[patch, fortran] PR 38220 Mark c_loc as pure and having a prototype

[Tobias Burnus]

Thomas Koenig wrote:
> c_loc and c_funloc are actually implemented somewhat strangely:  They
> are marked as external functions, so their first argument gets, via the
> g77 calling convention, the address of the argument.  This was done for
< simplicity's sake, and having looked at the code, I can understand
> why :-)
>
> This patch does nothing about this, but it is minimally invasive, and
> does work.  As this is a rejects-valid, I think this could potentiall go
> into current trunk.
>
> Regression-tested on i686-pc-linux-gnu.
> OK for trunk?  Or should I wait for 4.5 to branch?

OK for 4.4. Thanks for the patch.

Tobias