Ok not really sure if it is memory corruption or just a bug in my code, but here is the description of the problem: Single threaded program registers a SIGALRM handler:
struct sigaction action;
sigset_t blocked;
sigfillset(&blocked);
sigdelset(&blocked, SIGKILL);
sigdelset(&blocked, SIGSTOP);
action.sa_handler = &handler_function;
action.sa_flags = SA_RESTART;
action.sa_mask = blocked;
// We are not interested in 开发者_Go百科the previous action:
if (sigaction(SIGALRM, &action, NULL) == -1)
....
handler_function is a simple function which calls a function of an object:
void handler_function()
{
some_object->tick();
}
tick()
{
++variable;
}
Variable is of type long. my program has only one instance of some_object and only signal handler calls function tick and only tick function modifies variable. Other functions only read value of variable. Forgot to mention that I create a timer:
struct timeval interval;
interval.tv_sec = 0;
interval.tv_usec = (100000);
struct itimerval timer;
timer.it_interval = interval;
timer.it_value = interval;
if (setitimer(ITIMER_REAL, &timer, NULL) == -1)
...
so tick function is triggered 10 times per second. Now the problem is that intermittently i get other data members of object some_object corrupted, they are POD types(long), i.e. all of a sudden they contain random values. I've tried to isolate a problem in a small program, but to no avail. Any advise would be appreciated. thanks
There's a buffer overflow in your socket reading code. Fix that and the problem will go away.
精彩评论