strongswan/src/libstrongswan/tests/suites/test_threading.c

1778 lines
33 KiB
C

/*
* Copyright (C) 2013-2018 Tobias Brunner
* Copyright (C) 2008 Martin Willi
* HSR Hochschule fuer Technik Rapperswil
*
* 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. See <http://www.fsf.org/copyleft/gpl.txt>.
*
* 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.
*/
#include "test_suite.h"
#include <unistd.h>
#include <threading/thread.h>
#include <threading/mutex.h>
#include <threading/condvar.h>
#include <threading/rwlock.h>
#include <threading/rwlock_condvar.h>
#include <threading/spinlock.h>
#include <threading/semaphore.h>
#include <threading/thread_value.h>
#ifdef WIN32
/* when running on AppVeyor the wait functions seem to frequently trigger a bit
* early, allow this if the difference is within 5ms. */
static inline void time_is_at_least(timeval_t *expected, timeval_t *actual)
{
if (!timercmp(actual, expected, >))
{
timeval_t diff;
timersub(expected, actual, &diff);
if (!diff.tv_sec && diff.tv_usec <= 5000)
{
warn("allow timer event %dus too early on Windows (expected: %u.%u, "
"actual: %u.%u)", diff.tv_usec, expected->tv_sec,
expected->tv_usec, actual->tv_sec, actual->tv_usec);
return;
}
fail("expected: %u.%u, actual: %u.%u", expected->tv_sec,
expected->tv_usec, actual->tv_sec, actual->tv_usec);
}
}
#else /* WIN32 */
static inline void time_is_at_least(timeval_t *expected, timeval_t *actual)
{
ck_assert_msg(timercmp(actual, expected, >), "expected: %u.%u, actual: "
"%u.%u", expected->tv_sec, expected->tv_usec, actual->tv_sec,
actual->tv_usec);
}
#endif /* WIN32 */
/*******************************************************************************
* recursive mutex test
*/
#define THREADS 20
/**
* Thread barrier data
*/
typedef struct {
mutex_t *mutex;
condvar_t *cond;
int count;
int current;
bool active;
} barrier_t;
/**
* Create a thread barrier for count threads
*/
static barrier_t* barrier_create(int count)
{
barrier_t *this;
INIT(this,
.mutex = mutex_create(MUTEX_TYPE_DEFAULT),
.cond = condvar_create(CONDVAR_TYPE_DEFAULT),
.count = count,
);
return this;
}
/**
* Destroy a thread barrier
*/
static void barrier_destroy(barrier_t *this)
{
this->mutex->destroy(this->mutex);
this->cond->destroy(this->cond);
free(this);
}
/**
* Wait to have configured number of threads in barrier
*/
static bool barrier_wait(barrier_t *this)
{
bool winner = FALSE;
this->mutex->lock(this->mutex);
if (!this->active)
{ /* first, reset */
this->active = TRUE;
this->current = 0;
}
this->current++;
while (this->current < this->count)
{
this->cond->wait(this->cond, this->mutex);
}
if (this->active)
{ /* first, win */
winner = TRUE;
this->active = FALSE;
}
this->mutex->unlock(this->mutex);
this->cond->broadcast(this->cond);
sched_yield();
return winner;
}
/**
* Barrier for some tests
*/
static barrier_t *barrier;
/**
* A mutex for tests requiring one
*/
static mutex_t *mutex;
/**
* A condvar for tests requiring one
*/
static condvar_t *condvar;
/**
* A counter for signaling
*/
static int sigcount;
static void *mutex_run(void *data)
{
int locked = 0;
int i;
/* wait for all threads before getting in action */
barrier_wait(barrier);
for (i = 0; i < 100; i++)
{
mutex->lock(mutex);
mutex->lock(mutex);
mutex->lock(mutex);
locked++;
sched_yield();
if (locked > 1)
{
fail("two threads locked the mutex concurrently");
}
locked--;
mutex->unlock(mutex);
mutex->unlock(mutex);
mutex->unlock(mutex);
}
return NULL;
}
START_TEST(test_mutex)
{
thread_t *threads[THREADS];
int i;
barrier = barrier_create(THREADS);
mutex = mutex_create(MUTEX_TYPE_RECURSIVE);
for (i = 0; i < 10; i++)
{
mutex->lock(mutex);
mutex->unlock(mutex);
}
for (i = 0; i < 10; i++)
{
mutex->lock(mutex);
}
for (i = 0; i < 10; i++)
{
mutex->unlock(mutex);
}
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create(mutex_run, NULL);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
}
mutex->destroy(mutex);
barrier_destroy(barrier);
}
END_TEST
/**
* Spinlock for testing
*/
static spinlock_t *spinlock;
static void *spinlock_run(void *data)
{
int i, *locked = (int*)data;
barrier_wait(barrier);
for (i = 0; i < 1000; i++)
{
spinlock->lock(spinlock);
(*locked)++;
ck_assert_int_eq(*locked, 1);
(*locked)--;
spinlock->unlock(spinlock);
}
return NULL;
}
START_TEST(test_spinlock)
{
thread_t *threads[THREADS];
int i, locked = 0;
barrier = barrier_create(THREADS);
spinlock = spinlock_create();
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create(spinlock_run, &locked);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
}
spinlock->destroy(spinlock);
barrier_destroy(barrier);
}
END_TEST
static void *condvar_run(void *data)
{
mutex->lock(mutex);
sigcount++;
condvar->signal(condvar);
mutex->unlock(mutex);
return NULL;
}
START_TEST(test_condvar)
{
thread_t *threads[THREADS];
int i;
mutex = mutex_create(MUTEX_TYPE_DEFAULT);
condvar = condvar_create(CONDVAR_TYPE_DEFAULT);
sigcount = 0;
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create(condvar_run, NULL);
}
mutex->lock(mutex);
while (sigcount < THREADS)
{
condvar->wait(condvar, mutex);
}
mutex->unlock(mutex);
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
}
mutex->destroy(mutex);
condvar->destroy(condvar);
}
END_TEST
static void *condvar_recursive_run(void *data)
{
mutex->lock(mutex);
mutex->lock(mutex);
mutex->lock(mutex);
sigcount++;
condvar->signal(condvar);
mutex->unlock(mutex);
mutex->unlock(mutex);
mutex->unlock(mutex);
return NULL;
}
START_TEST(test_condvar_recursive)
{
thread_t *threads[THREADS];
int i;
mutex = mutex_create(MUTEX_TYPE_RECURSIVE);
condvar = condvar_create(CONDVAR_TYPE_DEFAULT);
sigcount = 0;
mutex->lock(mutex);
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create(condvar_recursive_run, NULL);
}
mutex->lock(mutex);
mutex->lock(mutex);
while (sigcount < THREADS)
{
condvar->wait(condvar, mutex);
}
mutex->unlock(mutex);
mutex->unlock(mutex);
mutex->unlock(mutex);
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
}
mutex->destroy(mutex);
condvar->destroy(condvar);
}
END_TEST
static void *condvar_run_broad(void *data)
{
mutex->lock(mutex);
while (sigcount < 0)
{
condvar->wait(condvar, mutex);
}
mutex->unlock(mutex);
return NULL;
}
START_TEST(test_condvar_broad)
{
thread_t *threads[THREADS];
int i;
mutex = mutex_create(MUTEX_TYPE_DEFAULT);
condvar = condvar_create(CONDVAR_TYPE_DEFAULT);
sigcount = 0;
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create(condvar_run_broad, NULL);
}
sched_yield();
mutex->lock(mutex);
sigcount = 1;
condvar->broadcast(condvar);
mutex->unlock(mutex);
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
}
mutex->destroy(mutex);
condvar->destroy(condvar);
}
END_TEST
START_TEST(test_condvar_timed)
{
thread_t *thread;
timeval_t start, end, diff = { .tv_usec = 50000 };
mutex = mutex_create(MUTEX_TYPE_DEFAULT);
condvar = condvar_create(CONDVAR_TYPE_DEFAULT);
sigcount = 0;
mutex->lock(mutex);
while (TRUE)
{
time_monotonic(&start);
if (condvar->timed_wait(condvar, mutex, diff.tv_usec / 1000))
{
break;
}
}
time_monotonic(&end);
mutex->unlock(mutex);
timersub(&end, &start, &end);
time_is_at_least(&diff, &end);
thread = thread_create(condvar_run, NULL);
mutex->lock(mutex);
while (sigcount == 0)
{
ck_assert(!condvar->timed_wait(condvar, mutex, 1000));
}
mutex->unlock(mutex);
thread->join(thread);
mutex->destroy(mutex);
condvar->destroy(condvar);
}
END_TEST
START_TEST(test_condvar_timed_abs)
{
thread_t *thread;
timeval_t start, end, abso, diff = { .tv_usec = 50000 };
mutex = mutex_create(MUTEX_TYPE_DEFAULT);
condvar = condvar_create(CONDVAR_TYPE_DEFAULT);
sigcount = 0;
mutex->lock(mutex);
while (TRUE)
{
time_monotonic(&start);
timeradd(&start, &diff, &abso);
if (condvar->timed_wait_abs(condvar, mutex, abso))
{
break;
}
}
time_monotonic(&end);
mutex->unlock(mutex);
time_is_at_least(&diff, &end);
thread = thread_create(condvar_run, NULL);
time_monotonic(&start);
diff.tv_sec = 1;
timeradd(&start, &diff, &abso);
mutex->lock(mutex);
while (sigcount == 0)
{
ck_assert(!condvar->timed_wait_abs(condvar, mutex, abso));
}
mutex->unlock(mutex);
thread->join(thread);
mutex->destroy(mutex);
condvar->destroy(condvar);
}
END_TEST
static void *condvar_cancel_run(void *data)
{
thread_cancelability(FALSE);
mutex->lock(mutex);
sigcount++;
condvar->broadcast(condvar);
thread_cleanup_push((void*)mutex->unlock, mutex);
thread_cancelability(TRUE);
while (TRUE)
{
condvar->wait(condvar, mutex);
}
thread_cleanup_pop(TRUE);
return NULL;
}
START_TEST(test_condvar_cancel)
{
thread_t *threads[THREADS];
int i;
mutex = mutex_create(MUTEX_TYPE_DEFAULT);
condvar = condvar_create(CONDVAR_TYPE_DEFAULT);
sigcount = 0;
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create(condvar_cancel_run, NULL);
}
/* wait for all threads */
mutex->lock(mutex);
while (sigcount < THREADS)
{
condvar->wait(condvar, mutex);
}
mutex->unlock(mutex);
for (i = 0; i < THREADS; i++)
{
threads[i]->cancel(threads[i]);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
}
mutex->destroy(mutex);
condvar->destroy(condvar);
}
END_TEST
/**
* RWlock for different tests
*/
static rwlock_t *rwlock;
static void *rwlock_run(refcount_t *refs)
{
rwlock->read_lock(rwlock);
ref_get(refs);
sched_yield();
ignore_result(ref_put(refs));
rwlock->unlock(rwlock);
if (rwlock->try_write_lock(rwlock))
{
ck_assert_int_eq(*refs, 0);
sched_yield();
rwlock->unlock(rwlock);
}
rwlock->write_lock(rwlock);
ck_assert_int_eq(*refs, 0);
sched_yield();
rwlock->unlock(rwlock);
rwlock->read_lock(rwlock);
rwlock->read_lock(rwlock);
ref_get(refs);
sched_yield();
ignore_result(ref_put(refs));
rwlock->unlock(rwlock);
rwlock->unlock(rwlock);
return NULL;
}
START_TEST(test_rwlock)
{
thread_t *threads[THREADS];
refcount_t refs = 0;
int i;
rwlock = rwlock_create(RWLOCK_TYPE_DEFAULT);
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create((void*)rwlock_run, &refs);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
}
rwlock->destroy(rwlock);
}
END_TEST
static void *rwlock_try_run(void *param)
{
if (rwlock->try_write_lock(rwlock))
{
rwlock->unlock(rwlock);
return param;
}
return NULL;
}
START_TEST(test_rwlock_try)
{
uintptr_t magic = 0xcafebabe;
thread_t *thread;
rwlock = rwlock_create(RWLOCK_TYPE_DEFAULT);
thread = thread_create(rwlock_try_run, (void*)magic);
ck_assert_int_eq((uintptr_t)thread->join(thread), magic);
rwlock->read_lock(rwlock);
thread = thread_create(rwlock_try_run, (void*)magic);
ck_assert(thread->join(thread) == NULL);
rwlock->unlock(rwlock);
rwlock->read_lock(rwlock);
rwlock->read_lock(rwlock);
rwlock->read_lock(rwlock);
thread = thread_create(rwlock_try_run, (void*)magic);
ck_assert(thread->join(thread) == NULL);
rwlock->unlock(rwlock);
rwlock->unlock(rwlock);
rwlock->unlock(rwlock);
rwlock->write_lock(rwlock);
thread = thread_create(rwlock_try_run, (void*)magic);
ck_assert(thread->join(thread) == NULL);
rwlock->unlock(rwlock);
rwlock->destroy(rwlock);
}
END_TEST
/**
* Rwlock condvar
*/
static rwlock_condvar_t *rwcond;
static void *rwlock_condvar_run(void *data)
{
rwlock->write_lock(rwlock);
sigcount++;
rwcond->signal(rwcond);
rwlock->unlock(rwlock);
return NULL;
}
START_TEST(test_rwlock_condvar)
{
thread_t *threads[THREADS];
int i;
rwlock = rwlock_create(RWLOCK_TYPE_DEFAULT);
rwcond = rwlock_condvar_create();
sigcount = 0;
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create(rwlock_condvar_run, NULL);
}
rwlock->write_lock(rwlock);
while (sigcount < THREADS)
{
rwcond->wait(rwcond, rwlock);
}
rwlock->unlock(rwlock);
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
}
rwlock->destroy(rwlock);
rwcond->destroy(rwcond);
}
END_TEST
static void *rwlock_condvar_run_broad(void *data)
{
rwlock->write_lock(rwlock);
while (sigcount < 0)
{
rwcond->wait(rwcond, rwlock);
}
rwlock->unlock(rwlock);
return NULL;
}
START_TEST(test_rwlock_condvar_broad)
{
thread_t *threads[THREADS];
int i;
rwlock = rwlock_create(RWLOCK_TYPE_DEFAULT);
rwcond = rwlock_condvar_create();
sigcount = 0;
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create(rwlock_condvar_run_broad, NULL);
}
sched_yield();
rwlock->write_lock(rwlock);
sigcount = 1;
rwcond->broadcast(rwcond);
rwlock->unlock(rwlock);
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
}
rwlock->destroy(rwlock);
rwcond->destroy(rwcond);
}
END_TEST
START_TEST(test_rwlock_condvar_timed)
{
thread_t *thread;
timeval_t start, end, diff = { .tv_usec = 50000 };
rwlock = rwlock_create(RWLOCK_TYPE_DEFAULT);
rwcond = rwlock_condvar_create();
sigcount = 0;
rwlock->write_lock(rwlock);
while (TRUE)
{
time_monotonic(&start);
if (rwcond->timed_wait(rwcond, rwlock, diff.tv_usec / 1000))
{
break;
}
}
rwlock->unlock(rwlock);
time_monotonic(&end);
timersub(&end, &start, &end);
time_is_at_least(&diff, &end);
thread = thread_create(rwlock_condvar_run, NULL);
rwlock->write_lock(rwlock);
while (sigcount == 0)
{
ck_assert(!rwcond->timed_wait(rwcond, rwlock, 1000));
}
rwlock->unlock(rwlock);
thread->join(thread);
rwlock->destroy(rwlock);
rwcond->destroy(rwcond);
}
END_TEST
START_TEST(test_rwlock_condvar_timed_abs)
{
thread_t *thread;
timeval_t start, end, abso, diff = { .tv_usec = 50000 };
rwlock = rwlock_create(RWLOCK_TYPE_DEFAULT);
rwcond = rwlock_condvar_create();
sigcount = 0;
rwlock->write_lock(rwlock);
while (TRUE)
{
time_monotonic(&start);
timeradd(&start, &diff, &abso);
if (rwcond->timed_wait_abs(rwcond, rwlock, abso))
{
break;
}
}
rwlock->unlock(rwlock);
time_monotonic(&end);
time_is_at_least(&abso, &end);
thread = thread_create(rwlock_condvar_run, NULL);
time_monotonic(&start);
diff.tv_sec = 1;
timeradd(&start, &diff, &abso);
rwlock->write_lock(rwlock);
while (sigcount == 0)
{
ck_assert(!rwcond->timed_wait_abs(rwcond, rwlock, abso));
}
rwlock->unlock(rwlock);
thread->join(thread);
rwlock->destroy(rwlock);
rwcond->destroy(rwcond);
}
END_TEST
static void *rwlock_condvar_cancel_run(void *data)
{
thread_cancelability(FALSE);
rwlock->write_lock(rwlock);
sigcount++;
rwcond->broadcast(rwcond);
thread_cleanup_push((void*)rwlock->unlock, rwlock);
thread_cancelability(TRUE);
while (TRUE)
{
rwcond->wait(rwcond, rwlock);
}
thread_cleanup_pop(TRUE);
return NULL;
}
START_TEST(test_rwlock_condvar_cancel)
{
thread_t *threads[THREADS];
int i;
rwlock = rwlock_create(RWLOCK_TYPE_DEFAULT);
rwcond = rwlock_condvar_create();
sigcount = 0;
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create(rwlock_condvar_cancel_run, NULL);
}
/* wait for all threads */
rwlock->write_lock(rwlock);
while (sigcount < THREADS)
{
rwcond->wait(rwcond, rwlock);
}
rwlock->unlock(rwlock);
for (i = 0; i < THREADS; i++)
{
threads[i]->cancel(threads[i]);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
}
rwlock->destroy(rwlock);
rwcond->destroy(rwcond);
}
END_TEST
/**
* Semaphore for different tests
*/
static semaphore_t *semaphore;
static void *semaphore_run(void *data)
{
semaphore->post(semaphore);
return NULL;
}
START_TEST(test_semaphore)
{
thread_t *threads[THREADS];
int i, initial = 5;
semaphore = semaphore_create(initial);
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create(semaphore_run, NULL);
}
for (i = 0; i < THREADS + initial; i++)
{
semaphore->wait(semaphore);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
}
semaphore->destroy(semaphore);
}
END_TEST
START_TEST(test_semaphore_timed)
{
thread_t *thread;
timeval_t start, end, diff = { .tv_usec = 50000 };
semaphore = semaphore_create(0);
time_monotonic(&start);
ck_assert(semaphore->timed_wait(semaphore, diff.tv_usec / 1000));
time_monotonic(&end);
timersub(&end, &start, &end);
time_is_at_least(&diff, &end);
thread = thread_create(semaphore_run, NULL);
ck_assert(!semaphore->timed_wait(semaphore, 1000));
thread->join(thread);
semaphore->destroy(semaphore);
}
END_TEST
START_TEST(test_semaphore_timed_abs)
{
thread_t *thread;
timeval_t start, end, abso, diff = { .tv_usec = 50000 };
semaphore = semaphore_create(0);
time_monotonic(&start);
timeradd(&start, &diff, &abso);
ck_assert(semaphore->timed_wait_abs(semaphore, abso));
time_monotonic(&end);
time_is_at_least(&abso, &end);
thread = thread_create(semaphore_run, NULL);
time_monotonic(&start);
diff.tv_sec = 1;
timeradd(&start, &diff, &abso);
ck_assert(!semaphore->timed_wait_abs(semaphore, abso));
thread->join(thread);
semaphore->destroy(semaphore);
}
END_TEST
static void *semaphore_cancel_run(void *data)
{
refcount_t *ready = (refcount_t*)data;
thread_cancelability(FALSE);
ref_get(ready);
thread_cancelability(TRUE);
semaphore->wait(semaphore);
ck_assert(FALSE);
return NULL;
}
START_TEST(test_semaphore_cancel)
{
thread_t *threads[THREADS];
refcount_t ready = 0;
int i;
semaphore = semaphore_create(0);
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create(semaphore_cancel_run, &ready);
}
while (ready < THREADS)
{
sched_yield();
}
for (i = 0; i < THREADS; i++)
{
threads[i]->cancel(threads[i]);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
}
semaphore->destroy(semaphore);
}
END_TEST
static void *join_run(void *data)
{
/* force some context switches */
sched_yield();
return (void*)((uintptr_t)data + THREADS);
}
START_TEST(test_join)
{
thread_t *threads[THREADS];
int i;
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create(join_run, (void*)(uintptr_t)i);
}
for (i = 0; i < THREADS; i++)
{
ck_assert_int_eq((uintptr_t)threads[i]->join(threads[i]), i + THREADS);
}
}
END_TEST
static void *exit_join_run(void *data)
{
sched_yield();
thread_exit((void*)((uintptr_t)data + THREADS));
/* not reached */
ck_assert(FALSE);
return NULL;
}
START_TEST(test_join_exit)
{
thread_t *threads[THREADS];
int i;
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create(exit_join_run, (void*)(uintptr_t)i);
}
for (i = 0; i < THREADS; i++)
{
ck_assert_int_eq((uintptr_t)threads[i]->join(threads[i]), i + THREADS);
}
}
END_TEST
static void *detach_run(void *data)
{
refcount_t *running = (refcount_t*)data;
ignore_result(ref_put(running));
return NULL;
}
START_TEST(test_detach)
{
thread_t *threads[THREADS];
int i;
refcount_t running = 0;
for (i = 0; i < THREADS; i++)
{
ref_get(&running);
threads[i] = thread_create(detach_run, &running);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->detach(threads[i]);
}
while (running > 0)
{
sched_yield();
}
/* no checks done here, but we check that thread state gets cleaned
* up with leak detective. give the threads time to clean up. */
usleep(10000);
}
END_TEST
static void *detach_exit_run(void *data)
{
refcount_t *running = (refcount_t*)data;
ignore_result(ref_put(running));
thread_exit(NULL);
/* not reached */
ck_assert(FALSE);
return NULL;
}
START_TEST(test_detach_exit)
{
thread_t *threads[THREADS];
int i;
refcount_t running = 0;
for (i = 0; i < THREADS; i++)
{
ref_get(&running);
threads[i] = thread_create(detach_exit_run, &running);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->detach(threads[i]);
}
while (running > 0)
{
sched_yield();
}
/* no checks done here, but we check that thread state gets cleaned
* up with leak detective. give the threads time to clean up. */
usleep(10000);
}
END_TEST
static void *cancel_run(void *data)
{
/* default cancelability should be TRUE, so don't change it */
while (TRUE)
{
sleep(10);
}
return NULL;
}
START_TEST(test_cancel)
{
thread_t *threads[THREADS];
int i;
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create(cancel_run, NULL);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->cancel(threads[i]);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
}
}
END_TEST
typedef struct {
semaphore_t *sem;
bool cancellable;
} cancel_onoff_data_t;
static void *cancel_onoff_run(void *data_in)
{
cancel_onoff_data_t *data = (cancel_onoff_data_t*)data_in;
thread_cancelability(FALSE);
data->cancellable = FALSE;
/* we should not get cancelled here */
data->sem->wait(data->sem);
data->cancellable = TRUE;
thread_cancelability(TRUE);
/* but here */
while (TRUE)
{
sleep(10);
}
return NULL;
}
START_TEST(test_cancel_onoff)
{
thread_t *threads[THREADS];
cancel_onoff_data_t data[THREADS];
semaphore_t *sem;
int i;
sem = semaphore_create(0);
for (i = 0; i < THREADS; i++)
{
data[i].sem = sem;
data[i].cancellable = TRUE;
threads[i] = thread_create(cancel_onoff_run, &data[i]);
/* wait until thread has cleared its cancelability */
while (data[i].cancellable)
{
sched_yield();
}
}
for (i = 0; i < THREADS; i++)
{
threads[i]->cancel(threads[i]);
}
/* let all threads continue */
for (i = 0; i < THREADS; i++)
{
sem->post(sem);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
ck_assert(data[i].cancellable);
}
sem->destroy(sem);
}
END_TEST
static void *cancel_point_run(void *data)
{
thread_cancelability(FALSE);
while (TRUE)
{
/* implicitly enables cancelability */
thread_cancellation_point();
}
return NULL;
}
START_TEST(test_cancel_point)
{
thread_t *threads[THREADS];
int i;
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create(cancel_point_run, NULL);
}
sched_yield();
for (i = 0; i < THREADS; i++)
{
threads[i]->cancel(threads[i]);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
}
}
END_TEST
static void close_fd_ptr(void *fd)
{
close(*(int*)fd);
}
static void cancellation_recv()
{
int sv[2];
char buf[1];
ck_assert(socketpair(AF_UNIX, SOCK_STREAM, 0, sv) == 0);
thread_cleanup_push(close_fd_ptr, &sv[0]);
thread_cleanup_push(close_fd_ptr, &sv[1]);
thread_cancelability(TRUE);
while (TRUE)
{
ck_assert(recv(sv[0], buf, sizeof(buf), 0) == 1);
}
}
static void cancellation_read()
{
int sv[2];
char buf[1];
ck_assert(socketpair(AF_UNIX, SOCK_STREAM, 0, sv) == 0);
thread_cleanup_push(close_fd_ptr, &sv[0]);
thread_cleanup_push(close_fd_ptr, &sv[1]);
thread_cancelability(TRUE);
while (TRUE)
{
ck_assert(read(sv[0], buf, sizeof(buf)) == 1);
}
}
static void cancellation_select()
{
int sv[2];
fd_set set;
ck_assert(socketpair(AF_UNIX, SOCK_STREAM, 0, sv) == 0);
thread_cleanup_push(close_fd_ptr, &sv[0]);
thread_cleanup_push(close_fd_ptr, &sv[1]);
FD_ZERO(&set);
FD_SET(sv[0], &set);
thread_cancelability(TRUE);
while (TRUE)
{
ck_assert(select(sv[0] + 1, &set, NULL, NULL, NULL) == 1);
}
}
static void cancellation_poll()
{
int sv[2];
struct pollfd pfd;
ck_assert(socketpair(AF_UNIX, SOCK_STREAM, 0, sv) == 0);
thread_cleanup_push(close_fd_ptr, &sv[0]);
thread_cleanup_push(close_fd_ptr, &sv[1]);
pfd.fd = sv[0];
pfd.events = POLLIN;
thread_cancelability(TRUE);
while (TRUE)
{
ck_assert(poll(&pfd, 1, -1) == 1);
}
}
static void cancellation_accept()
{
host_t *host;
int fd, c;
fd = socket(AF_INET, SOCK_STREAM, 0);
ck_assert(fd >= 0);
host = host_create_from_string("127.0.0.1", 0);
ck_assert_msg(bind(fd, host->get_sockaddr(host),
*host->get_sockaddr_len(host)) == 0, "%m");
host->destroy(host);
ck_assert(listen(fd, 5) == 0);
thread_cleanup_push(close_fd_ptr, &fd);
thread_cancelability(TRUE);
while (TRUE)
{
c = accept(fd, NULL, NULL);
ck_assert(c >= 0);
close(c);
}
}
static void cancellation_cond()
{
mutex_t *mutex;
condvar_t *cond;
mutex = mutex_create(MUTEX_TYPE_DEFAULT);
cond = condvar_create(CONDVAR_TYPE_DEFAULT);
mutex->lock(mutex);
thread_cleanup_push((void*)mutex->destroy, mutex);
thread_cleanup_push((void*)cond->destroy, cond);
thread_cancelability(TRUE);
while (TRUE)
{
cond->wait(cond, mutex);
}
}
static void cancellation_rwcond()
{
rwlock_t *lock;
rwlock_condvar_t *cond;
lock = rwlock_create(RWLOCK_TYPE_DEFAULT);
cond = rwlock_condvar_create();
lock->write_lock(lock);
thread_cleanup_push((void*)lock->destroy, lock);
thread_cleanup_push((void*)cond->destroy, cond);
thread_cancelability(TRUE);
while (TRUE)
{
cond->wait(cond, lock);
}
}
static void (*cancellation_points[])() = {
cancellation_read,
cancellation_recv,
cancellation_select,
cancellation_poll,
cancellation_accept,
cancellation_cond,
cancellation_rwcond,
};
static void* run_cancellation_point(void (*fn)())
{
fn();
return NULL;
}
static void* run_cancellation_point_pre(void (*fn)())
{
usleep(5000);
fn();
return NULL;
}
START_TEST(test_cancellation_point)
{
thread_t *thread;
thread = thread_create((void*)run_cancellation_point,
cancellation_points[_i]);
usleep(5000);
thread->cancel(thread);
thread->join(thread);
}
END_TEST
START_TEST(test_cancellation_point_pre)
{
thread_t *thread;
thread = thread_create((void*)run_cancellation_point_pre,
cancellation_points[_i]);
thread->cancel(thread);
thread->join(thread);
}
END_TEST
static void cleanup1(void *data)
{
uintptr_t *value = (uintptr_t*)data;
ck_assert_int_eq(*value, 1);
(*value)++;
}
static void cleanup2(void *data)
{
uintptr_t *value = (uintptr_t*)data;
ck_assert_int_eq(*value, 2);
(*value)++;
}
static void cleanup3(void *data)
{
uintptr_t *value = (uintptr_t*)data;
ck_assert_int_eq(*value, 3);
(*value)++;
}
static void *cleanup_run(void *data)
{
thread_cleanup_push(cleanup3, data);
thread_cleanup_push(cleanup2, data);
thread_cleanup_push(cleanup1, data);
return NULL;
}
START_TEST(test_cleanup)
{
thread_t *threads[THREADS];
uintptr_t values[THREADS];
int i;
for (i = 0; i < THREADS; i++)
{
values[i] = 1;
threads[i] = thread_create(cleanup_run, &values[i]);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
ck_assert_int_eq(values[i], 4);
}
}
END_TEST
static void *cleanup_exit_run(void *data)
{
thread_cleanup_push(cleanup3, data);
thread_cleanup_push(cleanup2, data);
thread_cleanup_push(cleanup1, data);
thread_exit(NULL);
ck_assert(FALSE);
return NULL;
}
START_TEST(test_cleanup_exit)
{
thread_t *threads[THREADS];
uintptr_t values[THREADS];
int i;
for (i = 0; i < THREADS; i++)
{
values[i] = 1;
threads[i] = thread_create(cleanup_exit_run, &values[i]);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
ck_assert_int_eq(values[i], 4);
}
}
END_TEST
static void *cleanup_cancel_run(void *data)
{
thread_cancelability(FALSE);
barrier_wait(barrier);
thread_cleanup_push(cleanup3, data);
thread_cleanup_push(cleanup2, data);
thread_cleanup_push(cleanup1, data);
thread_cancelability(TRUE);
while (TRUE)
{
sleep(1);
}
return NULL;
}
START_TEST(test_cleanup_cancel)
{
thread_t *threads[THREADS];
uintptr_t values[THREADS];
int i;
barrier = barrier_create(THREADS+1);
for (i = 0; i < THREADS; i++)
{
values[i] = 1;
threads[i] = thread_create(cleanup_cancel_run, &values[i]);
}
barrier_wait(barrier);
for (i = 0; i < THREADS; i++)
{
threads[i]->cancel(threads[i]);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
ck_assert_int_eq(values[i], 4);
}
barrier_destroy(barrier);
}
END_TEST
static void *cleanup_pop_run(void *data)
{
thread_cleanup_push(cleanup3, data);
thread_cleanup_push(cleanup2, data);
thread_cleanup_push(cleanup1, data);
thread_cleanup_push(cleanup2, data);
thread_cleanup_pop(FALSE);
thread_cleanup_pop(TRUE);
return NULL;
}
START_TEST(test_cleanup_pop)
{
thread_t *threads[THREADS];
uintptr_t values[THREADS];
int i;
for (i = 0; i < THREADS; i++)
{
values[i] = 1;
threads[i] = thread_create(cleanup_pop_run, &values[i]);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
ck_assert_int_eq(values[i], 4);
}
}
END_TEST
static void *cleanup_popall_run(void *data)
{
thread_cleanup_push(cleanup3, data);
thread_cleanup_push(cleanup2, data);
thread_cleanup_push(cleanup1, data);
thread_cleanup_popall();
return NULL;
}
START_TEST(test_cleanup_popall)
{
thread_t *threads[THREADS];
uintptr_t values[THREADS];
int i;
for (i = 0; i < THREADS; i++)
{
values[i] = 1;
threads[i] = thread_create(cleanup_popall_run, &values[i]);
}
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
ck_assert_int_eq(values[i], 4);
}
}
END_TEST
static thread_value_t *tls[10];
static void *tls_run(void *data)
{
uintptr_t value = (uintptr_t)data;
int i, j;
for (i = 0; i < countof(tls); i++)
{
ck_assert(tls[i]->get(tls[i]) == NULL);
}
for (i = 0; i < countof(tls); i++)
{
tls[i]->set(tls[i], (void*)(value * i));
}
for (j = 0; j < 1000; j++)
{
for (i = 0; i < countof(tls); i++)
{
tls[i]->set(tls[i], (void*)(value * i));
ck_assert(tls[i]->get(tls[i]) == (void*)(value * i));
}
sched_yield();
}
for (i = 0; i < countof(tls); i++)
{
ck_assert(tls[i]->get(tls[i]) == (void*)(value * i));
}
return (void*)(value + 1);
}
START_TEST(test_tls)
{
thread_t *threads[THREADS];
int i;
for (i = 0; i < countof(tls); i++)
{
tls[i] = thread_value_create(NULL);
}
for (i = 0; i < THREADS; i++)
{
threads[i] = thread_create(tls_run, (void*)(uintptr_t)i);
}
ck_assert_int_eq((uintptr_t)tls_run((void*)(uintptr_t)(THREADS + 1)),
THREADS + 2);
for (i = 0; i < THREADS; i++)
{
ck_assert_int_eq((uintptr_t)threads[i]->join(threads[i]), i + 1);
}
for (i = 0; i < countof(tls); i++)
{
tls[i]->destroy(tls[i]);
}
}
END_TEST
static void tls_cleanup(void *data)
{
uintptr_t *value = (uintptr_t*)data;
(*value)--;
}
static void *tls_cleanup_run(void *data)
{
int i;
for (i = 0; i < countof(tls); i++)
{
tls[i]->set(tls[i], data);
}
return NULL;
}
START_TEST(test_tls_cleanup)
{
thread_t *threads[THREADS];
uintptr_t values[THREADS], main_value = countof(tls);
int i;
for (i = 0; i < countof(tls); i++)
{
tls[i] = thread_value_create(tls_cleanup);
}
for (i = 0; i < THREADS; i++)
{
values[i] = countof(tls);
threads[i] = thread_create(tls_cleanup_run, &values[i]);
}
tls_cleanup_run(&main_value);
for (i = 0; i < THREADS; i++)
{
threads[i]->join(threads[i]);
ck_assert_int_eq(values[i], 0);
}
for (i = 0; i < countof(tls); i++)
{
tls[i]->destroy(tls[i]);
}
ck_assert_int_eq(main_value, 0);
}
END_TEST
Suite *threading_suite_create()
{
Suite *s;
TCase *tc;
s = suite_create("threading");
tc = tcase_create("recursive mutex");
tcase_add_test(tc, test_mutex);
suite_add_tcase(s, tc);
tc = tcase_create("spinlock");
tcase_add_test(tc, test_spinlock);
suite_add_tcase(s, tc);
tc = tcase_create("condvar");
tcase_add_test(tc, test_condvar);
tcase_add_test(tc, test_condvar_recursive);
tcase_add_test(tc, test_condvar_broad);
tcase_add_test(tc, test_condvar_timed);
tcase_add_test(tc, test_condvar_timed_abs);
tcase_add_test(tc, test_condvar_cancel);
suite_add_tcase(s, tc);
tc = tcase_create("rwlock");
tcase_add_test(tc, test_rwlock);
tcase_add_test(tc, test_rwlock_try);
suite_add_tcase(s, tc);
tc = tcase_create("rwlock condvar");
tcase_add_test(tc, test_rwlock_condvar);
tcase_add_test(tc, test_rwlock_condvar_broad);
tcase_add_test(tc, test_rwlock_condvar_timed);
tcase_add_test(tc, test_rwlock_condvar_timed_abs);
tcase_add_test(tc, test_rwlock_condvar_cancel);
suite_add_tcase(s, tc);
tc = tcase_create("semaphore");
tcase_add_test(tc, test_semaphore);
tcase_add_test(tc, test_semaphore_timed);
tcase_add_test(tc, test_semaphore_timed_abs);
tcase_add_test(tc, test_semaphore_cancel);
suite_add_tcase(s, tc);
tc = tcase_create("thread joining");
tcase_add_test(tc, test_join);
tcase_add_test(tc, test_join_exit);
suite_add_tcase(s, tc);
tc = tcase_create("thread detaching");
tcase_add_test(tc, test_detach);
tcase_add_test(tc, test_detach_exit);
suite_add_tcase(s, tc);
tc = tcase_create("thread cancellation");
tcase_add_test(tc, test_cancel);
tcase_add_test(tc, test_cancel_onoff);
tcase_add_test(tc, test_cancel_point);
suite_add_tcase(s, tc);
tc = tcase_create("thread cancellation point");
tcase_add_loop_test(tc, test_cancellation_point,
0, countof(cancellation_points));
tcase_add_loop_test(tc, test_cancellation_point_pre,
0, countof(cancellation_points));
suite_add_tcase(s, tc);
tc = tcase_create("thread cleanup");
tcase_add_test(tc, test_cleanup);
tcase_add_test(tc, test_cleanup_exit);
tcase_add_test(tc, test_cleanup_cancel);
tcase_add_test(tc, test_cleanup_pop);
tcase_add_test(tc, test_cleanup_popall);
suite_add_tcase(s, tc);
tc = tcase_create("thread local storage");
tcase_add_test(tc, test_tls);
tcase_add_test(tc, test_tls_cleanup);
suite_add_tcase(s, tc);
return s;
}