linux-stable/tools/testing/selftests/timers/posix_timers.c

291 lines
5.9 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-only
/*
* Copyright (C) 2013 Red Hat, Inc., Frederic Weisbecker <fweisbec@redhat.com>
*
* Selftests for a few posix timers interface.
*
* Kernel loop code stolen from Steven Rostedt <srostedt@redhat.com>
*/
#include <sys/time.h>
#include <stdio.h>
#include <signal.h>
#include <unistd.h>
#include <time.h>
#include <pthread.h>
#include "../kselftest.h"
#define DELAY 2
#define USECS_PER_SEC 1000000
static volatile int done;
/* Busy loop in userspace to elapse ITIMER_VIRTUAL */
static void user_loop(void)
{
while (!done);
}
/*
* Try to spend as much time as possible in kernelspace
* to elapse ITIMER_PROF.
*/
static void kernel_loop(void)
{
void *addr = sbrk(0);
int err = 0;
while (!done && !err) {
err = brk(addr + 4096);
err |= brk(addr);
}
}
/*
* Sleep until ITIMER_REAL expiration.
*/
static void idle_loop(void)
{
pause();
}
static void sig_handler(int nr)
{
done = 1;
}
/*
* Check the expected timer expiration matches the GTOD elapsed delta since
* we armed the timer. Keep a 0.5 sec error margin due to various jitter.
*/
static int check_diff(struct timeval start, struct timeval end)
{
long long diff;
diff = end.tv_usec - start.tv_usec;
diff += (end.tv_sec - start.tv_sec) * USECS_PER_SEC;
if (llabs(diff - DELAY * USECS_PER_SEC) > USECS_PER_SEC / 2) {
printf("Diff too high: %lld..", diff);
return -1;
}
return 0;
}
static int check_itimer(int which)
{
const char *name;
int err;
struct timeval start, end;
struct itimerval val = {
.it_value.tv_sec = DELAY,
};
if (which == ITIMER_VIRTUAL)
name = "ITIMER_VIRTUAL";
else if (which == ITIMER_PROF)
name = "ITIMER_PROF";
else if (which == ITIMER_REAL)
name = "ITIMER_REAL";
else
return -1;
done = 0;
if (which == ITIMER_VIRTUAL)
signal(SIGVTALRM, sig_handler);
else if (which == ITIMER_PROF)
signal(SIGPROF, sig_handler);
else if (which == ITIMER_REAL)
signal(SIGALRM, sig_handler);
err = gettimeofday(&start, NULL);
if (err < 0) {
ksft_perror("Can't call gettimeofday()");
return -1;
}
err = setitimer(which, &val, NULL);
if (err < 0) {
ksft_perror("Can't set timer");
return -1;
}
if (which == ITIMER_VIRTUAL)
user_loop();
else if (which == ITIMER_PROF)
kernel_loop();
else if (which == ITIMER_REAL)
idle_loop();
err = gettimeofday(&end, NULL);
if (err < 0) {
ksft_perror("Can't call gettimeofday()");
return -1;
}
ksft_test_result(check_diff(start, end) == 0, "%s\n", name);
return 0;
}
static int check_timer_create(int which)
{
const char *type;
int err;
timer_t id;
struct timeval start, end;
struct itimerspec val = {
.it_value.tv_sec = DELAY,
};
if (which == CLOCK_THREAD_CPUTIME_ID) {
type = "thread";
} else if (which == CLOCK_PROCESS_CPUTIME_ID) {
type = "process";
} else {
ksft_print_msg("Unknown timer_create() type %d\n", which);
return -1;
}
done = 0;
err = timer_create(which, NULL, &id);
if (err < 0) {
ksft_perror("Can't create timer");
return -1;
}
signal(SIGALRM, sig_handler);
err = gettimeofday(&start, NULL);
if (err < 0) {
ksft_perror("Can't call gettimeofday()");
return -1;
}
err = timer_settime(id, 0, &val, NULL);
if (err < 0) {
ksft_perror("Can't set timer");
return -1;
}
user_loop();
err = gettimeofday(&end, NULL);
if (err < 0) {
ksft_perror("Can't call gettimeofday()");
return -1;
}
ksft_test_result(check_diff(start, end) == 0,
"timer_create() per %s\n", type);
return 0;
}
selftests/timers/posix_timers: Reimplement check_timer_distribution() check_timer_distribution() runs ten threads in a busy loop and tries to test that the kernel distributes a process posix CPU timer signal to every thread over time. There is not guarantee that this is true even after commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") because that commit only avoids waking up the sleeping process leader thread, but that has nothing to do with the actual signal delivery. As the signal is process wide the first thread which observes sigpending and wins the race to lock sighand will deliver the signal. Testing shows that this hangs on a regular base because some threads never win the race. The comment "This primarily tests that the kernel does not favour any one." is wrong. The kernel does favour a thread which hits the timer interrupt when CLOCK_PROCESS_CPUTIME_ID expires. Rewrite the test so it only checks that the group leader sleeping in join() never receives SIGALRM and the thread which burns CPU cycles receives all signals. In older kernels which do not have commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") the test-case fails immediately, the very 1st tick wakes the leader up. Otherwise it quickly succeeds after 100 ticks. CI testing wants to use newer selftest versions on stable kernels. In this case the test is guaranteed to fail. So check in the failure case whether the kernel version is less than v6.3 and skip the test result in that case. [ tglx: Massaged change log, renamed the version check helper ] Fixes: e797203fb3ba ("selftests/timers/posix_timers: Test delivery of signals across threads") Signed-off-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240409133802.GD29396@redhat.com
2024-04-09 13:38:03 +00:00
static pthread_t ctd_thread;
static volatile int ctd_count, ctd_failed;
selftests/timers/posix_timers: Reimplement check_timer_distribution() check_timer_distribution() runs ten threads in a busy loop and tries to test that the kernel distributes a process posix CPU timer signal to every thread over time. There is not guarantee that this is true even after commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") because that commit only avoids waking up the sleeping process leader thread, but that has nothing to do with the actual signal delivery. As the signal is process wide the first thread which observes sigpending and wins the race to lock sighand will deliver the signal. Testing shows that this hangs on a regular base because some threads never win the race. The comment "This primarily tests that the kernel does not favour any one." is wrong. The kernel does favour a thread which hits the timer interrupt when CLOCK_PROCESS_CPUTIME_ID expires. Rewrite the test so it only checks that the group leader sleeping in join() never receives SIGALRM and the thread which burns CPU cycles receives all signals. In older kernels which do not have commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") the test-case fails immediately, the very 1st tick wakes the leader up. Otherwise it quickly succeeds after 100 ticks. CI testing wants to use newer selftest versions on stable kernels. In this case the test is guaranteed to fail. So check in the failure case whether the kernel version is less than v6.3 and skip the test result in that case. [ tglx: Massaged change log, renamed the version check helper ] Fixes: e797203fb3ba ("selftests/timers/posix_timers: Test delivery of signals across threads") Signed-off-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240409133802.GD29396@redhat.com
2024-04-09 13:38:03 +00:00
static void ctd_sighandler(int sig)
{
selftests/timers/posix_timers: Reimplement check_timer_distribution() check_timer_distribution() runs ten threads in a busy loop and tries to test that the kernel distributes a process posix CPU timer signal to every thread over time. There is not guarantee that this is true even after commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") because that commit only avoids waking up the sleeping process leader thread, but that has nothing to do with the actual signal delivery. As the signal is process wide the first thread which observes sigpending and wins the race to lock sighand will deliver the signal. Testing shows that this hangs on a regular base because some threads never win the race. The comment "This primarily tests that the kernel does not favour any one." is wrong. The kernel does favour a thread which hits the timer interrupt when CLOCK_PROCESS_CPUTIME_ID expires. Rewrite the test so it only checks that the group leader sleeping in join() never receives SIGALRM and the thread which burns CPU cycles receives all signals. In older kernels which do not have commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") the test-case fails immediately, the very 1st tick wakes the leader up. Otherwise it quickly succeeds after 100 ticks. CI testing wants to use newer selftest versions on stable kernels. In this case the test is guaranteed to fail. So check in the failure case whether the kernel version is less than v6.3 and skip the test result in that case. [ tglx: Massaged change log, renamed the version check helper ] Fixes: e797203fb3ba ("selftests/timers/posix_timers: Test delivery of signals across threads") Signed-off-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240409133802.GD29396@redhat.com
2024-04-09 13:38:03 +00:00
if (pthread_self() != ctd_thread)
ctd_failed = 1;
ctd_count--;
}
selftests/timers/posix_timers: Reimplement check_timer_distribution() check_timer_distribution() runs ten threads in a busy loop and tries to test that the kernel distributes a process posix CPU timer signal to every thread over time. There is not guarantee that this is true even after commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") because that commit only avoids waking up the sleeping process leader thread, but that has nothing to do with the actual signal delivery. As the signal is process wide the first thread which observes sigpending and wins the race to lock sighand will deliver the signal. Testing shows that this hangs on a regular base because some threads never win the race. The comment "This primarily tests that the kernel does not favour any one." is wrong. The kernel does favour a thread which hits the timer interrupt when CLOCK_PROCESS_CPUTIME_ID expires. Rewrite the test so it only checks that the group leader sleeping in join() never receives SIGALRM and the thread which burns CPU cycles receives all signals. In older kernels which do not have commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") the test-case fails immediately, the very 1st tick wakes the leader up. Otherwise it quickly succeeds after 100 ticks. CI testing wants to use newer selftest versions on stable kernels. In this case the test is guaranteed to fail. So check in the failure case whether the kernel version is less than v6.3 and skip the test result in that case. [ tglx: Massaged change log, renamed the version check helper ] Fixes: e797203fb3ba ("selftests/timers/posix_timers: Test delivery of signals across threads") Signed-off-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240409133802.GD29396@redhat.com
2024-04-09 13:38:03 +00:00
static void *ctd_thread_func(void *arg)
{
struct itimerspec val = {
.it_value.tv_sec = 0,
.it_value.tv_nsec = 1000 * 1000,
.it_interval.tv_sec = 0,
.it_interval.tv_nsec = 1000 * 1000,
};
selftests/timers/posix_timers: Reimplement check_timer_distribution() check_timer_distribution() runs ten threads in a busy loop and tries to test that the kernel distributes a process posix CPU timer signal to every thread over time. There is not guarantee that this is true even after commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") because that commit only avoids waking up the sleeping process leader thread, but that has nothing to do with the actual signal delivery. As the signal is process wide the first thread which observes sigpending and wins the race to lock sighand will deliver the signal. Testing shows that this hangs on a regular base because some threads never win the race. The comment "This primarily tests that the kernel does not favour any one." is wrong. The kernel does favour a thread which hits the timer interrupt when CLOCK_PROCESS_CPUTIME_ID expires. Rewrite the test so it only checks that the group leader sleeping in join() never receives SIGALRM and the thread which burns CPU cycles receives all signals. In older kernels which do not have commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") the test-case fails immediately, the very 1st tick wakes the leader up. Otherwise it quickly succeeds after 100 ticks. CI testing wants to use newer selftest versions on stable kernels. In this case the test is guaranteed to fail. So check in the failure case whether the kernel version is less than v6.3 and skip the test result in that case. [ tglx: Massaged change log, renamed the version check helper ] Fixes: e797203fb3ba ("selftests/timers/posix_timers: Test delivery of signals across threads") Signed-off-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240409133802.GD29396@redhat.com
2024-04-09 13:38:03 +00:00
timer_t id;
selftests/timers/posix_timers: Reimplement check_timer_distribution() check_timer_distribution() runs ten threads in a busy loop and tries to test that the kernel distributes a process posix CPU timer signal to every thread over time. There is not guarantee that this is true even after commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") because that commit only avoids waking up the sleeping process leader thread, but that has nothing to do with the actual signal delivery. As the signal is process wide the first thread which observes sigpending and wins the race to lock sighand will deliver the signal. Testing shows that this hangs on a regular base because some threads never win the race. The comment "This primarily tests that the kernel does not favour any one." is wrong. The kernel does favour a thread which hits the timer interrupt when CLOCK_PROCESS_CPUTIME_ID expires. Rewrite the test so it only checks that the group leader sleeping in join() never receives SIGALRM and the thread which burns CPU cycles receives all signals. In older kernels which do not have commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") the test-case fails immediately, the very 1st tick wakes the leader up. Otherwise it quickly succeeds after 100 ticks. CI testing wants to use newer selftest versions on stable kernels. In this case the test is guaranteed to fail. So check in the failure case whether the kernel version is less than v6.3 and skip the test result in that case. [ tglx: Massaged change log, renamed the version check helper ] Fixes: e797203fb3ba ("selftests/timers/posix_timers: Test delivery of signals across threads") Signed-off-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240409133802.GD29396@redhat.com
2024-04-09 13:38:03 +00:00
/* 1/10 seconds to ensure the leader sleeps */
usleep(10000);
selftests/timers/posix_timers: Reimplement check_timer_distribution() check_timer_distribution() runs ten threads in a busy loop and tries to test that the kernel distributes a process posix CPU timer signal to every thread over time. There is not guarantee that this is true even after commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") because that commit only avoids waking up the sleeping process leader thread, but that has nothing to do with the actual signal delivery. As the signal is process wide the first thread which observes sigpending and wins the race to lock sighand will deliver the signal. Testing shows that this hangs on a regular base because some threads never win the race. The comment "This primarily tests that the kernel does not favour any one." is wrong. The kernel does favour a thread which hits the timer interrupt when CLOCK_PROCESS_CPUTIME_ID expires. Rewrite the test so it only checks that the group leader sleeping in join() never receives SIGALRM and the thread which burns CPU cycles receives all signals. In older kernels which do not have commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") the test-case fails immediately, the very 1st tick wakes the leader up. Otherwise it quickly succeeds after 100 ticks. CI testing wants to use newer selftest versions on stable kernels. In this case the test is guaranteed to fail. So check in the failure case whether the kernel version is less than v6.3 and skip the test result in that case. [ tglx: Massaged change log, renamed the version check helper ] Fixes: e797203fb3ba ("selftests/timers/posix_timers: Test delivery of signals across threads") Signed-off-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240409133802.GD29396@redhat.com
2024-04-09 13:38:03 +00:00
ctd_count = 100;
if (timer_create(CLOCK_PROCESS_CPUTIME_ID, NULL, &id))
return "Can't create timer\n";
if (timer_settime(id, 0, &val, NULL))
return "Can't set timer\n";
selftests/timers/posix_timers: Reimplement check_timer_distribution() check_timer_distribution() runs ten threads in a busy loop and tries to test that the kernel distributes a process posix CPU timer signal to every thread over time. There is not guarantee that this is true even after commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") because that commit only avoids waking up the sleeping process leader thread, but that has nothing to do with the actual signal delivery. As the signal is process wide the first thread which observes sigpending and wins the race to lock sighand will deliver the signal. Testing shows that this hangs on a regular base because some threads never win the race. The comment "This primarily tests that the kernel does not favour any one." is wrong. The kernel does favour a thread which hits the timer interrupt when CLOCK_PROCESS_CPUTIME_ID expires. Rewrite the test so it only checks that the group leader sleeping in join() never receives SIGALRM and the thread which burns CPU cycles receives all signals. In older kernels which do not have commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") the test-case fails immediately, the very 1st tick wakes the leader up. Otherwise it quickly succeeds after 100 ticks. CI testing wants to use newer selftest versions on stable kernels. In this case the test is guaranteed to fail. So check in the failure case whether the kernel version is less than v6.3 and skip the test result in that case. [ tglx: Massaged change log, renamed the version check helper ] Fixes: e797203fb3ba ("selftests/timers/posix_timers: Test delivery of signals across threads") Signed-off-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240409133802.GD29396@redhat.com
2024-04-09 13:38:03 +00:00
while (ctd_count > 0 && !ctd_failed)
;
selftests/timers/posix_timers: Reimplement check_timer_distribution() check_timer_distribution() runs ten threads in a busy loop and tries to test that the kernel distributes a process posix CPU timer signal to every thread over time. There is not guarantee that this is true even after commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") because that commit only avoids waking up the sleeping process leader thread, but that has nothing to do with the actual signal delivery. As the signal is process wide the first thread which observes sigpending and wins the race to lock sighand will deliver the signal. Testing shows that this hangs on a regular base because some threads never win the race. The comment "This primarily tests that the kernel does not favour any one." is wrong. The kernel does favour a thread which hits the timer interrupt when CLOCK_PROCESS_CPUTIME_ID expires. Rewrite the test so it only checks that the group leader sleeping in join() never receives SIGALRM and the thread which burns CPU cycles receives all signals. In older kernels which do not have commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") the test-case fails immediately, the very 1st tick wakes the leader up. Otherwise it quickly succeeds after 100 ticks. CI testing wants to use newer selftest versions on stable kernels. In this case the test is guaranteed to fail. So check in the failure case whether the kernel version is less than v6.3 and skip the test result in that case. [ tglx: Massaged change log, renamed the version check helper ] Fixes: e797203fb3ba ("selftests/timers/posix_timers: Test delivery of signals across threads") Signed-off-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240409133802.GD29396@redhat.com
2024-04-09 13:38:03 +00:00
if (timer_delete(id))
return "Can't delete timer\n";
selftests/timers/posix_timers: Reimplement check_timer_distribution() check_timer_distribution() runs ten threads in a busy loop and tries to test that the kernel distributes a process posix CPU timer signal to every thread over time. There is not guarantee that this is true even after commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") because that commit only avoids waking up the sleeping process leader thread, but that has nothing to do with the actual signal delivery. As the signal is process wide the first thread which observes sigpending and wins the race to lock sighand will deliver the signal. Testing shows that this hangs on a regular base because some threads never win the race. The comment "This primarily tests that the kernel does not favour any one." is wrong. The kernel does favour a thread which hits the timer interrupt when CLOCK_PROCESS_CPUTIME_ID expires. Rewrite the test so it only checks that the group leader sleeping in join() never receives SIGALRM and the thread which burns CPU cycles receives all signals. In older kernels which do not have commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") the test-case fails immediately, the very 1st tick wakes the leader up. Otherwise it quickly succeeds after 100 ticks. CI testing wants to use newer selftest versions on stable kernels. In this case the test is guaranteed to fail. So check in the failure case whether the kernel version is less than v6.3 and skip the test result in that case. [ tglx: Massaged change log, renamed the version check helper ] Fixes: e797203fb3ba ("selftests/timers/posix_timers: Test delivery of signals across threads") Signed-off-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240409133802.GD29396@redhat.com
2024-04-09 13:38:03 +00:00
return NULL;
}
/*
* Test that only the running thread receives the timer signal.
*/
static int check_timer_distribution(void)
{
const char *errmsg;
selftests/timers/posix_timers: Reimplement check_timer_distribution() check_timer_distribution() runs ten threads in a busy loop and tries to test that the kernel distributes a process posix CPU timer signal to every thread over time. There is not guarantee that this is true even after commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") because that commit only avoids waking up the sleeping process leader thread, but that has nothing to do with the actual signal delivery. As the signal is process wide the first thread which observes sigpending and wins the race to lock sighand will deliver the signal. Testing shows that this hangs on a regular base because some threads never win the race. The comment "This primarily tests that the kernel does not favour any one." is wrong. The kernel does favour a thread which hits the timer interrupt when CLOCK_PROCESS_CPUTIME_ID expires. Rewrite the test so it only checks that the group leader sleeping in join() never receives SIGALRM and the thread which burns CPU cycles receives all signals. In older kernels which do not have commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") the test-case fails immediately, the very 1st tick wakes the leader up. Otherwise it quickly succeeds after 100 ticks. CI testing wants to use newer selftest versions on stable kernels. In this case the test is guaranteed to fail. So check in the failure case whether the kernel version is less than v6.3 and skip the test result in that case. [ tglx: Massaged change log, renamed the version check helper ] Fixes: e797203fb3ba ("selftests/timers/posix_timers: Test delivery of signals across threads") Signed-off-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240409133802.GD29396@redhat.com
2024-04-09 13:38:03 +00:00
signal(SIGALRM, ctd_sighandler);
errmsg = "Can't create thread\n";
if (pthread_create(&ctd_thread, NULL, ctd_thread_func, NULL))
goto err;
errmsg = "Can't join thread\n";
if (pthread_join(ctd_thread, (void **)&errmsg) || errmsg)
goto err;
if (!ctd_failed)
ksft_test_result_pass("check signal distribution\n");
else if (ksft_min_kernel_version(6, 3))
ksft_test_result_fail("check signal distribution\n");
else
ksft_test_result_skip("check signal distribution (old kernel)\n");
return 0;
selftests/timers/posix_timers: Reimplement check_timer_distribution() check_timer_distribution() runs ten threads in a busy loop and tries to test that the kernel distributes a process posix CPU timer signal to every thread over time. There is not guarantee that this is true even after commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") because that commit only avoids waking up the sleeping process leader thread, but that has nothing to do with the actual signal delivery. As the signal is process wide the first thread which observes sigpending and wins the race to lock sighand will deliver the signal. Testing shows that this hangs on a regular base because some threads never win the race. The comment "This primarily tests that the kernel does not favour any one." is wrong. The kernel does favour a thread which hits the timer interrupt when CLOCK_PROCESS_CPUTIME_ID expires. Rewrite the test so it only checks that the group leader sleeping in join() never receives SIGALRM and the thread which burns CPU cycles receives all signals. In older kernels which do not have commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") the test-case fails immediately, the very 1st tick wakes the leader up. Otherwise it quickly succeeds after 100 ticks. CI testing wants to use newer selftest versions on stable kernels. In this case the test is guaranteed to fail. So check in the failure case whether the kernel version is less than v6.3 and skip the test result in that case. [ tglx: Massaged change log, renamed the version check helper ] Fixes: e797203fb3ba ("selftests/timers/posix_timers: Test delivery of signals across threads") Signed-off-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240409133802.GD29396@redhat.com
2024-04-09 13:38:03 +00:00
err:
ksft_print_msg("%s", errmsg);
selftests/timers/posix_timers: Reimplement check_timer_distribution() check_timer_distribution() runs ten threads in a busy loop and tries to test that the kernel distributes a process posix CPU timer signal to every thread over time. There is not guarantee that this is true even after commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") because that commit only avoids waking up the sleeping process leader thread, but that has nothing to do with the actual signal delivery. As the signal is process wide the first thread which observes sigpending and wins the race to lock sighand will deliver the signal. Testing shows that this hangs on a regular base because some threads never win the race. The comment "This primarily tests that the kernel does not favour any one." is wrong. The kernel does favour a thread which hits the timer interrupt when CLOCK_PROCESS_CPUTIME_ID expires. Rewrite the test so it only checks that the group leader sleeping in join() never receives SIGALRM and the thread which burns CPU cycles receives all signals. In older kernels which do not have commit bcb7ee79029d ("posix-timers: Prefer delivery of signals to the current thread") the test-case fails immediately, the very 1st tick wakes the leader up. Otherwise it quickly succeeds after 100 ticks. CI testing wants to use newer selftest versions on stable kernels. In this case the test is guaranteed to fail. So check in the failure case whether the kernel version is less than v6.3 and skip the test result in that case. [ tglx: Massaged change log, renamed the version check helper ] Fixes: e797203fb3ba ("selftests/timers/posix_timers: Test delivery of signals across threads") Signed-off-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240409133802.GD29396@redhat.com
2024-04-09 13:38:03 +00:00
return -1;
}
int main(int argc, char **argv)
{
ksft_print_header();
ksft_set_plan(6);
ksft_print_msg("Testing posix timers. False negative may happen on CPU execution \n");
ksft_print_msg("based timers if other threads run on the CPU...\n");
if (check_itimer(ITIMER_VIRTUAL) < 0)
return ksft_exit_fail();
if (check_itimer(ITIMER_PROF) < 0)
return ksft_exit_fail();
if (check_itimer(ITIMER_REAL) < 0)
return ksft_exit_fail();
if (check_timer_create(CLOCK_THREAD_CPUTIME_ID) < 0)
return ksft_exit_fail();
/*
* It's unfortunately hard to reliably test a timer expiration
* on parallel multithread cputime. We could arm it to expire
* on DELAY * nr_threads, with nr_threads busy looping, then wait
* the normal DELAY since the time is elapsing nr_threads faster.
* But for that we need to ensure we have real physical free CPUs
* to ensure true parallelism. So test only one thread until we
* find a better solution.
*/
if (check_timer_create(CLOCK_PROCESS_CPUTIME_ID) < 0)
return ksft_exit_fail();
if (check_timer_distribution() < 0)
return ksft_exit_fail();
ksft_finished();
}