GNU MP to embedded system?

[Hon.Hwang@csiro.au]

Hi all,

Has anyone attempted to build GMP on an embedded system?

I am currently attempting to do this and have a question regarding the insertion of compiler and linker flags to the GMP build process.

Specifically I like to know how to add '-lfloat' (without quotes) to the GMP build process.

I need this because on my initial builds, some symbols relating to floating points are undefined in libgmp.a.  These symbols are:

__floatsidf
__fixunsdfsi
__muldf3

The '-lfloat' (without quotes), I assume is a linker flag to import the code from libfloat, which means I'll also like to know how to add -L flag to the GMP build process.

Do I specify these flags during ./configure or do I have to change the Makefiles?

BTW, the target platform does not support hardware floating point (on CPU and kernel level).  So floating point facilities are provided by user level library (I think it's the GNU one).

Thanks.

Hon Hwang.