This is the mail archive of the gcc-help@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Tutorial Time - part 2


Hi All

As a comparison, I have spent some time today attempting a cross build of
an ARM-ELF target.

It was interesting in as much as it got passed the stage which the Mcore
had problems with and says that it is using the sjlj exception model
(short-jump-long-jump as I discovered in
GCC-3.0.4/LIBSTDC++-V3/DOCS/HTLM/configopts.html).

This time the error is due to:
configure: error: installation or configuration problem: C compiler can not
create executable.

I checked out some details and found that the problem is this time in
libiberty, where the compiler is tested by compiling a test program called
conftest.c with the content:


#line 1945 "configure"
#include "confdefs.h"

main(){return(0);}


The command used is:

/usr/local/src/gnu/BUILD/gcc/gcc/xgcc -B/usr/local/src/gnu/BUILD/gcc/gcc/
-B/usr/local/arm/arm-elf/bin/ -B/usr/local/arm/arm-elf/lib/ -isystem
/usr/local/arm/arm-elf/include -g -O2 conftest.c



I have run this command and receive the same compiler error (actually a
linker error)

/usr/local/arm/arm-elf/bin/ld: cannot open crt0.o: No such file or
directory
collect2: ld returned 1 exit status


At this point I am not sure what has gone wrong since crt0 does not seem to
be a part of ARM (I have seen that crtend.o and crtbegin.o, I assume the
ARM equivalents, have been created in the build directory  gcc and also
gcc/thumb). It seems as though these programs are needed to create an
executable - hopefully for the target and not the host...



Now I assume that many more gcc-users have experience with ARM than with
Mcore, so there may be someone out there who knows why this can go wrong -
possibly a configuration error ? Since it takes about 5 hours on my slow PC
at home to retry with different configuration parameters I will wait to see
whether someone sends a recommendation first (in the lists of successful
test reports I have found several examples - for example one using
--disable-nls and another using --with-newlib plus --enable-obselete%g
[haven't found a description of this one] - neither of which seems to be
critical).

Cheers

Mark in Switzerland



Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]