1 Star 0 Fork 2

gaoxuelong / stress-ng

forked from HoperunHarmony / stress-ng 
加入 Gitee
与超过 1200万 开发者一起发现、参与优秀开源项目,私有仓库也完全免费 :)
免费加入
克隆/下载
stress-mq.c 13.95 KB
一键复制 编辑 原始数据 按行查看 历史
/*
* Copyright (C) 2013-2021 Canonical, Ltd.
* Copyright (C) 2022 Colin Ian King.
*
* This program 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; either version 2
* of the License, or (at your option) any later version.
*
* This program 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.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
*
*/
#include "stress-ng.h"
#if defined(HAVE_MQUEUE_H)
#include <mqueue.h>
#endif
#if defined(HAVE_POLL_H)
#include <poll.h>
#endif
#define MIN_MQ_SIZE (1)
#define MAX_MQ_SIZE (32)
#define DEFAULT_MQ_SIZE (10)
static const stress_help_t help[] = {
{ NULL, "mq N", "start N workers passing messages using POSIX messages" },
{ NULL, "mq-ops N", "stop mq workers after N bogo messages" },
{ NULL, "mq-size N", "specify the size of the POSIX message queue" },
{ NULL, NULL, NULL }
};
static int stress_set_mq_size(const char *opt)
{
uint64_t sz;
int mq_size;
sz = stress_get_uint64(opt);
stress_check_range("mq-size", sz, MIN_MQ_SIZE, MAX_MQ_SIZE);
mq_size = (int)sz;
return stress_set_setting("mq-size", TYPE_ID_INT, &mq_size);
}
static const stress_opt_set_func_t opt_set_funcs[] = {
{ OPT_mq_size, stress_set_mq_size },
{ 0, NULL }
};
#if defined(HAVE_MQUEUE_H) && \
defined(HAVE_LIB_RT) && \
defined(HAVE_MQ_POSIX)
#define PRIOS_MAX (128)
typedef struct {
uint64_t value;
} stress_msg_t;
static void stress_mq_notify_func(union sigval s)
{
(void)s;
}
#if defined(SIGUSR2)
static void MLOCKED_TEXT stress_sigusr2_handler(int signum)
{
(void)signum;
}
#endif
/*
* stress_mq_invalid_open()
* perform invalid mq open and perform tidy up
* if it somehow succeeded
*/
static void stress_mq_invalid_open(
const char *name,
int oflag,
mode_t mode,
struct mq_attr *attr)
{
mqd_t mq;
mq = mq_open(name, oflag, mode, attr);
if (mq >= 0) {
(void)mq_close(mq);
(void)mq_unlink(name);
}
}
/*
* stress_mq
* stress POSIX message queues
*/
static int stress_mq(const stress_args_t *args)
{
pid_t pid;
mqd_t mq = -1;
int sz, max_sz, mq_size = DEFAULT_MQ_SIZE;
FILE *fp;
struct mq_attr attr;
char mq_name[64], mq_tmp_name[64];
bool do_timed;
time_t time_start;
struct timespec abs_timeout;
unsigned int max_prio = UINT_MAX;
#if defined(SIGUSR2)
if (stress_sighandler(args->name, SIGUSR2, stress_sigusr2_handler, NULL) < 0)
return EXIT_NO_RESOURCE;
#endif
#if defined(_SC_MQ_PRIO_MAX)
{
long sysconf_ret;
sysconf_ret = sysconf(_SC_MQ_PRIO_MAX);
if ((sysconf_ret > 0) && (sysconf_ret < (long)UINT_MAX))
max_prio = sysconf_ret + 1;
}
#endif
if (!stress_get_setting("mq-size", &mq_size)) {
if (g_opt_flags & OPT_FLAGS_MAXIMIZE)
mq_size = MAX_MQ_SIZE;
if (g_opt_flags & OPT_FLAGS_MINIMIZE)
mq_size = MIN_MQ_SIZE;
}
sz = mq_size;
(void)snprintf(mq_tmp_name, sizeof(mq_tmp_name), "/%s-%" PRIdMAX "-%" PRIu32,
args->name, (intmax_t)args->pid, args->instance + 10000);
(void)snprintf(mq_name, sizeof(mq_name), "/%s-%" PRIdMAX "-%" PRIu32,
args->name, (intmax_t)args->pid, args->instance);
fp = fopen("/proc/sys/fs/mqueue/msg_default", "r");
if (fp) {
if (fscanf(fp, "%20d", &max_sz) != 1)
max_sz = MAX_MQ_SIZE;
(void)fclose(fp);
if (max_sz < MIN_MQ_SIZE)
max_sz = MIN_MQ_SIZE;
if (max_sz > MAX_MQ_SIZE)
max_sz = MAX_MQ_SIZE;
} else {
max_sz = MAX_MQ_SIZE;
}
if (sz > max_sz)
sz = max_sz;
/*
* Determine a workable MQ size if we can't determine it from /proc
*/
while (sz > 0) {
attr.mq_flags = 0;
attr.mq_maxmsg = sz;
attr.mq_msgsize = sizeof(stress_msg_t);
attr.mq_curmsgs = 0;
mq = mq_open(mq_name, O_CREAT | O_RDWR, S_IRUSR | S_IWUSR, &attr);
if (mq >= 0)
break;
if (errno == ENOSYS) {
if (args->instance == 0)
pr_inf_skip("%s: POSIX message queues not implemented, skipping stressor\n",
args->name);
return EXIT_NOT_IMPLEMENTED;
}
sz--;
}
if (mq < 0) {
pr_fail("%s: mq_open failed, errno=%d (%s)\n",
args->name, errno, strerror(errno));
return EXIT_FAILURE;
}
if (sz < mq_size) {
pr_inf("%s: POSIX message queue requested "
"size %d messages, maximum of %d allowed\n",
args->name, mq_size, sz);
}
pr_dbg("%s: POSIX message queue %s with %lu messages\n",
args->name, mq_name, (unsigned long)attr.mq_maxmsg);
if (time(&time_start) == ((time_t)-1)) {
do_timed = false;
pr_inf("%s: mq_timed send and receive skipped, can't get time, errno=%d (%s)\n",
args->name, errno, strerror(errno));
} else {
do_timed = true;
abs_timeout.tv_sec = time_start + (time_t)g_opt_timeout + 1;
abs_timeout.tv_nsec = 0;
}
stress_set_proc_state(args->name, STRESS_STATE_RUN);
again:
pid = fork();
if (pid < 0) {
if (stress_redo_fork(errno))
goto again;
(void)mq_close(mq);
(void)mq_unlink(mq_name);
if (!keep_stressing(args))
goto finish;
pr_fail("%s: fork failed, errno=%d (%s)\n",
args->name, errno, strerror(errno));
return EXIT_FAILURE;
} else if (pid == 0) {
struct sigevent sigev;
uint64_t values[PRIOS_MAX];
(void)setpgid(0, g_pgrp);
stress_parent_died_alarm();
(void)sched_settings_apply(true);
(void)memset(&values, 0, sizeof(values));
while (keep_stressing_flag()) {
uint64_t i;
for (i = 0; ; i++) {
stress_msg_t ALIGN64 msg;
ssize_t sret;
const uint64_t timed = (i & 1);
unsigned int prio;
if (!(i & 1023)) {
#if defined(__linux__)
char buffer[1024];
off_t off;
struct stat statbuf;
void *ptr;
int ret;
#if defined(HAVE_POLL_H)
struct pollfd fds[1];
#endif
/* On Linux, one can seek on a mq descriptor */
off = lseek(mq, 0, SEEK_SET);
(void)off;
/* Attempt a fstat too */
ret = fstat(mq, &statbuf);
(void)ret;
/* illegal mmap, should be ENODEV */
ptr = mmap(NULL, 16, PROT_READ, MAP_SHARED, mq, 0);
if (ptr != MAP_FAILED)
(void)munmap(ptr, 16);
#if defined(HAVE_POLL_H)
/* ..and poll too */
fds[0].fd = mq;
fds[0].events = POLLIN;
fds[0].revents = 0;
ret = poll(fds, 1, 0);
(void)ret;
#endif
/* Read state of queue from mq fd */
sret = read(mq, buffer, sizeof(buffer));
if (sret < 0) {
if (errno == EINTR)
break;
pr_fail("%s: mq read failed, errno=%d (%s)\n",
args->name, errno, strerror(errno));
}
#endif
(void)memset(&sigev, 0, sizeof sigev);
switch (stress_mwc8() % 5) {
case 3:
/* Illegal signal */
sigev.sigev_notify = SIGEV_SIGNAL;
sigev.sigev_signo = ~0;
break;
case 2:
/* Illegal notify event */
sigev.sigev_notify = ~0;
break;
case 1:
sigev.sigev_notify = SIGEV_NONE;
break;
case 0:
#if defined(SIGUSR2)
sigev.sigev_notify = SIGEV_SIGNAL;
sigev.sigev_signo = SIGUSR2;
break;
#else
CASE_FALLTHROUGH;
#endif
default:
sigev.sigev_notify = SIGEV_THREAD;
sigev.sigev_notify_function = stress_mq_notify_func;
sigev.sigev_notify_attributes = NULL;
break;
}
(void)mq_notify(mq, &sigev);
/* Exercise illegal mq descriptor */
(void)mq_notify(~0, &sigev);
(void)mq_notify(0, &sigev);
/* Exercise illegal mq_open name */
attr.mq_flags = 0;
attr.mq_maxmsg = sz;
attr.mq_msgsize = sizeof(stress_msg_t);
attr.mq_curmsgs = 0;
stress_mq_invalid_open("/bad/bad/bad", O_CREAT | O_RDWR, S_IRUSR | S_IWUSR, &attr);
/* Exercise illegal mq_maxmsg */
attr.mq_flags = 0;
attr.mq_maxmsg = ~0;
attr.mq_msgsize = sizeof(stress_msg_t);
attr.mq_curmsgs = 0;
stress_mq_invalid_open(mq_tmp_name, O_CREAT | O_RDWR, S_IRUSR | S_IWUSR, &attr);
attr.mq_flags = 0;
attr.mq_maxmsg = LONG_MAX;
attr.mq_msgsize = sizeof(stress_msg_t);
attr.mq_curmsgs = 0;
stress_mq_invalid_open(mq_tmp_name, O_CREAT | O_RDWR, S_IRUSR | S_IWUSR, &attr);
attr.mq_flags = 0;
attr.mq_maxmsg = 0;
attr.mq_msgsize = sizeof(stress_msg_t);
attr.mq_curmsgs = 0;
stress_mq_invalid_open(mq_tmp_name, O_CREAT | O_RDWR, S_IRUSR | S_IWUSR, &attr);
/* Exercise illegal mq_msgsize */
attr.mq_flags = 0;
attr.mq_maxmsg = sz;
attr.mq_msgsize = ~0;
attr.mq_curmsgs = 0;
stress_mq_invalid_open(mq_tmp_name, O_CREAT | O_RDWR, S_IRUSR | S_IWUSR, &attr);
attr.mq_flags = 0;
attr.mq_maxmsg = sz;
attr.mq_msgsize = LONG_MAX;
attr.mq_curmsgs = 0;
stress_mq_invalid_open(mq_tmp_name, O_CREAT | O_RDWR, S_IRUSR | S_IWUSR, &attr);
attr.mq_flags = 0;
attr.mq_maxmsg = sz;
attr.mq_msgsize = 0;
attr.mq_curmsgs = 0;
stress_mq_invalid_open(mq_tmp_name, O_CREAT | O_RDWR, S_IRUSR | S_IWUSR, &attr);
/* Exercise invalid mq_unlink */
(void)mq_unlink(mq_tmp_name);
(void)mq_unlink("/bad/bad/bad");
(void)mq_unlink("");
}
/*
* toggle between timedreceive and receive
*/
if (do_timed && (timed))
sret = mq_timedreceive(mq, (char *)&msg, sizeof(msg), &prio, &abs_timeout);
else
sret = mq_receive(mq, (char *)&msg, sizeof(msg), &prio);
if (sret < 0) {
if (errno != EINTR) {
pr_fail("%s: %s failed, errno=%d (%s)\n",
args->name,
timed ? "mq_timedreceive" : "mq_receive",
errno, strerror(errno));
}
break;
}
if (prio >= PRIOS_MAX) {
pr_fail("%s: mq_receive: unexpected priority %u, expected 0..%d\n",
args->name, prio, PRIOS_MAX - 1);
} else {
if (g_opt_flags & OPT_FLAGS_VERIFY) {
if (msg.value != values[prio]) {
pr_fail("%s: mq_receive: expected message "
"containing 0x%" PRIx64
" but received 0x%" PRIx64 " instead\n",
args->name, values[prio], msg.value);
}
values[prio]++;
}
}
if (do_timed && (timed)) {
/* Exercise mq_timedreceive on invalid mq descriptors */
(void)mq_timedreceive(-1, (char *)&msg, sizeof(msg), &prio, &abs_timeout);
(void)mq_timedreceive(0, (char *)&msg, sizeof(msg), &prio, &abs_timeout);
/* Exercise mq_timedreceive on invalid mq size */
(void)mq_timedreceive(mq, (char *)&msg, (size_t)0, &prio, &abs_timeout);
} else {
/* Exercise mq_receive on invalid mq descriptors */
(void)mq_receive(-1, (char *)&msg, sizeof(msg), &prio);
(void)mq_receive(0, (char *)&msg, sizeof(msg), &prio);
/* Exercise mq_receive on invalid mq size */
(void)mq_receive(mq, (char *)&msg, (size_t)0, &prio);
}
}
_exit(EXIT_SUCCESS);
}
} else {
int status;
int attr_count = 0;
stress_msg_t ALIGN64 msg;
uint64_t values[PRIOS_MAX];
/* Parent */
(void)setpgid(pid, g_pgrp);
(void)memset(&msg, 0, sizeof(msg));
(void)memset(&values, 0, sizeof(values));
do {
int ret;
unsigned int prio = stress_mwc8() % PRIOS_MAX;
const uint64_t timed = (msg.value & 1);
uint64_t i = 0;
if ((attr_count++ & 31) == 0) {
struct mq_attr old_attr;
if (mq_getattr(mq, &attr) < 0) {
pr_fail("%s: mq_getattr failed, errno=%d (%s)\n",
args->name, errno, strerror(errno));
} else {
(void)mq_setattr(mq, &attr, &old_attr);
}
/* Exercise invalid mq id */
if ((mq_getattr(~0, &attr) < 0) && (errno == EBADF))
(void)mq_setattr(~0, &attr, &old_attr);
/* Exercise invalid mq id */
if ((mq_getattr(0, &attr) == EBADF) && (errno == EBADF))
(void)mq_setattr(0, &attr, &old_attr);
}
msg.value = values[prio];
values[prio]++;
/*
* toggle between timedsend and send
*/
if (do_timed && (timed))
ret = mq_timedsend(mq, (char *)&msg, sizeof(msg), prio, &abs_timeout);
else
ret = mq_send(mq, (char *)&msg, sizeof(msg), prio);
if (ret < 0) {
if ((errno != EINTR) && (errno != ETIMEDOUT))
pr_fail("%s: %s failed, errno=%d (%s)\n",
args->name,
timed ? "mq_timedsend" : "mq_send",
errno, strerror(errno));
break;
}
if (!(i & 1023)) {
if (do_timed && (timed)) {
/* Exercise mq_timedsend on invalid mq descriptors */
(void)mq_timedsend(-1, (char *)&msg, sizeof(msg), prio, &abs_timeout);
(void)mq_timedsend(0, (char *)&msg, sizeof(msg), prio, &abs_timeout);
/* Exercise mq_timedsend on invalid mq size */
(void)mq_timedsend(mq, (char *)&msg, ~(size_t)0, prio, &abs_timeout);
/* Exercise mq_timedsend on invalid priority size */
(void)mq_timedsend(mq, (char *)&msg, sizeof(msg), max_prio, &abs_timeout);
} else {
/* Exercise mq_send on invalid mq descriptors */
(void)mq_send(-1, (char *)&msg, sizeof(msg), prio);
(void)mq_send(0, (char *)&msg, sizeof(msg), prio);
/* Exercise mq_send on invalid mq size */
(void)mq_send(mq, (char *)&msg, ~(size_t)0, prio);
/* Exercise mq_send on invalid priority size */
(void)mq_send(mq, (char *)&msg, sizeof(msg), max_prio);
}
(void)mq_close(-1);
}
i++;
inc_counter(args);
} while (keep_stressing(args));
(void)kill(pid, SIGKILL);
(void)shim_waitpid(pid, &status, 0);
if (mq_close(mq) < 0)
pr_fail("%s: mq_close failed, errno=%d (%s)\n",
args->name, errno, strerror(errno));
if (mq_unlink(mq_name) < 0)
pr_fail("%s: mq_unlink failed, errno=%d (%s)\n",
args->name, errno, strerror(errno));
/* Exercise invalid mq_close, already closed mq */
(void)mq_close(mq);
/* Exercise invalid mq_unlink */
(void)mq_unlink(mq_name);
/* Exercise invalid mq_unlink */
(void)mq_unlink("/");
}
finish:
stress_set_proc_state(args->name, STRESS_STATE_DEINIT);
return EXIT_SUCCESS;
}
stressor_info_t stress_mq_info = {
.stressor = stress_mq,
.class = CLASS_SCHEDULER | CLASS_OS,
.opt_set_funcs = opt_set_funcs,
.verify = VERIFY_OPTIONAL,
.help = help
};
#else
stressor_info_t stress_mq_info = {
.stressor = stress_not_implemented,
.class = CLASS_SCHEDULER | CLASS_OS,
.opt_set_funcs = opt_set_funcs,
.help = help
};
#endif
1
https://gitee.com/gaoxuelong/stress-ng.git
git@gitee.com:gaoxuelong/stress-ng.git
gaoxuelong
stress-ng
stress-ng
master

搜索帮助