Thread priority with posix - c

Question

I am trying to play with the priority with POSIX. My goal is to have the following program. Two thread are running Thread0 with a priority of 10 and Thread1 with the priority of 50. The Thread1 block in a finite loop (like 3 seconds) and in this time interval Thread0 try to execute himself. The result should be the the Thread0 will be blocked because a thread with higher priority is executing.
My result is that the priority doesn't change the behaviour of the thread... I compile with the following command gcc -Wall -o scheduling scheduling3.c -pthread
and with the sudo su command on ubuntu.
Result :
Prio min = 1, Prio max = 99
SCHED_FIFO
Priority of the thread 0 : 10
SCHED_FIFO
Priority of the thread 1 : 50
Value of test_thread_0 1
Value of test_thread_1 1
Result that I want :
Prio min = 1, Prio max = 99
SCHED_FIFO
Priority of the thread 0 : 10
SCHED_FIFO
Priority of the thread 1 : 50
Value of test_thread_0 1
Value of test_thread_1 1
Program :
// The thread with high priority wil block the thread with low priority
// Run with super user privilege (sudo su with ubuntu)
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <sys/time.h>
#include <time.h>
#include <pthread.h>
#include <errno.h>
#define NUM_THREADS 2
int test_thread_0 = 0;
int test_thread_1 = 0;
void *BlockThread0(void *threadid) {
test_thread_0 = 1;
}
void *BlockThread1(void *threadid) {
struct timeval start, end;
long secs_used;
gettimeofday(&start, NULL);
test_thread_1 = 1;
while(1) {
gettimeofday(&end, NULL);
secs_used=(end.tv_sec - start.tv_sec); //avoid overflow by subtracting first
if(secs_used > 3)
break;
}
}
int main(int argc, char *argv[]) {
int i, policy;
pthread_t tid[NUM_THREADS];
pthread_attr_t attr[NUM_THREADS];
struct sched_param param[NUM_THREADS], test;
int prio_max, prio_min;
// Get the range of the policy
prio_max = sched_get_priority_max(SCHED_FIFO);
prio_min = sched_get_priority_min(SCHED_FIFO);
printf("Prio min = %d, Prio max = %d \n", prio_min, prio_max);
// Set the different priority
param[0].sched_priority = 10;
param[1].sched_priority = 50;
// Set all the attribute (policy + priority)
for(i = 0; i < NUM_THREADS; i++) {
pthread_attr_init(&attr[i]);
if(pthread_attr_setinheritsched(&attr[i], PTHREAD_EXPLICIT_SCHED) != 0)
fprintf(stderr, "Unable to set EXPLICIT SCHEDULER.\n");
pthread_attr_setdetachstate(&attr[i], PTHREAD_CREATE_JOINABLE);
/* The attribute get the new policy */
if(pthread_attr_setschedpolicy(&attr[i], SCHED_FIFO) != 0)
fprintf(stderr, "Unable to set policy.\n");
/* Test to change the priority of each task */
if(pthread_attr_setschedparam(&attr[i], &param[i]) != 0)
fprintf(stderr, "Unable to set priority.\n");
}
// Get all the attribute (policy + priority)
for(i = 0; i < NUM_THREADS; i++) {
// Get the policy
if(pthread_attr_getschedpolicy(&attr[i], &policy) != 0)
fprintf(stderr, "Unable to get policy.\n");
else{
if(policy == SCHED_OTHER)
printf("SCHED_OTHER\n");
else if(policy == SCHED_RR)
printf("SCHED_RR\n");
else if(policy == SCHED_FIFO)
printf("SCHED_FIFO\n");
}
/* Get the priority */
pthread_attr_getschedparam(&attr[i], &test);
printf("Priority of the thread %d : %d \n",i,test.sched_priority);
}
// Thread1 with the most important priority is executing
pthread_create(&tid[1], &attr[1], BlockThread1, (void *)1);
// To be sure that the thread1 is running
sleep(1);
//Thread2 with lower priority attempt to execute himself but he is blocked because thread1 is executing
pthread_create(&tid[0], &attr[0], BlockThread0, (void *)0);
/* now join on each thread */
for(i = 0; i < NUM_THREADS; i++)
pthread_join(tid[i], NULL);
printf("Value of test_thread_0 %d \n",test_thread_0);
printf("Value of test_thread_1 %d \n",test_thread_1);
}

Answer 1

That loop is called a "busy loop", it does not give the OS' scheduler a way to schedule other threads. You should use some variant of sleep().

Answer 2

'The result should be the the Thread0 will be blocked because a thread with higher priority is executing'
No. Not with nearly all hardware commonly available today.
Threads will only remain ready, waiting for execution, if there are more ready/running threads than cores available to run them.
With two threads, you should notice something like you were expecting on a processor with only one core.
They are very rare now.