openMP parallelism for FORTRAN subroutines

[sahiljhawar]

I am trying to write another C/C++ wrapper to computer Lstar shell splitting. However while using openmp, I run into issues, which is most likely due to thread safety. For example, I run the parallel for loop for N time steps, with #pragma omp parallel, just using this produces bad values (-1e+31), however if I put make_lstar_shell_splitting1_ in #pragma omp critical, it runs but on single thread, hence I loose the parallelism. Given this, is it safe to say that FORTRAN code is not thread-safe. I also see COMMON blocks in the code (onera_desp_lib.f)

[AntoineBrunet]

Yes, IRBEM is not thread safe, mostly due to the commons blocks scattered in the code. There are OpenMP guards which make commons blocks thread-local, and they could be used here.

[sahiljhawar]

Can you give some guidance on how to use to these OpenMP guards?