Hi! Philip Herron-2 wrote: > > Thing is compiling without warnings doesnt guarentee no sigsegv :-). > I totally agree with you, if I didn't, I think I could throw my diploma in computer science in the garbage :-D. Philip Herron-2 wrote: > > This is all too big, try to build a small test case showing the error. > I was sure of that =^D. No problem, I reduced it to a simple function with a double for loop (no more packages or library, only openmp and pthread). I hope it's ok now, I can't reduce it more . I added two parameters to precise the numbers of processors and clients (not to recompile each time) ;-). http://old.nabble.com/file/p29181707/main.cpp main.cpp http://old.nabble.com/file/p29181707/Classe1.h Classe1.h http://old.nabble.com/file/p29181707/Classe1.cpp Classe1.cpp Philip Herron-2 wrote: > > You may want to run your program through gdb or valgrind might give you > some ideas I had already run it with valgrind and the error message it returns (only when the program segfaults, when it's a deadlock there is no message) is the following (it doesn't help me to solve the problem) : Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 0x41802950 (LWP 5109)] 0x00007f5b2656fbd0 in GOMP_loop_runtime_next () from /usr/lib/libgomp.so.1 I hope all that is ok for you to try to see the problem :-). Thanks! -- View this message in context: http://old.nabble.com/Problem-with-an-openMP-and-pthread-program-compiled-with-gcc-tp29173546p29181707.html Sent from the gcc - Help mailing list archive at Nabble.com.
