Improper locking bugs due to the unrelesed locks before destorying

[ycaibb]

Hi, developers, thank you for your checking! Should the lock be released before destroying it(Line 158 and 164)? According to this, it says It shall be safe to destroy an initialized mutex that is unlocked.

neko/vm/threads.c

Lines 150 to 165 in 1df580c

	pthread_attr_t attr;
	pthread_attr_init(&attr);
	pthread_attr_setdetachstate(&attr,PTHREAD_CREATE_DETACHED);
	pthread_mutex_init(&p.lock,NULL);
	pthread_mutex_lock(&p.lock);
	// force the use of a the GC method to capture created threads
	// this function should be defined in gc/gc.h
	if( GC_pthread_create((pthread_t*)handle,&attr,&ThreadMain,&p) != 0 ) {
	pthread_attr_destroy(&attr);
	pthread_mutex_destroy(&p.lock);
	return 0;
	}
	pthread_mutex_lock(&p.lock);
	pthread_attr_destroy(&attr);
	pthread_mutex_destroy(&p.lock);
	return 1;

[ycaibb]

Also, it seems the lock(Line 154) would be released by the thread(Line 117), is my understanding correct? If so, I think the attribute of mutex should be set to pthread_mutexattr_setpshared to enable this:

The process-shared attribute is set to PTHREAD_PROCESS_SHARED to permit a mutex to be operated upon by any thread that has access to the memory where the mutex is allocated, even if the mutex is allocated in memory that is shared by multiple processes.

neko/vm/threads.c

Lines 110 to 124 in 1df580c

	static THREAD_FUN ThreadMain( void *_p ) {
	tparams *lp = (tparams*)_p;
	tparams p = *lp;
	p.init(p.param);
	// we have the 'param' value on this thread C stack
	// so it's safe to give back control to main thread
	# ifdef NEKO_WINDOWS
	ReleaseSemaphore(p.lock,1,NULL);
	# else
	pthread_mutex_unlock(&lp->lock);
	# endif
	clean_c_stack(40,clean_c_stack);
	p.main(p.param);
	return 0;
	}

Thank you for your checking! Looking forward to the discussion.