[Patch, Fortra, Coarray] Free allocated memory for static coarrays.

Tobias Burnus tobias.burnus@physik.fu-berlin.de
Thu Jun 16 12:55:00 GMT 2011

Hi Daniel,

Daniel Carrera wrote:
> This patch adds a linked list to keep track of the allocated
> memory of all static coarrays and ensures that at the end of
> the program the memory is freed.

+_gfortran_caf_register (ptrdiff_t size,	caf_register_t type,
 			void **token)

Between "size," and "caf_register_t" is a <tab> instead of a simple " ".

 _gfortran_caf_init (int *argc, char ***argv, int *this_image, int *num_images)
+  if (this_image)
+    this_image = caf_this_image;
+  if (num_images)
+    num_images = caf_num_images;

As you mentioned yourself (in a private mail, commenting on my code):
There is a "*" missing at the left side of the assignment lines.

+_gfortran_caf_register (ptrdiff_t size, caf_register_t type,
                         void **token)
+  token = malloc (sizeof (void*) * caf_num_images);
+  /* Start MPI if not already started.  */
+  if (caf_num_images == 0)
+    _gfortran_caf_init (NULL, NULL, NULL, NULL);

Wrong order for the "token" allocation: You first need to know how many
images exists, before you can use "caf_num_images".

+  MPI_Allgather (local,    sizeof (void*), MPI_BYTE,

There is a <tab> character between "local," and "sizeof"; it should
have been a space.

Have you tested the modifications with MPI? I mean: Compiling mpi.c,
and then in gcc/testsuite/gfortran.dg/coarray:

for I in *.f90; do echo ====== $I =====; \
  mpif90 -fcoarray=lib $I mpi.o && mpiexec -np 3 ./a.out; done

Additionally, have you run test suite? It is sufficient to run recompile
gfortran (to create the latest libcaf_single.a from single.c) and then to
  make check-gfortran RUNTESTFLAGS="caf.exp"
(caf.exp is sufficient as the rest of the testsuite does not use
libgfortran/caf/single.c, which is the only thing your patch modifies.)

Otherwise, your patch looks OK.

who has currently no access to the GCC source code and can hence not
test himself.

More information about the Gcc-patches mailing list