GMP on Arm architecture
Jonathan A. Zylstra
jon at jzylstra.com
Tue Aug 2 18:51:14 CEST 2005
Ganeriwal Saurabh wrote:
>I am trying to cross compile the GMP library for arm architecture. I am
>doing it on a Pentium 4 machine running Fedora 4 distribution. None of the
>steps: configure, make and make all gives me any problem. In fact, I am
>also able to see the final .h and .a files.
>However, when i do "make check" in between make and make install as
>mentioned in the documentation, it gives me "All the checks failed: 5 out
>of 5 failed". Now is this something to worry. Is the problem with the fact
>that I am doing cross compiling and not performing the checks on the
>intended platform (arm out here).
That's the problem. By cross compiling, you are making an executable
with machine code that only an ARM machine
can understand. This machine code is different than the x86 machine code
that your Pentium runs on. The 'make check' command
actually creates dozens of small executables which run a few
calculations each, and it checks the results of these and compares them to
the correct answers. Since these executables can't run, the checks
therefore fail. They have to be run on the ARM machine.
Now, as far as I remember, once a test fails, the 'make check' quits,
and doesn't compile any further test programs.
If anyone else knows of a quick way to just compile all of the test
programs without running them that would help.
[my first guess would be 'make -k check', which continues to run make,
ignoring errors -- but that's just a guess]
That way you could compile all the programs on your P4, move them to the
ARM, and then just run them on the ARM without worrying about compiling.
Note that if you're creating a dynamic library, you would also have to
copy over the main gmp library over to the ARM machine before running
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 3166 bytes
Desc: S/MIME Cryptographic Signature
Url : http://gmplib.org/list-archives/gmp-discuss/attachments/20050802/c6e9794b/smime.bin
More information about the gmp-discuss