[Carpet] grid array is NULL pointer after DECLARE_CCTK_ARGUMENTS in local mode?

[Jonathan Thornburg]

I'm trying to debug a Cactus run which crashes, and I wonder if anyone
can suggest what might be going wrong.

I can reproduce the crash in a uniprocessor DEBUG=yes OPTIMISE=no
executable, and I have a stack traceback from gdb.  The thorn where the
crash occurs has the following schedule.ccl:

	storage: AH_shape
	storage: AH_origin
	storage: AH_ever_found
	storage: AH_found
	storage: AH_appearance_time
	storage: AH_rmin, AH_rmax

	schedule AHUtil_reset_vars at basegrid
		{
		lang: C
		} "reset variables"

The interface.ccl declares the relevant grid array like this:

	cctk_int   AH_ever_found[101] type=scalar "this is to tell if a given AH has ever been found"

The source code reads (in part) as follows:

	extern "C" void  AHUtil_reset_vars(CCTK_ARGUMENTS)
	{
	  DECLARE_CCTK_ARGUMENTS;

	  // initialize AH-related variables

	  memset(AH_ever_found, 0, 100*sizeof(CCTK_INT));

The crash occurs becaue the value of  AH_ever_found  is a NULL pointer
at the memset call. :(  I have verified this both in the core-dump
and by running cactus under gdb.

>From looking at the grid::[xyzr] grid functons inside gdb, the current
grid is the coarsest refinement level.


So... questions:

I had thought that with storage turned on outside any schedule block,
AH_ever_found would always be a non-NULL pointer.  Obviously I'm wrong.
Is it supposed to be this way?  That is, is this a Cactus/Carpet feature,
or a bug in some piece of software (possibly corrupting memory and
confusing things)?

ciao,

-- 
-- Jonathan Thornburg <jthorn at aei.mpg.de>      
   Max-Planck-Institut fuer Gravitationsphysik (Albert-Einstein-Institut),
   Golm, Germany, "Old Europe"     http://www.aei.mpg.de/~jthorn/home.html      
   "Washing one's hands of the conflict between the powerful and the
    powerless means to side with the powerful, not to be neutral."
                                      -- quote by Freire / poster by Oxfam