Quellcodebibliothek Statistik Leitseite products/sources/formale Sprachen/C/Linux/tools/testing/selftests/mqueue/   (Open Source Betriebssystem Version 6.17.9©)  Datei vom 24.10.2025 mit Größe 21 kB image not shown  

Quelle  mq_perf_tests.c   Sprache: C

 
/*
 * This application is Copyright 2012 Red Hat, Inc.
 * Doug Ledford <dledford@redhat.com>
 *
 * mq_perf_tests is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, version 3.
 *
 * mq_perf_tests is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * For the full text of the license, see <http://www.gnu.org/licenses/>.
 *
 * mq_perf_tests.c
 *   Tests various types of message queue workloads, concentrating on those
 *   situations that invole large message sizes, large message queue depths,
 *   or both, and reports back useful metrics about kernel message queue
 *   performance.
 *
 */

#define _GNU_SOURCE
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <fcntl.h>
#include <string.h>
#include <limits.h>
#include <errno.h>
#include <signal.h>
#include <pthread.h>
#include <sched.h>
#include <sys/types.h>
#include <sys/time.h>
#include <sys/resource.h>
#include <sys/stat.h>
#include <sys/param.h>
#include <mqueue.h>
#include <popt.h>
#include <error.h>

#include "../kselftest.h"

static char *usage =
"Usage:\n"
" %s [-c #[,#..] -f] path\n"
"\n"
" -c # Skip most tests and go straight to a high queue depth test\n"
" and then run that test continuously (useful for running at\n"
" the same time as some other workload to see how much the\n"
" cache thrashing caused by adding messages to a very deep\n"
" queue impacts the performance of other programs). The number\n"
" indicates which CPU core we should bind the process to during\n"
" the run. If you have more than one physical CPU, then you\n"
" will need one copy per physical CPU package, and you should\n"
" specify the CPU cores to pin ourself to via a comma separated\n"
" list of CPU values.\n"
" -f Only usable with continuous mode. Pin ourself to the CPUs\n"
" as requested, then instead of looping doing a high mq\n"
" workload, just busy loop. This will allow us to lock up a\n"
" single CPU just like we normally would, but without actually\n"
" thrashing the CPU cache. This is to make it easier to get\n"
" comparable numbers from some other workload running on the\n"
" other CPUs. One set of numbers with # CPUs locked up running\n"
" an mq workload, and another set of numbers with those same\n"
" CPUs locked away from the test workload, but not doing\n"
" anything to trash the cache like the mq workload might.\n"
" path Path name of the message queue to create\n"
"\n"
" Note: this program must be run as root in order to enable all tests\n"
"\n";

char *MAX_MSGS = "/proc/sys/fs/mqueue/msg_max";
char *MAX_MSGSIZE = "/proc/sys/fs/mqueue/msgsize_max";

#define MAX_CPUS 64
char *cpu_option_string;
int cpus_to_pin[MAX_CPUS];
int num_cpus_to_pin;
pthread_t cpu_threads[MAX_CPUS];
pthread_t main_thread;
cpu_set_t *cpu_set;
int cpu_set_size;
int cpus_online;

#define MSG_SIZE 16
#define TEST1_LOOPS 10000000
#define TEST2_LOOPS 100000
int continuous_mode;
int continuous_mode_fake;

struct rlimit saved_limits, cur_limits;
int saved_max_msgs, saved_max_msgsize;
int cur_max_msgs, cur_max_msgsize;
FILE *max_msgs, *max_msgsize;
int cur_nice;
char *queue_path = "/mq_perf_tests";
mqd_t queue = -1;
struct mq_attr result;
int mq_prio_max;

const struct poptOption options[] = {
 {
  .longName = "continuous",
  .shortName = 'c',
  .argInfo = POPT_ARG_STRING,
  .arg = &cpu_option_string,
  .val = 'c',
  .descrip = "Run continuous tests at a high queue depth in "
   "order to test the effects of cache thrashing on "
   "other tasks on the system. This test is intended "
   "to be run on one core of each physical CPU while "
   "some other CPU intensive task is run on all the other "
   "cores of that same physical CPU and the other task "
   "is timed. It is assumed that the process of adding "
   "messages to the message queue in a tight loop will "
   "impact that other task to some degree. Once the "
   "tests are performed in this way, you should then "
   "re-run the tests using fake mode in order to check "
   "the difference in time required to perform the CPU "
   "intensive task",
  .argDescrip = "cpu[,cpu]",
 },
 {
  .longName = "fake",
  .shortName = 'f',
  .argInfo = POPT_ARG_NONE,
  .arg = &continuous_mode_fake,
  .val = 0,
  .descrip = "Tie up the CPUs that we would normally tie up in"
   "continuous mode, but don't actually do any mq stuff, "
   "just keep the CPU busy so it can't be used to process "
   "system level tasks as this would free up resources on "
   "the other CPU cores and skew the comparison between "
   "the no-mqueue work and mqueue work tests",
  .argDescrip = NULL,
 },
 {
  .longName = "path",
  .shortName = 'p',
  .argInfo = POPT_ARG_STRING | POPT_ARGFLAG_SHOW_DEFAULT,
  .arg = &queue_path,
  .val = 'p',
  .descrip = "The name of the path to use in the mqueue "
   "filesystem for our tests",
  .argDescrip = "pathname",
 },
 POPT_AUTOHELP
 POPT_TABLEEND
};

static inline void __set(FILE *stream, int value, char *err_msg);
void shutdown(int exit_val, char *err_cause, int line_no);
void sig_action_SIGUSR1(int signum, siginfo_t *info, void *context);
void sig_action(int signum, siginfo_t *info, void *context);
static inline int get(FILE *stream);
static inline void set(FILE *stream, int value);
static inline int try_set(FILE *stream, int value);
static inline void getr(int type, struct rlimit *rlim);
static inline void setr(int type, struct rlimit *rlim);
static inline void open_queue(struct mq_attr *attr);
void increase_limits(void);

static inline void __set(FILE *stream, int value, char *err_msg)
{
 rewind(stream);
 if (fprintf(stream, "%d", value) < 0)
  perror(err_msg);
}


void shutdown(int exit_val, char *err_cause, int line_no)
{
 static int in_shutdown = 0;
 int errno_at_shutdown = errno;
 int i;

 /* In case we get called by multiple threads or from an sighandler */
 if (in_shutdown++)
  return;

 /* Free the cpu_set allocated using CPU_ALLOC in main function */
 CPU_FREE(cpu_set);

 for (i = 0; i < num_cpus_to_pin; i++)
  if (cpu_threads[i]) {
   pthread_kill(cpu_threads[i], SIGUSR1);
   pthread_join(cpu_threads[i], NULL);
  }

 if (queue != -1)
  if (mq_close(queue))
   perror("mq_close() during shutdown");
 if (queue_path)
  /*
 * Be silent if this fails, if we cleaned up already it's
 * expected to fail
 */

  mq_unlink(queue_path);
 if (saved_max_msgs)
  __set(max_msgs, saved_max_msgs,
        "failed to restore saved_max_msgs");
 if (saved_max_msgsize)
  __set(max_msgsize, saved_max_msgsize,
        "failed to restore saved_max_msgsize");
 if (exit_val)
  error(exit_val, errno_at_shutdown, "%s at %d",
        err_cause, line_no);
 exit(0);
}

void sig_action_SIGUSR1(int signum, siginfo_t *info, void *context)
{
 if (pthread_self() != main_thread)
  pthread_exit(0);
 else {
  fprintf(stderr, "Caught signal %d in SIGUSR1 handler, "
    "exiting\n", signum);
  shutdown(0, "", 0);
  fprintf(stderr, "\n\nReturned from shutdown?!?!\n\n");
  exit(0);
 }
}

void sig_action(int signum, siginfo_t *info, void *context)
{
 if (pthread_self() != main_thread)
  pthread_kill(main_thread, signum);
 else {
  fprintf(stderr, "Caught signal %d, exiting\n", signum);
  shutdown(0, "", 0);
  fprintf(stderr, "\n\nReturned from shutdown?!?!\n\n");
  exit(0);
 }
}

static inline int get(FILE *stream)
{
 int value;
 rewind(stream);
 if (fscanf(stream, "%d", &value) != 1)
  shutdown(4, "Error reading /proc entry", __LINE__);
 return value;
}

static inline void set(FILE *stream, int value)
{
 int new_value;

 rewind(stream);
 if (fprintf(stream, "%d", value) < 0)
  return shutdown(5, "Failed writing to /proc file", __LINE__);
 new_value = get(stream);
 if (new_value != value)
  return shutdown(5, "We didn't get what we wrote to /proc back",
    __LINE__);
}

static inline int try_set(FILE *stream, int value)
{
 int new_value;

 rewind(stream);
 fprintf(stream, "%d", value);
 new_value = get(stream);
 return new_value == value;
}

static inline void getr(int type, struct rlimit *rlim)
{
 if (getrlimit(type, rlim))
  shutdown(6, "getrlimit()", __LINE__);
}

static inline void setr(int type, struct rlimit *rlim)
{
 if (setrlimit(type, rlim))
  shutdown(7, "setrlimit()", __LINE__);
}

/**
 * open_queue - open the global queue for testing
 * @attr - An attr struct specifying the desired queue traits
 * @result - An attr struct that lists the actual traits the queue has
 *
 * This open is not allowed to fail, failure will result in an orderly
 * shutdown of the program.  The global queue_path is used to set what
 * queue to open, the queue descriptor is saved in the global queue
 * variable.
 */

static inline void open_queue(struct mq_attr *attr)
{
 int flags = O_RDWR | O_EXCL | O_CREAT | O_NONBLOCK;
 int perms = DEFFILEMODE;

 queue = mq_open(queue_path, flags, perms, attr);
 if (queue == -1)
  shutdown(1, "mq_open()", __LINE__);
 if (mq_getattr(queue, &result))
  shutdown(1, "mq_getattr()", __LINE__);
 printf("\n\tQueue %s created:\n", queue_path);
 printf("\t\tmq_flags:\t\t\t%s\n", result.mq_flags & O_NONBLOCK ?
        "O_NONBLOCK" : "(null)");
 printf("\t\tmq_maxmsg:\t\t\t%lu\n", result.mq_maxmsg);
 printf("\t\tmq_msgsize:\t\t\t%lu\n", result.mq_msgsize);
 printf("\t\tmq_curmsgs:\t\t\t%lu\n", result.mq_curmsgs);
}

void *fake_cont_thread(void *arg)
{
 int i;

 for (i = 0; i < num_cpus_to_pin; i++)
  if (cpu_threads[i] == pthread_self())
   break;
 printf("\tStarted fake continuous mode thread %d on CPU %d\n", i,
        cpus_to_pin[i]);
 while (1)
  ;
}

void *cont_thread(void *arg)
{
 char buff[MSG_SIZE];
 int i;
 unsigned int priority;

 for (i = 0; i < num_cpus_to_pin; i++)
  if (cpu_threads[i] == pthread_self())
   break;
 printf("\tStarted continuous mode thread %d on CPU %d\n", i,
        cpus_to_pin[i]);
 while (1) {
  while (mq_send(queue, buff, sizeof(buff), 0) == 0)
   ;
  mq_receive(queue, buff, sizeof(buff), &priority);
 }
}

#define drain_queue() \
 while (mq_receive(queue, buff, MSG_SIZE, &prio_in) == MSG_SIZE)

#define do_untimed_send() \
 do { \
  if (mq_send(queue, buff, MSG_SIZE, prio_out)) \
   shutdown(3, "Test send failure", __LINE__); \
 } while (0)

#define do_send_recv() \
 do { \
  clock_gettime(clock, &start); \
  if (mq_send(queue, buff, MSG_SIZE, prio_out)) \
   shutdown(3, "Test send failure", __LINE__); \
  clock_gettime(clock, &middle); \
  if (mq_receive(queue, buff, MSG_SIZE, &prio_in) != MSG_SIZE) \
   shutdown(3, "Test receive failure", __LINE__); \
  clock_gettime(clock, &end); \
  nsec = ((middle.tv_sec - start.tv_sec) * 1000000000) + \
   (middle.tv_nsec - start.tv_nsec); \
  send_total.tv_nsec += nsec; \
  if (send_total.tv_nsec >= 1000000000) { \
   send_total.tv_sec++; \
   send_total.tv_nsec -= 1000000000; \
  } \
  nsec = ((end.tv_sec - middle.tv_sec) * 1000000000) + \
   (end.tv_nsec - middle.tv_nsec); \
  recv_total.tv_nsec += nsec; \
  if (recv_total.tv_nsec >= 1000000000) { \
   recv_total.tv_sec++; \
   recv_total.tv_nsec -= 1000000000; \
  } \
 } while (0)

struct test {
 char *desc;
 void (*func)(int *);
};

void const_prio(int *prio)
{
 return;
}

void inc_prio(int *prio)
{
 if (++*prio == mq_prio_max)
  *prio = 0;
}

void dec_prio(int *prio)
{
 if (--*prio < 0)
  *prio = mq_prio_max - 1;
}

void random_prio(int *prio)
{
 *prio = random() % mq_prio_max;
}

struct test test2[] = {
 {"\n\tTest #2a: Time send/recv message, queue full, constant prio\n",
  const_prio},
 {"\n\tTest #2b: Time send/recv message, queue full, increasing prio\n",
  inc_prio},
 {"\n\tTest #2c: Time send/recv message, queue full, decreasing prio\n",
  dec_prio},
 {"\n\tTest #2d: Time send/recv message, queue full, random prio\n",
  random_prio},
 {NULL, NULL}
};

/**
 * Tests to perform (all done with MSG_SIZE messages):
 *
 * 1) Time to add/remove message with 0 messages on queue
 * 1a) with constant prio
 * 2) Time to add/remove message when queue close to capacity:
 * 2a) with constant prio
 * 2b) with increasing prio
 * 2c) with decreasing prio
 * 2d) with random prio
 * 3) Test limits of priorities honored (double check _SC_MQ_PRIO_MAX)
 */

void *perf_test_thread(void *arg)
{
 char buff[MSG_SIZE];
 int prio_out;
 unsigned int prio_in;
 int i;
 clockid_t clock;
 pthread_t *t;
 struct timespec res, start, middle, end, send_total, recv_total;
 unsigned long long nsec;
 struct test *cur_test;

 t = &cpu_threads[0];
 printf("\n\tStarted mqueue performance test thread on CPU %d\n",
        cpus_to_pin[0]);
 mq_prio_max = sysconf(_SC_MQ_PRIO_MAX);
 if (mq_prio_max == -1)
  shutdown(2, "sysconf(_SC_MQ_PRIO_MAX)", __LINE__);
 if (pthread_getcpuclockid(cpu_threads[0], &clock) != 0)
  shutdown(2, "pthread_getcpuclockid", __LINE__);

 if (clock_getres(clock, &res))
  shutdown(2, "clock_getres()", __LINE__);

 printf("\t\tMax priorities:\t\t\t%d\n", mq_prio_max);
 printf("\t\tClock resolution:\t\t%lu nsec%s\n", res.tv_nsec,
        res.tv_nsec > 1 ? "s" : "");



 printf("\n\tTest #1: Time send/recv message, queue empty\n");
 printf("\t\t(%d iterations)\n", TEST1_LOOPS);
 prio_out = 0;
 send_total.tv_sec = 0;
 send_total.tv_nsec = 0;
 recv_total.tv_sec = 0;
 recv_total.tv_nsec = 0;
 for (i = 0; i < TEST1_LOOPS; i++)
  do_send_recv();
 printf("\t\tSend msg:\t\t\t%ld.%lus total time\n",
        send_total.tv_sec, send_total.tv_nsec);
 nsec = ((unsigned long long)send_total.tv_sec * 1000000000 +
   send_total.tv_nsec) / TEST1_LOOPS;
 printf("\t\t\t\t\t\t%lld nsec/msg\n", nsec);
 printf("\t\tRecv msg:\t\t\t%ld.%lus total time\n",
        recv_total.tv_sec, recv_total.tv_nsec);
 nsec = ((unsigned long long)recv_total.tv_sec * 1000000000 +
  recv_total.tv_nsec) / TEST1_LOOPS;
 printf("\t\t\t\t\t\t%lld nsec/msg\n", nsec);


 for (cur_test = test2; cur_test->desc != NULL; cur_test++) {
  printf("%s:\n", cur_test->desc);
  printf("\t\t(%d iterations)\n", TEST2_LOOPS);
  prio_out = 0;
  send_total.tv_sec = 0;
  send_total.tv_nsec = 0;
  recv_total.tv_sec = 0;
  recv_total.tv_nsec = 0;
  printf("\t\tFilling queue...");
  fflush(stdout);
  clock_gettime(clock, &start);
  for (i = 0; i < result.mq_maxmsg - 1; i++) {
   do_untimed_send();
   cur_test->func(&prio_out);
  }
  clock_gettime(clock, &end);
  nsec = ((unsigned long long)(end.tv_sec - start.tv_sec) *
   1000000000) + (end.tv_nsec - start.tv_nsec);
  printf("done.\t\t%lld.%llds\n", nsec / 1000000000,
         nsec % 1000000000);
  printf("\t\tTesting...");
  fflush(stdout);
  for (i = 0; i < TEST2_LOOPS; i++) {
   do_send_recv();
   cur_test->func(&prio_out);
  }
  printf("done.\n");
  printf("\t\tSend msg:\t\t\t%ld.%lus total time\n",
         send_total.tv_sec, send_total.tv_nsec);
  nsec = ((unsigned long long)send_total.tv_sec * 1000000000 +
    send_total.tv_nsec) / TEST2_LOOPS;
  printf("\t\t\t\t\t\t%lld nsec/msg\n", nsec);
  printf("\t\tRecv msg:\t\t\t%ld.%lus total time\n",
         recv_total.tv_sec, recv_total.tv_nsec);
  nsec = ((unsigned long long)recv_total.tv_sec * 1000000000 +
   recv_total.tv_nsec) / TEST2_LOOPS;
  printf("\t\t\t\t\t\t%lld nsec/msg\n", nsec);
  printf("\t\tDraining queue...");
  fflush(stdout);
  clock_gettime(clock, &start);
  drain_queue();
  clock_gettime(clock, &end);
  nsec = ((unsigned long long)(end.tv_sec - start.tv_sec) *
   1000000000) + (end.tv_nsec - start.tv_nsec);
  printf("done.\t\t%lld.%llds\n", nsec / 1000000000,
         nsec % 1000000000);
 }
 return 0;
}

void increase_limits(void)
{
 cur_limits.rlim_cur = RLIM_INFINITY;
 cur_limits.rlim_max = RLIM_INFINITY;
 setr(RLIMIT_MSGQUEUE, &cur_limits);
 while (try_set(max_msgs, cur_max_msgs += 10))
  ;
 cur_max_msgs = get(max_msgs);
 while (try_set(max_msgsize, cur_max_msgsize += 1024))
  ;
 cur_max_msgsize = get(max_msgsize);
 if (setpriority(PRIO_PROCESS, 0, -20) != 0)
  shutdown(2, "setpriority()", __LINE__);
 cur_nice = -20;
}

int main(int argc, char *argv[])
{
 struct mq_attr attr;
 char *option, *next_option;
 int i, cpu, rc;
 struct sigaction sa;
 poptContext popt_context;
 void *retval;

 main_thread = pthread_self();
 num_cpus_to_pin = 0;

 if (sysconf(_SC_NPROCESSORS_ONLN) == -1) {
  perror("sysconf(_SC_NPROCESSORS_ONLN)");
  exit(1);
 }

 if (getuid() != 0)
  ksft_exit_skip("Not running as root, but almost all tests "
   "require root in order to modify\nsystem settings. "
   "Exiting.\n");

 cpus_online = MIN(MAX_CPUS, sysconf(_SC_NPROCESSORS_ONLN));
 cpu_set = CPU_ALLOC(cpus_online);
 if (cpu_set == NULL) {
  perror("CPU_ALLOC()");
  exit(1);
 }
 cpu_set_size = CPU_ALLOC_SIZE(cpus_online);
 CPU_ZERO_S(cpu_set_size, cpu_set);

 popt_context = poptGetContext(NULL, argc, (const char **)argv,
          options, 0);

 while ((rc = poptGetNextOpt(popt_context)) > 0) {
  switch (rc) {
  case 'c':
   continuous_mode = 1;
   option = cpu_option_string;
   do {
    next_option = strchr(option, ',');
    if (next_option)
     *next_option = '\0';
    cpu = atoi(option);
    if (cpu >= cpus_online)
     fprintf(stderr, "CPU %d exceeds "
      "cpus online, ignoring.\n",
      cpu);
    else
     cpus_to_pin[num_cpus_to_pin++] = cpu;
    if (next_option)
     option = ++next_option;
   } while (next_option && num_cpus_to_pin < MAX_CPUS);
   /* Double check that they didn't give us the same CPU
 * more than once */

   for (cpu = 0; cpu < num_cpus_to_pin; cpu++) {
    if (CPU_ISSET_S(cpus_to_pin[cpu], cpu_set_size,
      cpu_set)) {
     fprintf(stderr, "Any given CPU may "
      "only be given once.\n");
     goto err_code;
    } else
     CPU_SET_S(cpus_to_pin[cpu],
        cpu_set_size, cpu_set);
   }
   break;
  case 'p':
   /*
 * Although we can create a msg queue with a
 * non-absolute path name, unlink will fail.  So,
 * if the name doesn't start with a /, add one
 * when we save it.
 */

   option = queue_path;
   if (*option != '/') {
    queue_path = malloc(strlen(option) + 2);
    if (!queue_path) {
     perror("malloc()");
     goto err_code;
    }
    queue_path[0] = '/';
    queue_path[1] = 0;
    strcat(queue_path, option);
    free(option);
   }
   break;
  }
 }

 if (continuous_mode && num_cpus_to_pin == 0) {
  fprintf(stderr, "Must pass at least one CPU to continuous "
   "mode.\n");
  poptPrintUsage(popt_context, stderr, 0);
  goto err_code;
 } else if (!continuous_mode) {
  num_cpus_to_pin = 1;
  cpus_to_pin[0] = cpus_online - 1;
 }

 max_msgs = fopen(MAX_MSGS, "r+");
 max_msgsize = fopen(MAX_MSGSIZE, "r+");
 if (!max_msgs)
  shutdown(2, "Failed to open msg_max", __LINE__);
 if (!max_msgsize)
  shutdown(2, "Failed to open msgsize_max", __LINE__);

 /* Load up the current system values for everything we can */
 getr(RLIMIT_MSGQUEUE, &saved_limits);
 cur_limits = saved_limits;
 saved_max_msgs = cur_max_msgs = get(max_msgs);
 saved_max_msgsize = cur_max_msgsize = get(max_msgsize);
 errno = 0;
 cur_nice = getpriority(PRIO_PROCESS, 0);
 if (errno)
  shutdown(2, "getpriority()", __LINE__);

 /* Tell the user our initial state */
 printf("\nInitial system state:\n");
 printf("\tUsing queue path:\t\t\t%s\n", queue_path);
 printf("\tRLIMIT_MSGQUEUE(soft):\t\t\t%ld\n",
  (long) saved_limits.rlim_cur);
 printf("\tRLIMIT_MSGQUEUE(hard):\t\t\t%ld\n",
  (long) saved_limits.rlim_max);
 printf("\tMaximum Message Size:\t\t\t%d\n", saved_max_msgsize);
 printf("\tMaximum Queue Size:\t\t\t%d\n", saved_max_msgs);
 printf("\tNice value:\t\t\t\t%d\n", cur_nice);
 printf("\n");

 increase_limits();

 printf("Adjusted system state for testing:\n");
 if (cur_limits.rlim_cur == RLIM_INFINITY) {
  printf("\tRLIMIT_MSGQUEUE(soft):\t\t\t(unlimited)\n");
  printf("\tRLIMIT_MSGQUEUE(hard):\t\t\t(unlimited)\n");
 } else {
  printf("\tRLIMIT_MSGQUEUE(soft):\t\t\t%ld\n",
         (long) cur_limits.rlim_cur);
  printf("\tRLIMIT_MSGQUEUE(hard):\t\t\t%ld\n",
         (long) cur_limits.rlim_max);
 }
 printf("\tMaximum Message Size:\t\t\t%d\n", cur_max_msgsize);
 printf("\tMaximum Queue Size:\t\t\t%d\n", cur_max_msgs);
 printf("\tNice value:\t\t\t\t%d\n", cur_nice);
 printf("\tContinuous mode:\t\t\t(%s)\n", continuous_mode ?
        (continuous_mode_fake ? "fake mode" : "enabled") :
        "disabled");
 printf("\tCPUs to pin:\t\t\t\t%d", cpus_to_pin[0]);
 for (cpu = 1; cpu < num_cpus_to_pin; cpu++)
   printf(",%d", cpus_to_pin[cpu]);
 printf("\n");

 sa.sa_sigaction = sig_action_SIGUSR1;
 sigemptyset(&sa.sa_mask);
 sigaddset(&sa.sa_mask, SIGHUP);
 sigaddset(&sa.sa_mask, SIGINT);
 sigaddset(&sa.sa_mask, SIGQUIT);
 sigaddset(&sa.sa_mask, SIGTERM);
 sa.sa_flags = SA_SIGINFO;
 if (sigaction(SIGUSR1, &sa, NULL) == -1)
  shutdown(1, "sigaction(SIGUSR1)", __LINE__);
 sa.sa_sigaction = sig_action;
 if (sigaction(SIGHUP, &sa, NULL) == -1)
  shutdown(1, "sigaction(SIGHUP)", __LINE__);
 if (sigaction(SIGINT, &sa, NULL) == -1)
  shutdown(1, "sigaction(SIGINT)", __LINE__);
 if (sigaction(SIGQUIT, &sa, NULL) == -1)
  shutdown(1, "sigaction(SIGQUIT)", __LINE__);
 if (sigaction(SIGTERM, &sa, NULL) == -1)
  shutdown(1, "sigaction(SIGTERM)", __LINE__);

 if (!continuous_mode_fake) {
  attr.mq_flags = O_NONBLOCK;
  attr.mq_maxmsg = cur_max_msgs;
  attr.mq_msgsize = MSG_SIZE;
  open_queue(&attr);
 }
 for (i = 0; i < num_cpus_to_pin; i++) {
  pthread_attr_t thread_attr;
  void *thread_func;

  if (continuous_mode_fake)
   thread_func = &fake_cont_thread;
  else if (continuous_mode)
   thread_func = &cont_thread;
  else
   thread_func = &perf_test_thread;

  CPU_ZERO_S(cpu_set_size, cpu_set);
  CPU_SET_S(cpus_to_pin[i], cpu_set_size, cpu_set);
  pthread_attr_init(&thread_attr);
  pthread_attr_setaffinity_np(&thread_attr, cpu_set_size,
         cpu_set);
  if (pthread_create(&cpu_threads[i], &thread_attr, thread_func,
       NULL))
   shutdown(1, "pthread_create()", __LINE__);
  pthread_attr_destroy(&thread_attr);
 }

 if (!continuous_mode) {
  pthread_join(cpu_threads[0], &retval);
  shutdown((long)retval, "perf_test_thread()", __LINE__);
 } else {
  while (1)
   sleep(1);
 }
 shutdown(0, "", 0);

err_code:
 CPU_FREE(cpu_set);
 exit(1);

}

Messung V0.5
C=94 H=89 G=91

¤ Dauer der Verarbeitung: 0.6 Sekunden  ¤

*© Formatika GbR, Deutschland






Wurzel

Suchen

Beweissystem der NASA

Beweissystem Isabelle

NIST Cobol Testsuite

Cephes Mathematical Library

Wiener Entwicklungsmethode

Haftungshinweis

Die Informationen auf dieser Webseite wurden nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit, noch Qualität der bereit gestellten Informationen zugesichert.

Bemerkung:

Die farbliche Syntaxdarstellung und die Messung sind noch experimentell.