当两个线程拥有相同优先级时,linux c的线程调度策略问题

        当主线程和A,B优先级相同时,结果为aaaaabbbbb ,这个实在无法理解,我觉得应该也是bbbbbaaaaa         当主线程的优先级比另外两个高时, 结果为bbbbbaaaaa, 这个好理解,man手册是这样说的:          A call to sched_setscheduler(2), sched_setparam(2), or sched_setattr(2) will  put the SCHED_FIFO (or SCHED_RR) thread identified by pid at the start of the  list if it was runnable. As a consequence, it may preempt the currently running  thread if it has the same priority. (POSIX.1 specifies that the thread should go to the end of the list.)

/* critical.c
 *
 * compile with  gcc critical.c -o critical -lrt -lpthread
 *
 *  当主线程和A,B优先级相同时,结果为aaaaabbbbb
 *  当主线程的优先级比另外两个高时, 结果为bbbbbaaaaa 
 */

#include <pthread.h>    /* header file for pthreads */
#include <unistd.h>     /* header file for POSIX conformance */
#include <time.h>       /* header file for POSIX time management */
#include <sched.h>      /* header file for POSIX scheduling */
#include <stdio.h>      /* header file for standard input/outputlibrary */
#define _REENTRANT      /* macro to ensure system calls are reentrant */

void *threadA(void *);   /* predefine threadA routine */
void *threadB(void *);   /* predefine threadB routine */

pthread_t threadA_id,threadB_id,main_id;   /* thread identifiers */
pthread_attr_t attrA,attrB;        /* thread attribute structures */
struct sched_param param;          /* scheduling structure for thread attributes */

int policy=SCHED_FIFO;
int priority_min,priority_max;     /* for range of priority levels */

/* main routine */
int main()
{
  struct timespec start;
  int status;                        /* check that system calls return ok */

  clock_gettime(CLOCK_REALTIME, &start);        /* get the time   */
  printf("Start time is: %d seconds %d nano_seconds\n",start.tv_sec,start.tv_nsec);

  /* Set processor affinity */
  unsigned long mask = 1; /* use only 1 CPU core */
  unsigned int len = sizeof(mask);
  status = sched_setaffinity(0, len, &mask);
  if (status < 0) perror("sched_setaffinity");
  status = sched_getaffinity(0, len, &mask);
  if (status < 0) perror("sched_getaffinity");

  /* Find priority limits */

  priority_max = sched_get_priority_max(policy);
  priority_min = sched_get_priority_min(policy);

  /* Change priority and policy of main thread */

  main_id = pthread_self();
  param.sched_priority=priority_min;
  status = pthread_setschedparam(main_id, policy, &param);
  if (status != 0) perror("pthread_setschedparam"); /* error check */

  /* Create threadA */

  param.sched_priority = priority_min;
  pthread_attr_init(&attrA);
  status = pthread_attr_setschedpolicy(&attrA,policy);
  if (status != 0) perror("pthread_attr_setschedpolicy"); /* error check */
  status = pthread_attr_setschedparam(&attrA,&param);
  if (status != 0) perror("pthread_attr_setschedparam"); /* error check */
  status = pthread_create(&threadA_id, &attrA, threadA, NULL);
  if (status != 0) perror("pthread_create"); /* error check */
  status = pthread_setschedparam(threadA_id,policy,&param);
  if (status != 0) perror("pthread_setschedparam");

  /* Create threadB */
  param.sched_priority = priority_min; /* so that B runs with a higher priority than A */
  pthread_attr_init(&attrB);
  status = pthread_attr_setschedpolicy(&attrB,policy);
  if (status != 0) perror("pthread_attr_setschedpolicy"); /* error check */
  status = pthread_attr_setschedparam(&attrB,&param);
  if (status != 0) perror("pthread_attr_setschedparam"); /* error check */
  status = pthread_create(&threadB_id, &attrB, threadB, NULL);
  if (status != 0) perror("pthread_create"); /* error check */
  status = pthread_setschedparam(threadB_id,policy,&param);
  if (status != 0) perror("pthread_setschedparam");

  /* Join threads - force main to wait for the thread to terminate */
  printf("main() waiting for threads\n");

  status = pthread_join(threadA_id, NULL);
  if (status != 0) perror("pthread_join(threadA_id, NULL)"); /* error check */
  status = pthread_join(threadB_id, NULL);
  if (status != 0) perror("pthread_join(threadB_id, NULL)"); /* error check */

  printf("\nmain() reporting that all threads have terminated\n");
  return(0);

}  /* end of main */

void *threadA(void *arg)
{
  int j;
  for(j=1;j<=5;j++){
    printf("a");
  }
  return (NULL);
}

void *threadB(void *arg)
{
  int j;
  for(j=1;j<=5;j++){
    printf("b");
  }
  return (NULL);
}

fish - Hadooper

赞同来自:

优先级相同时为何aaaaabbbbb 不能理解?

fish - Hadooper

赞同来自:

打印的内容很少,在B执行setschedparam(想达到插队目的)之前,是不是A都已经执行完了?

要回复问题请先登录注册