
the powerpc-specific headers are used.

 Ok, in this case it was weird that the '--with-newlib' was not
used in the GCC configure although newlib was the C library
and creating executables with the defaults doesn't succeed.

 BTW, the bug with the '__end' is in all those '.ld' files for PPC coming
with newlib-1.11.0...

Cheers, Kai


------
Want more information?  See the CrossGCC FAQ, http://www.objsw.com/CrossGCC/
Want to unsubscribe? Send a note to crossgcc-unsubscribe@sources.redhat.com