[Omp] A problem with guard objects and exceptions on Intel

[Michael Suess]

Hi,

I have just published a blog entry comparing the merits of guard objects vs. 
the critical construct (with C++ and OpenMP). The problem I have now is: My 
test programs with exceptions and guard objects do not work for the Intel 
Compiler. Since this would not be the first time I missed something, I 
figured I would double check here before filing a bug with Intel.

So if you find the time, please take a look at the code for the guard object, 
enclosed in the following article:
http://www.thinkingparallel.com/2006/08/21/scoped-locking-vs-critical-in-openmp-a-personal-shootout/

Combining this with the following short test program, 
test_omp_guard_exceptions.cpp:

#include <omp.h>
#include <cassert>

#include "omp_guard.hpp"

int main (void)
{
	omp_lock_t my_lock;

	// how many exceptions have been thrown (should be one for each thread)
	int ex_count = 0;

	// how many threads will be created (most likely)
	int numthreads = omp_get_max_threads ();

	omp_init_lock (&my_lock);

	#pragma omp parallel
	{
		try {
			omp_guard my_guard (my_lock);
			throw omp_get_thread_num ();
		}
		catch(int e) {
			#pragma omp atomic
			++ex_count;
		}
	}
	assert (ex_count == numthreads);
	omp_destroy_lock (&my_lock);

	return 0;
}

Running this program on the Intel Compiler, I get a segmentation fault. The 
Sun and Portland compilers run it just fine, so I think it may be a bug in 
the compiler. But since I have had this thought one too many times in the 
past and usually it is me and not the compiler who is wrong, I thought I 
would double-check.

Thanks for any insights,
Michael Suess

-- 
My blog on parallel programming: http://www.thinkingparallel.com
Michael Suess
Public PGP key: http://www.suessnetz.de/michael/michaelsuess.gpg
PGP key fingerprint: A744 AFBA CA93 620B 8701 AB98 D4CF 4F3C 945A 61FE