Problem in Timers and signal

Question

I have implemented a POSIX timer using timer_create( ) API, and this will generate SIGUSR1 when the timer expires for which i have put a handler code. Now the problem is, if this program receives another SIGUSR1, then the same signal handler will be invoked and caught.

Is there any way to prevent this, so that the handler can catch signals only generated by the timer?

Rumple Stiltskin · Accepted Answer · 2011-04-21T10:35:28.053

Will this work for you? (Modified the code from example in timer_create man page.)

#include <stdlib.h>
#include <unistd.h>
#include <stdio.h>
#include <signal.h>
#include <time.h>

#define CLOCKID CLOCK_REALTIME
#define SIG SIGUSR1
timer_t timerid;


static void handler(int sig, siginfo_t *si, void *uc)
{
    if(si->si_value.sival_ptr != &timerid){
        printf("Stray signal\n");
    } else {
        printf("Caught signal %d from timer\n", sig);
    }
}

int main(int argc, char *argv[])
{
    struct sigevent sev;
    struct itimerspec its;
    long long freq_nanosecs;
    sigset_t mask;
    struct sigaction sa;

    printf("Establishing handler for signal %d\n", SIG);
    sa.sa_flags = SA_SIGINFO;
    sa.sa_sigaction = handler;
    sigemptyset(&sa.sa_mask);
    sigaction(SIG, &sa, NULL);

    sev.sigev_notify = SIGEV_SIGNAL;
    sev.sigev_signo = SIG;
    sev.sigev_value.sival_ptr = &timerid;
    timer_create(CLOCKID, &sev, &timerid);
    /* Start the timer */

    its.it_value.tv_sec = 10;
    its.it_value.tv_nsec = 0;
    its.it_interval.tv_sec = its.it_value.tv_sec;
    its.it_interval.tv_nsec = its.it_value.tv_nsec;

    timer_settime(timerid, 0, &its, NULL);
    sleep(100);
    exit(EXIT_SUCCESS);
}

When signal from timer is caught Caught signal 10 from timer will be displayed. Otherwise Stray signal will be displayed.

Extremely sorry, i saw your post late. You have given an excellent solution. Bravo!!!! And also loads and loads of thanks :) — RajSanpui, Apr 27 '11 at 08:17
+1 Great thanks! It turned out that timer ins UNIX is simple, just a bit confusing — Viet, May 29 '12 at 23:45
what if i have to use multiple timers using this API? I tried but after executing handler once, the program exits. is there any way? — mihirj, Nov 21 '13 at 21:25
Sure, you just change `struct sigaction sa;` to make it an array (and you rename all of those variables to meaningful, maintainable, names ;-) — Mawg says reinstate Monica, Jul 13 '16 at 09:03

score 3 · Answer 2 · edited Jul 13 '16 at 09:06

3

The question is whether you really need to use signals. You may think of using callback that will be called when the timer expires:

void cbf(union sigval);
struct sigevent sev;
timer_t timer;

sev.sigev_notify = SIGEV_THREAD;
sev.sigev_notify_function = cbf; //this function will be called when timer expires
sev.sigev_value.sival_ptr = (void*) arg;//this argument will be passed to cbf
timer_create(CLOCK_MONOTONIC, &sev, &timer);

The callback function will be called in a new thread.

edited Jul 13 '16 at 09:06

Mawg says reinstate Monica

38,334
103
306
551

answered Apr 21 '11 at 08:08

Patryk

1,421
8
21

No, we can't use SIGEV_THREAD, because our library does not support that. Also, SIGEV_THREAD creates useless threads which in turn cause memory leaks. – RajSanpui Apr 21 '11 at 08:29
In what way do threads created by system calls cause memeory leaks? Have you reported this? – Mawg says reinstate Monica Jul 13 '16 at 09:05
While I _do_ like this (+1), as it is much clearer than the solution using signals, I have to ask - if you cancel the timer before expiry, will the callback association also be cancelled? – Mawg says reinstate Monica Jul 13 '16 at 09:35

score 1 · Answer 3 · answered Apr 21 '11 at 08:09

1

No, there is no easy way. Why don't you use SIGUSR2 instead for your timers if you have something else generating SIGUSR1 together with your timer. If that is not enough, use one of the real time signals for your application.

If it must be able to handle the same signal from the timer and some other source, then depending on how fast, how many, what system, etc etc you could try setting a timestamp before registering a timer on when the timer will approximately exit, and then in the signal handler try to deduce if it was within time margin. I would strongly advise not to use this approach, but instead redesign what you are doing.

answered Apr 21 '11 at 08:09

Milan

15,389
20
57
65

One of the real-time signals? Can you please elaborate? What if the test code also generates a real time signal? Won't the same handler catch it? – RajSanpui Apr 21 '11 at 08:32
Yes, the handler will catch the signal. The point with real time signals is that they should be used within your own application where you know what happens. Scroll down to real time signals on the man page and read more http://linux.die.net/man/7/signal – Milan Apr 21 '11 at 08:52

score 1 · Answer 4 · edited May 23 '17 at 12:32

1

Use another RT signals. See answers on Is there any way to create a user defined signal in Linux?

edited May 23 '17 at 12:32

Community

1
1

answered Apr 21 '11 at 13:45

Yann Droneaud

5,277
1
23
39

Problem in Timers and signal

4 Answers4

Linked