I am running the following program which implements a timer. When a thread awake after receiving a signal on condition variable from the previous running thread, it creates a timer and send a signal to the next thread on timer expiration. I want it to run for some time, but the timer stops ticking after some runs.
//Import
#define _POSIX_C_SOURCE 199309
#include <sched.h>
#include <unistd.h>
#include <sys/wait.h>
#include <stdio.h>
#include <stdlib.h>
#include <pthread.h>
#include <unistd.h>
#include <time.h>
#include <sys/time.h>
#include <signal.h>
#include <errno.h>
#define NUM_THREADS 10
#define CLOCKID CLOCK_REALTIME
#define SIG SIGUSR1
timer_t timerid;
pthread_cond_t condA[NUM_THREADS+1] = PTHREAD_COND_INITIALIZER;
pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
pthread_t tid[NUM_THREADS];
int state = 0;
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);
}
pthread_cond_signal(&condA[state]);
}
void *threadA(void *data_)
{
int i = 0, s;
long int loopNum, j;
int turn = (intptr_t)data_;
struct timeval tval_result;
// Timer's part starts
struct sigevent sev;
struct itimerspec its;
long long freq_nanosecs;
sigset_t mask;
struct sigaction sa;
// TImer'spart ends
while(1)
{
/* Wait for state A */
pthread_mutex_lock(&mutex);
for (;state != turn;)
{
s = pthread_cond_wait(&condA[turn], &mutex);
if (s != 0)
perror("pthread_cond_wait");
// printf("main(): state = %d\n", state);
}
pthread_mutex_unlock(&mutex);
//do stuff
for(j=0;j<10000;j++)
{//some dummy time consuming works}
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 = 0;
its.it_value.tv_nsec = 2000;
its.it_interval.tv_sec = 0;
its.it_interval.tv_nsec = 0;
timer_settime(timerid, 0, &its, NULL);
pthread_mutex_lock(&mutex);
state = (state +1)%NUM_THREADS;
//pthread_cond_signal(&condA[state]);
pthread_mutex_unlock(&mutex);
// Timer's code ends
}
}
int main(int argc, char *argv[])
{
int data = 0;
int err;
while(data < NUM_THREADS)
{
//create our threads
err = pthread_create(&tid[data], NULL, threadA, (void *)(intptr_t)data);
if(err != 0)
printf("\ncan't create thread :[%s]", strerror(err));
else
// printf("\n Thread created successfully\n");
data++;
}
pthread_exit(NULL);
}
Although no printf statements are executing, why is it freezing after some time?
If no. of timers are limited, what other strategy should I use to redress this issue?
POSIX says:
It is not safe to use the pthread_cond_signal() function in a signal handler that is invoked asynchronously.
Most likely you end up corrupting the state of pthread_cond_wait/pthread_cond_signal and anything can happen.
Don't mix threads and signal handlers, it leads only to madness. There are very few things you're allowed to do inside a signal handler, even fewer that are thread related, it's very hard to ensure that the right thread ends up handling the right signal, etc.
If you're doing threads anyway implement a timer in one thread that calculates how much time it needs to sleep to deliver the next event (don't just hardcode it to your timer period since that will make your timer drift), sleep that much and call pthread_cond_signal.
Also, it's bad form to have naked pthread_cond_signal calls and most often a bug. You might get unlucky and call it just before the other thread does the pthread_cond_wait and your signal will get lost. The normal thing to do is to set a variable (protected by a mutex, that's why pthread_cond_signal wants a mutex) and then signal that the variable is set.
If you think this is too much work, condition variables are probably not the right mechanism in this case and you should use semaphores instead. Incidentally sem_post is legal to call from a signal handler according to POSIX, but I still think it's a bad idea to mix threads with signals.
Related
So reading this topic, the common way to exit would be using a flag. My question is, how is the waiting handled? Say the thread is only to run every 30s, how would you wait those 30s properly?
Using sem_timedwait() isn't ideal as it relies on the system clock and any change to the clock can severely impact your application. This topic explains using condition variables instead. The problem is, it relies on a mutex. You can't safely use pthread_mutex_lock() and pthread_mutex_unlock() in a signal handler. So in terms of my example above of 30s, if you want to exit immediately, who is handling the mutex unlock?
My guess would be another thread that sole purpose is to check the exit flag and if true it would unlock the mutex. However, what is that thread like? Would it not be wasted resources to just sit there constantly checking a flag? Would you use sleep() and check every 1s for example?
I don't believe my guess is a good one. It seems very inefficient and I run into similar "how do I wait" type of question. I feel like I'm missing something, but my searching is leading to topics similar to what I linked where it talks about flags, but nothing on waiting.
#include <stdlib.h>
#include <stdio.h>
#include <stdbool.h>
#include <errno.h>
#include <pthread.h>
#include <signal.h>
#include <unistd.h>
pthread_mutex_t my_mutex;
volatile sig_atomic_t exitRequested = 0;
void signal_handler(int signum) {
exitRequested = 1;
}
bool my_timedwait(pthread_mutex_t *mutex, int seconds) {
pthread_condattr_t attr;
pthread_condattr_init(&attr);
pthread_condattr_setclock(&attr, CLOCK_MONOTONIC);
pthread_cond_t cond;
pthread_cond_init(&cond, &attr);
struct timespec ts;
clock_gettime(CLOCK_MONOTONIC, &ts);
ts.tv_sec += seconds;
int status = pthread_cond_timedwait(&cond, mutex, &ts);
if (status == 0) {
return false; // mutex unlocked
}
if ((status < 0) && (status != ETIMEDOUT)) {
// error, do something
return false;
}
return true; // timedout
}
void *exitThread(void *ptr) {
// constant check???
while (1) {
if (exitRequested) {
pthread_mutex_unlock(&my_mutex);
break;
}
}
}
void *myThread(void *ptr) {
while (1) {
// do work
printf("test\n");
// wait and check for exit (how?)
if (!my_timedwait(&my_mutex, 30)) {
// exiting
break;
}
}
}
int main(void) {
// init and setup signals
struct sigaction sa;
sa.sa_handler = signal_handler;
sigaction(SIGINT, &sa, NULL);
// init the mutex and lock it
pthread_mutex_init(&my_mutex, NULL);
pthread_mutex_lock(&my_mutex);
// start exit thread
pthread_t exitHandler;
pthread_create(&exitHandler, NULL, exitThread, NULL);
// start thread
pthread_t threadHandler;
pthread_create(&threadHandler, NULL, myThread, NULL);
// wait for thread to exit
pthread_join(threadHandler, NULL);
pthread_join(exitHandler, NULL);
return EXIT_SUCCESS;
}
The solution is simple. Instead of having the first thread block in pthread_join, block that thread waiting for signals. That will ensure that a SIGINT can be handled synchronously.
You need a global structure protected by a mutex. It should count the number of outstanding threads and whether or not a shutdown is requested.
When a thread finishes, have it acquire the mutex, decrement the number of outstanding threads and, if it's zero, send a SIGINT. The main thread can loop waiting for a signal. If it's from the thread count going to zero, let the process terminate. If it's from an external signal, set the shutdown flag, broadcast the condition variable, unlock the mutex, and continue waiting for the thread count to hit zero.
Here's a start:
pthread_mutex_t my_mutex; // protects shared state
pthread_cond_t my_cond; // allows threads to wait some time
bool exitRequested = 0; // protected by mutex
int threadsRunning = 0; // protected by mutex
pthread_t main_thread; // set in main
bool my_timedwait(int seconds)
{
struct timespec ts;
clock_gettime(CLOCK_MONOTONIC, &ts);
ts.tv_sec += seconds;
pthread_mutex_lock (&my_mutex);
while (exitRequested == 0)
{
int status = pthread_cond_timedwait(&my_cond, &my_mutex, &ts);
if (status == ETIMEDOUT) // we waited as long as supposed to
break;
}
bool ret = ! exitRequested;
pthread_mutex_unlock (&my_mutex);
return ret; // timedout
}
bool shuttingDown()
{
pthread_mutex_lock (&my_mutex);
bool ret = exitRequested;
pthread_mutex_unlock (&my_mutex);
return ret;
}
void requestShutdown()
{
// call from the main thread if a SIGINT is received
pthread_mutex_lock (&my_mutex);
exitRequested = 1;
pthread_cond_broadcast (&my_cond);
pthread_mutex_unlock (&my_mutex);
}
void threadDone()
{
// call when a thread is done
pthread_mutex_lock (&my_mutex);
if (--threadsRunning == 0)
pthread_kill(main_thread, SIGINT); // make the main thread end
pthread_mutex_unlock (&my_mutex);
}
Here i am trying achieve a timer say 0-10secs and an interval of 2secs each,so that i need to generate an interrupt # every 2secs(total 5 times) saying 2secs completed. which i have been cross checking in handler() function by using printf(). But i am not able to achieve the desired result.please put me touch if anyone is aware of it.
Thanks in advance.
#include <stdlib.h>
#include <unistd.h>
#include <stdio.h>
#include <signal.h>
#include <time.h>
#define SIGTIMER (SIGRTMAX)
#define errExit(msg) do { perror(msg); exit(EXIT_FAILURE); \
} while (0)
void handler(int sig, siginfo_t *si, void *uc)
{
printf("Caught signal %d\n", sig);
}
int
main(int argc, char *argv[])
{
timer_t timerid;
struct sigevent sev;
struct itimerspec its;
struct itimerspec oitval;
struct sigaction sa;
/* Establish handler for timer signal */
printf("Establishing handler for signal %d\n", SIGTIMER);
sa.sa_flags = SA_SIGINFO;
sa.sa_sigaction = handler;
sigemptyset(&sa.sa_mask);
if (sigaction(SIGTIMER, &sa, NULL) == -1)
errExit("sigaction");
/* Create the timer */
sev.sigev_notify = SIGEV_SIGNAL;
sev.sigev_signo = SIGTIMER;
sev.sigev_value.sival_ptr = &timerid;
if (timer_create(CLOCK_REALTIME, &sev, &timerid) == 0)
{
printf("timer ID is 0x%lx\n", (long) timerid);
/* Start the timer */
its.it_value.tv_sec = 10;
its.it_value.tv_nsec =0;
its.it_interval.tv_sec = 2;
its.it_interval.tv_nsec = 0;
if (timer_settime(timerid, 0, &its, &oitval) == -1)
errExit("timer_settime");
}
else
{
errExit("timer_create");
}
return 0;
}
First of all you should properly setup timeouts:
/* Start the timer */
its.it_value.tv_sec = 2;
its.it_value.tv_nsec = 0;
its.it_interval.tv_sec = 2;
its.it_interval.tv_nsec = 0;
it_value is the duration before first triggering, and it_interval is time between all subsequent triggerings. See here. Then, you shouldn't return from main() due it leads to process exit and you won't be able to observe timer ticks. You need to block execution somehow, e.g.
while(1) sleep(INT_MAX);
After each tick sleep() returns with errno set to EINTR so it should we wrapped into loop to allow timer to go ahead. Later you can decide when to leave this loop and exit.
P.S. Using printf() from signal handler is not a good idea. You should be very carefully with what you are doing there. It would be better just to write to some global variable and immediately return. And that var may be tested immediately after sleep() giving you an idea if you should sleep again or return.
I have been dealing with a problem in a thread pool for some days now. I tried all types of different things but I can't seem to solve the issue. I have made a simple version that reproduces the problem.
Code:
#include <unistd.h>
#include <signal.h>
#include <stdio.h>
#include <stdlib.h>
#include <pthread.h>
#include <time.h>
struct bsem_t bsem;
pthread_t threads[2];
/* Binary semaphore */
typedef struct bsem_t {
pthread_mutex_t mutex;
pthread_cond_t cond;
int v;
} bsem_t;
void bsem_post(bsem_t *bsem) {
pthread_mutex_lock(&bsem->mutex);
bsem->v = 1;
pthread_cond_broadcast(&bsem->cond);
pthread_mutex_unlock(&bsem->mutex);
}
void bsem_wait(bsem_t *bsem) {
pthread_mutex_lock(&bsem->mutex);
while (bsem->v != 1) {
pthread_cond_wait(&bsem->cond, &bsem->mutex);
}
bsem->v = 0;
pthread_mutex_unlock(&bsem->mutex);
}
/* Being called by each thread on SIGUSR1 */
void thread_exit(){
printf("%u: pthread_exit()\n", (int)pthread_self());
pthread_exit(NULL);
}
/* Startpoint for each thread */
void thread_do(){
struct sigaction act;
act.sa_handler = thread_exit;
sigaction(SIGUSR1, &act, NULL);
while(1){
bsem_wait(&bsem); // Each thread is blocked here
puts("Passed semaphore");
}
}
/* Main */
int main(){
bsem.v = 0;
pthread_create(&threads[0], NULL, (void *)thread_do, NULL);
pthread_create(&threads[1], NULL, (void *)thread_do, NULL);
pthread_detach(threads[0]);
pthread_detach(threads[1]);
puts("Created threads");
sleep(2);
pthread_kill(threads[0], SIGUSR1);
pthread_kill(threads[1], SIGUSR1);
puts("Killed threads");
sleep(10);
return 0;
}
What the code does, is create two threads. Both threads wait on a binary semaphore (bsem_wait). Then while they are waiting I send a SIGUSR1 signal to both resulting on pthread_exit() being executed on each thread. On my terminal it shows that everything goes as planned..
Output:
Created threads
Killed threads
2695145216: pthread_exit()
2686752512: pthread_exit()
The problem
Although the output seems correct, using pstree shows that only one of two threads die. The other thread stays alive until the whole program exits. Why is this?
Update
Replacing my custom binary semaphore with a normal semaphore seems to solve this for no apparent reason..
#include <unistd.h>
#include <signal.h>
#include <stdio.h>
#include <stdlib.h>
#include <pthread.h>
#include <time.h>
#include <semaphore.h>
sem_t sem;
pthread_t threads[2];
/* Caller thread will exit */
void thread_exit(){
printf("%u: pthread_exit()\n", (int)pthread_self());
pthread_exit(NULL);
}
/* Startpoint for each thread */
void* thread_do(){
struct sigaction act;
act.sa_handler = thread_exit;
sigaction(SIGUSR1, &act, NULL);
while(1){
sem_wait(&sem); // Each thread is blocked here
puts("Passed semaphore");
}
}
/* Main */
int main(){
sem_init(&sem, 0, 0); // Normal semaphore
pthread_create(&threads[0], NULL, thread_do, NULL);
pthread_create(&threads[1], NULL, thread_do, NULL);
pthread_detach(threads[0]);
pthread_detach(threads[1]);
puts("Created threads in pool");
sleep(2);
//PROBLEM
pthread_kill(threads[0], SIGUSR1);
pthread_kill(threads[1], SIGUSR1);
puts("Destroyed pool");
sleep(10);
return 0;
}
You can't get there from here
pthread_exit() is not listed in the "signal safe funtions" of the signal(7) man page.
rewrite your code to have the pthread_exit call outside of the signal handler.
So the issue seemed to be a deadlock!
The problem is that each thread is waiting inside the bsem_wait of the binary semaphore in different locations:
void bsem_wait(bsem_t *bsem) {
pthread_mutex_lock(&bsem->mutex); // THREAD 2 BLOCKED HERE
while (bsem->v != 1) {
pthread_cond_wait(&bsem->cond, &bsem->mutex); // THREAD 1 WAITING HERE
}
bsem->v = 0;
pthread_mutex_unlock(&bsem->mutex);
}
Thread 1 in this case is the fastest thread. Thread 2 is the slower one. When I run the signal to kill a thread, the waiting thread unblocks and exits as expected. The problem is that it never unlocks the mutex. So the blocked thread (2) remains blocked forever. For some reason, the thread won't be terminated because it is waiting on the mutex.
Just adding an unblock before exiting, solves the issue:
void thread_exit(){
printf("%u: pthread_exit()\n", (int)pthread_self());
pthread_mutex_unlock(&mutex); // NEW CODE
pthread_exit(NULL);
}
This ofcourse is a hack to demonstrate what is happening and shouldn't be used. I am going to follow Jasen's advice and get rid of the signal handler all together and solve it some other way. Namely I have to assure that the thread goes through the whole bsem_wait!
I'm trying to understand timers in linux to use it in my application, I've collected code from multiple sources and made the following program
#include <stdlib.h>
#include <unistd.h>
#include <stdio.h>
#include <signal.h>
#include <time.h>
#include <pthread.h>
pthread_mutex_t mutex1 = PTHREAD_MUTEX_INITIALIZER;
#define CLOCKID CLOCK_REALTIME
#define SIG SIGUSR1
timer_t timerid;
void *print_message_function( void *ptr );
static void handler(int sig, siginfo_t *si, void *uc)
{
pthread_mutex_lock( &mutex1 );
printf("Caught signal %d from timer\n", sig);
//it will be working like this in real application
// 1. check the front of the queue, if timeout then detete it from queue
pthread_mutex_unlock( &mutex1 );
}
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 = 1;
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);
pthread_t thread1, thread2;
char *message1 = "Thread 1";
int iret1 ;
/* Create independent threads each of which will execute function */
iret1 = pthread_create( &thread1, NULL, print_message_function, (void*) message1);
pthread_join( thread1, NULL);
printf("Thread 1 returns: %d\n",iret1);
exit(0);
}
void *print_message_function( void *ptr )
{
char *message;
message = (char *) ptr;
int i;
for(i=0;; i++){
pthread_mutex_lock( &mutex1 );
printf("%s \n", message);
//it will be working like this in real application
// 1. check if a response received from the network
// 2. if received then delete from queue
pthread_mutex_unlock( &mutex1 );
usleep(50022);
}
pause();
}
I'm trying to proctect the critical section which is in this case is the printf, the printf here is just an example, in my application it's actually a queue, so the thread will be working on the printf all the time and when the timer is ready it has to lock the mutex and then print, is the above the right way to do it?
In the real application, the thread will be waiting for a response and when the response is received by other application then I'll access the queue, the timer will always be checking the queue on a certain interval (e.g every 2 seconds) for timedout message and delete it if any found.
No, that is not the right way to do it. It is only allowed to call async-signal-safe functions from a singla handler. Otherwise the behavior is undefined:
«A signal handler function must be very careful, since processing
elsewhere may be interrupted at some arbitrary point in the execution
of the program. POSIX has the concept of “safe function”. If a signal
interrupts the execution of an unsafe function, and handler calls an
unsafe function, then the behavior of the program is undefined.»
For a bit more details, see man 7 signal.
If you are writing for Linux, consider using signalfd. Other operating systems have similar alternatives. For a long story, see «How Not To Write a Signal Handler».
Hope it helps. Good Luck!
I am implementing a simple timer that throws a RT signal upon expiration. What I want to do is to register a signal handler (using sigaction) that gets called when the signal occurs. Meanwhile the main code waits until the signal is called using sigwaitinfo.
Implementing either a signal handler or sigwaitinfo exclusively works fine. However when both are used, the signal handler is never called. I tried switching the order; i.e. registering the handler before blocking the signal. Makes no difference.
Here is the code
// gcc -Wall -o sigwait_example sigwait_example.c -lrt
#include <stdlib.h>
#include <unistd.h>
#include <stdio.h>
#include <signal.h>
#include <time.h>
#include <errno.h>
#include <string.h>
#define install_handler(sig,sa) if( sigaction(sig, &sa, NULL) == -1 ){ \
perror("sigaction"); }
#define SIG SIGRTMIN+1
volatile int flag=0;
void handler(int signum){
flag++;
}
int main(void){
struct itimerspec its;
sigset_t blocked;
siginfo_t si;
timer_t timerid;
struct sigevent evt;
struct sigaction sa;
evt.sigev_notify = SIGEV_SIGNAL;
evt.sigev_signo = SIG;
evt.sigev_value.sival_ptr = &timerid;
if ( timer_create(CLOCK_REALTIME, &evt, &timerid) ){
perror("timer_create");
}
//setup timer
its.it_value.tv_sec = 0;
its.it_value.tv_nsec = 0.1*1E9;
its.it_interval.tv_sec = 0;
its.it_interval.tv_nsec = 0;
//arm the timer
if ( timer_settime(timerid, 0, &its, NULL) )
perror("timer_settime");
sigemptyset(&blocked);
sigaddset(&blocked, SIG);
//add SIG to blocked signals
pthread_sigmask(SIG_BLOCK, &blocked, NULL);
sa.sa_flags = SA_SIGINFO; //use this flag to set custom handler
sa.sa_sigaction = handler;
sigemptyset(&sa.sa_mask);
install_handler(SIG,sa);
while ( sigwaitinfo(&blocked, &si) == -1 && errno == EINTR );
printf("received signal: %s, flag=%d\n",strsignal(si.si_signo),flag);
//while(flag==0) sleep(1); //use this when only signal handler is used
timer_delete(timerid);
return 0;
}
I am doing this mostly for educational purposes, since I need to learn as much as possible about how threads are sent/blocked as I will be using them in threads.
It's not possible because sigwaitinfo() removes the signal from the queue.
You can, however, use sigaction(SIG, NULL, &sa) to retrieve the sigaction struct of this signal and execute the handler.