zircon/kernel/tests/thread_tests.cpp

// Copyright 2016, 2018 The Fuchsia Authors
// Copyright (c) 2008-2015 Travis Geiselbrecht
//
// Use of this source code is governed by a MIT-style
// license that can be found in the LICENSE file or at
// https://opensource.org/licenses/MIT

#include "tests.h"

#include <assert.h>
#include <debug.h>
#include <err.h>
#include <fbl/algorithm.h>
#include <inttypes.h>
#include <kernel/event.h>
#include <kernel/mp.h>
#include <kernel/mutex.h>
#include <kernel/thread.h>
#include <platform.h>
#include <pow2.h>
#include <rand.h>
#include <string.h>
#include <trace.h>
#include <zircon/types.h>

static uint rand_range(uint low, uint high) {
    uint r = rand();
    uint result = ((r ^ (r >> 16)) % (high - low + 1u)) + low;

    return result;
}

static int sleep_thread(void* arg) {
    for (;;) {
        printf("sleeper %p\n", get_current_thread());
        thread_sleep_relative(ZX_MSEC(rand() % 500));
    }
    return 0;
}

static int sleep_test(void) {
    int i;
    for (i = 0; i < 16; i++)
        thread_detach_and_resume(thread_create("sleeper", &sleep_thread, NULL, DEFAULT_PRIORITY));
    return 0;
}

static int mutex_thread(void* arg) {
    int i;
    const int iterations = 1000000;
    int count = 0;

    static volatile uintptr_t shared = 0;

    auto m = reinterpret_cast<Mutex*>(arg);

    printf("mutex tester thread %p starting up, will go for %d iterations\n", get_current_thread(), iterations);

    for (i = 0; i < iterations; i++) {
        m->Acquire();

        if (shared != 0)
            panic("someone else has messed with the shared data\n");

        shared = (intptr_t)get_current_thread();
        if ((rand() % 5) == 0)
            thread_yield();

        if (++count % 10000 == 0)
            printf("%p: count %d\n", get_current_thread(), count);
        shared = 0;

        m->Release();
        if ((rand() % 5) == 0)
            thread_yield();
    }

    printf("mutex tester %p done\n", get_current_thread());

    return 0;
}

static int mutex_test(void) {
    static Mutex imutex;
    printf("preinitialized mutex:\n");
    hexdump(&imutex, sizeof(imutex));

    Mutex m;

    thread_t* threads[5];

    for (uint i = 0; i < fbl::count_of(threads); i++) {
        threads[i] = thread_create("mutex tester", &mutex_thread, &m,
                                   get_current_thread()->base_priority);
        thread_resume(threads[i]);
    }

    for (uint i = 0; i < fbl::count_of(threads); i++) {
        thread_join(threads[i], NULL, ZX_TIME_INFINITE);
    }

    thread_sleep_relative(ZX_MSEC(100));

    printf("done with mutex tests\n");

    return 0;
}

static int mutex_inherit_test() {
    printf("running mutex inheritance test\n");

    constexpr uint inherit_test_mutex_count = 4;
    constexpr uint inherit_test_thread_count = 5;

    {  // Explicit scope to control when the destruction of |args| happens
        // working variables to pass the working thread
        struct args {
            event_t test_blocker = EVENT_INITIAL_VALUE(test_blocker, false, 0);
            Mutex test_mutex[inherit_test_mutex_count];
        } args;

        // worker thread to stress the priority inheritance mechanism
        auto inherit_worker = [](void* arg) TA_NO_THREAD_SAFETY_ANALYSIS -> int {
            struct args* args = static_cast<struct args*>(arg);

            for (int count = 0; count < 100000; count++) {
                uint r = rand_range(1, inherit_test_mutex_count);

                // pick a random priority
                thread_set_priority(
                    get_current_thread(), rand_range(DEFAULT_PRIORITY - 4, DEFAULT_PRIORITY + 4));

                // grab a random number of mutexes
                for (uint j = 0; j < r; j++) {
                    args->test_mutex[j].Acquire();
                }

                if (count % 1000 == 0)
                    printf("%p: count %d\n", get_current_thread(), count);

                // wait on a event for a period of time, to try to have other grabber threads
                // need to tweak our priority in either one of the mutexes we hold or the
                // blocking event
                event_wait_deadline(&args->test_blocker, current_time() + ZX_USEC(rand() % 10u), true);

                // release in reverse order
                for (int j = r - 1; j >= 0; j--) {
                    args->test_mutex[j].Release();
                }
            }

            return 0;
        };

        // create a stack of mutexes and a few threads
        thread_t* test_thread[inherit_test_thread_count];
        for (auto &t: test_thread) {
            t = thread_create("mutex tester", inherit_worker, &args,
                                       get_current_thread()->base_priority);
            thread_resume(t);
        }

        for (auto &t: test_thread) {
            thread_join(t, NULL, ZX_TIME_INFINITE);
        }
    }

    thread_sleep_relative(ZX_MSEC(100));

    printf("done with mutex inheirit test\n");

    return 0;
}

static event_t e;

static int event_signaler(void* arg) {
    printf("event signaler pausing\n");
    thread_sleep_relative(ZX_SEC(1));

    //  for (;;) {
    printf("signaling event\n");
    event_signal(&e, true);
    printf("done signaling event\n");
    thread_yield();
    //  }

    return 0;
}

static int event_waiter(void* arg) {
    uintptr_t count = (uintptr_t)arg;

    while (count > 0) {
        printf("thread %p: waiting on event...\n", get_current_thread());
        zx_status_t status = event_wait_deadline(&e, ZX_TIME_INFINITE, true);
        if (status == ZX_ERR_INTERNAL_INTR_KILLED) {
            printf("thread %p: killed\n", get_current_thread());
            return -1;
        } else if (status != ZX_OK) {
            printf("thread %p: event_wait() returned error %d\n", get_current_thread(), status);
            return -1;
        }
        printf("thread %p: done waiting on event\n", get_current_thread());
        thread_yield();
        count--;
    }

    return 0;
}

static void event_test(void) {
    thread_t* threads[5];

    static event_t ievent = EVENT_INITIAL_VALUE(ievent, true, 0x1234);
    printf("preinitialized event:\n");
    hexdump(&ievent, sizeof(ievent));

    printf("event tests starting\n");

    /* make sure signaling the event wakes up all the threads and stays signaled */
    printf("creating event, waiting on it with 4 threads, signaling it and making sure all threads fall through twice\n");
    event_init(&e, false, 0);
    threads[0] = thread_create("event signaler", &event_signaler, NULL, DEFAULT_PRIORITY);
    threads[1] = thread_create("event waiter 0", &event_waiter, (void*)2, DEFAULT_PRIORITY);
    threads[2] = thread_create("event waiter 1", &event_waiter, (void*)2, DEFAULT_PRIORITY);
    threads[3] = thread_create("event waiter 2", &event_waiter, (void*)2, DEFAULT_PRIORITY);
    threads[4] = thread_create("event waiter 3", &event_waiter, (void*)2, DEFAULT_PRIORITY);

    for (uint i = 0; i < fbl::count_of(threads); i++)
        thread_resume(threads[i]);

    for (uint i = 0; i < fbl::count_of(threads); i++)
        thread_join(threads[i], NULL, ZX_TIME_INFINITE);

    thread_sleep_relative(ZX_SEC(2));
    printf("destroying event\n");
    event_destroy(&e);

    /* make sure signaling the event wakes up precisely one thread */
    printf("creating event, waiting on it with 4 threads, signaling it and making sure only one thread wakes up\n");
    event_init(&e, false, EVENT_FLAG_AUTOUNSIGNAL);
    threads[0] = thread_create("event signaler", &event_signaler, NULL, DEFAULT_PRIORITY);
    threads[1] = thread_create("event waiter 0", &event_waiter, (void*)99, DEFAULT_PRIORITY);
    threads[2] = thread_create("event waiter 1", &event_waiter, (void*)99, DEFAULT_PRIORITY);
    threads[3] = thread_create("event waiter 2", &event_waiter, (void*)99, DEFAULT_PRIORITY);
    threads[4] = thread_create("event waiter 3", &event_waiter, (void*)99, DEFAULT_PRIORITY);

    for (uint i = 0; i < fbl::count_of(threads); i++)
        thread_resume(threads[i]);

    thread_sleep_relative(ZX_SEC(2));

    for (uint i = 0; i < fbl::count_of(threads); i++) {
        thread_kill(threads[i]);
        thread_join(threads[i], NULL, ZX_TIME_INFINITE);
    }

    event_destroy(&e);

    printf("event tests done\n");
}

static int quantum_tester(void* arg) {
    for (;;) {
        printf("%p: in this thread. rq %" PRIi64 "\n", get_current_thread(), get_current_thread()->remaining_time_slice);
    }
    return 0;
}

static void quantum_test(void) {
    thread_detach_and_resume(thread_create("quantum tester 0", &quantum_tester, NULL, DEFAULT_PRIORITY));
    thread_detach_and_resume(thread_create("quantum tester 1", &quantum_tester, NULL, DEFAULT_PRIORITY));
    thread_detach_and_resume(thread_create("quantum tester 2", &quantum_tester, NULL, DEFAULT_PRIORITY));
    thread_detach_and_resume(thread_create("quantum tester 3", &quantum_tester, NULL, DEFAULT_PRIORITY));
}

static event_t context_switch_event;
static event_t context_switch_done_event;

static int context_switch_tester(void* arg) {
    int i;
    uint64_t total_count = 0;
    const int iter = 100000;
    uintptr_t thread_count = (uintptr_t)arg;

    event_wait(&context_switch_event);

    uint64_t count = arch_cycle_count();
    for (i = 0; i < iter; i++) {
        thread_yield();
    }
    total_count += arch_cycle_count() - count;
    thread_sleep_relative(ZX_SEC(1));
    printf("took %" PRIu64 " cycles to yield %d times, %" PRIu64 " per yield, %" PRIu64 " per yield per thread\n",
           total_count, iter, total_count / iter, total_count / iter / thread_count);

    event_signal(&context_switch_done_event, true);

    return 0;
}

static void context_switch_test(void) {
    event_init(&context_switch_event, false, 0);
    event_init(&context_switch_done_event, false, 0);

    thread_detach_and_resume(thread_create("context switch idle", &context_switch_tester, (void*)1, DEFAULT_PRIORITY));
    thread_sleep_relative(ZX_MSEC(100));
    event_signal(&context_switch_event, true);
    event_wait(&context_switch_done_event);
    thread_sleep_relative(ZX_MSEC(100));

    event_unsignal(&context_switch_event);
    event_unsignal(&context_switch_done_event);
    thread_detach_and_resume(thread_create("context switch 2a", &context_switch_tester, (void*)2, DEFAULT_PRIORITY));
    thread_detach_and_resume(thread_create("context switch 2b", &context_switch_tester, (void*)2, DEFAULT_PRIORITY));
    thread_sleep_relative(ZX_MSEC(100));
    event_signal(&context_switch_event, true);
    event_wait(&context_switch_done_event);
    thread_sleep_relative(ZX_MSEC(100));

    event_unsignal(&context_switch_event);
    event_unsignal(&context_switch_done_event);
    thread_detach_and_resume(thread_create("context switch 4a", &context_switch_tester, (void*)4, DEFAULT_PRIORITY));
    thread_detach_and_resume(thread_create("context switch 4b", &context_switch_tester, (void*)4, DEFAULT_PRIORITY));
    thread_detach_and_resume(thread_create("context switch 4c", &context_switch_tester, (void*)4, DEFAULT_PRIORITY));
    thread_detach_and_resume(thread_create("context switch 4d", &context_switch_tester, (void*)4, DEFAULT_PRIORITY));
    thread_sleep_relative(ZX_MSEC(100));
    event_signal(&context_switch_event, true);
    event_wait(&context_switch_done_event);
    thread_sleep_relative(ZX_MSEC(100));
}

static volatile int atomic;
static volatile int atomic_count;

static int atomic_tester(void* arg) {
    int add = (int)(uintptr_t)arg;
    int i;

    const int iter = 10000000;

    TRACEF("add %d, %d iterations\n", add, iter);

    for (i = 0; i < iter; i++) {
        atomic_add(&atomic, add);
    }

    int old = atomic_add(&atomic_count, -1);
    TRACEF("exiting, old count %d\n", old);

    return 0;
}

static void atomic_test(void) {
    atomic = 0;
    atomic_count = 8;

    printf("testing atomic routines\n");

    thread_t* threads[8];
    threads[0] = thread_create("atomic tester 1", &atomic_tester, (void*)1, LOW_PRIORITY);
    threads[1] = thread_create("atomic tester 1", &atomic_tester, (void*)1, LOW_PRIORITY);
    threads[2] = thread_create("atomic tester 1", &atomic_tester, (void*)1, LOW_PRIORITY);
    threads[3] = thread_create("atomic tester 1", &atomic_tester, (void*)1, LOW_PRIORITY);
    threads[4] = thread_create("atomic tester 2", &atomic_tester, (void*)-1, LOW_PRIORITY);
    threads[5] = thread_create("atomic tester 2", &atomic_tester, (void*)-1, LOW_PRIORITY);
    threads[6] = thread_create("atomic tester 2", &atomic_tester, (void*)-1, LOW_PRIORITY);
    threads[7] = thread_create("atomic tester 2", &atomic_tester, (void*)-1, LOW_PRIORITY);

    /* start all the threads */
    for (uint i = 0; i < fbl::count_of(threads); i++)
        thread_resume(threads[i]);

    /* wait for them to all stop */
    for (uint i = 0; i < fbl::count_of(threads); i++) {
        thread_join(threads[i], NULL, ZX_TIME_INFINITE);
    }

    printf("atomic count == %d (should be zero)\n", atomic);
}

static volatile int preempt_count;

static int preempt_tester(void* arg) {
    spin(1000000);

    printf("exiting ts %" PRIi64 " ns\n", current_time());

    atomic_add(&preempt_count, -1);

    return 0;
}

static void preempt_test(void) {
    /* create 5 threads, let them run. If the system is properly timer preempting,
     * the threads should interleave each other at a fine enough granularity so
     * that they complete at roughly the same time. */
    printf("testing preemption\n");

    preempt_count = 5;

    for (int i = 0; i < preempt_count; i++)
        thread_detach_and_resume(thread_create("preempt tester", &preempt_tester, NULL, LOW_PRIORITY));

    while (preempt_count > 0) {
        thread_sleep_relative(ZX_SEC(1));
    }

    printf("done with preempt test, above time stamps should be very close\n");

    /* do the same as above, but mark the threads as real time, which should
     * effectively disable timer based preemption for them. They should
     * complete in order, about a second apart. */
    printf("testing real time preemption\n");

    const int num_threads = 5;
    preempt_count = num_threads;

    for (int i = 0; i < num_threads; i++) {
        thread_t* t = thread_create("preempt tester", &preempt_tester, NULL, LOW_PRIORITY);
        thread_set_real_time(t);
        thread_set_cpu_affinity(t, cpu_num_to_mask(0));
        thread_detach_and_resume(t);
    }

    while (preempt_count > 0) {
        thread_sleep_relative(ZX_SEC(1));
    }

    printf("done with real-time preempt test, above time stamps should be 1 second apart\n");
}

static int join_tester(void* arg) {
    int val = (int)(uintptr_t)arg;

    printf("\t\tjoin tester starting\n");
    thread_sleep_relative(ZX_MSEC(500));
    printf("\t\tjoin tester exiting with result %d\n", val);

    return val;
}

static int join_tester_server(void* arg) {
    int ret;
    zx_status_t err;
    thread_t* t;

    printf("\ttesting thread_join/thread_detach\n");

    printf("\tcreating and waiting on thread to exit with thread_join\n");
    t = thread_create("join tester", &join_tester, (void*)1, DEFAULT_PRIORITY);
    thread_resume(t);
    ret = 99;
    printf("\tthread magic is 0x%x (should be 0x%x)\n", (unsigned)t->magic, (unsigned)THREAD_MAGIC);
    err = thread_join(t, &ret, ZX_TIME_INFINITE);
    printf("\tthread_join returns err %d, retval %d\n", err, ret);
    printf("\tthread magic is 0x%x (should be 0)\n", (unsigned)t->magic);

    printf("\tcreating and waiting on thread to exit with thread_join, after thread has exited\n");
    t = thread_create("join tester", &join_tester, (void*)2, DEFAULT_PRIORITY);
    thread_resume(t);
    thread_sleep_relative(ZX_SEC(1)); // wait until thread is already dead
    ret = 99;
    printf("\tthread magic is 0x%x (should be 0x%x)\n", (unsigned)t->magic, (unsigned)THREAD_MAGIC);
    err = thread_join(t, &ret, ZX_TIME_INFINITE);
    printf("\tthread_join returns err %d, retval %d\n", err, ret);
    printf("\tthread magic is 0x%x (should be 0)\n", (unsigned)t->magic);

    printf("\tcreating a thread, detaching it, let it exit on its own\n");
    t = thread_create("join tester", &join_tester, (void*)3, DEFAULT_PRIORITY);
    thread_detach(t);
    thread_resume(t);
    thread_sleep_relative(ZX_SEC(1)); // wait until the thread should be dead
    printf("\tthread magic is 0x%x (should be 0)\n", (unsigned)t->magic);

    printf("\tcreating a thread, detaching it after it should be dead\n");
    t = thread_create("join tester", &join_tester, (void*)4, DEFAULT_PRIORITY);
    thread_resume(t);
    thread_sleep_relative(ZX_SEC(1)); // wait until thread is already dead
    printf("\tthread magic is 0x%x (should be 0x%x)\n", (unsigned)t->magic, (unsigned)THREAD_MAGIC);
    thread_detach(t);
    printf("\tthread magic is 0x%x\n", (unsigned)t->magic);

    printf("\texiting join tester server\n");

    return 55;
}

static void join_test(void) {
    int ret;
    zx_status_t err;
    thread_t* t;

    printf("testing thread_join/thread_detach\n");

    printf("creating thread join server thread\n");
    t = thread_create("join tester server", &join_tester_server, (void*)1, DEFAULT_PRIORITY);
    thread_resume(t);
    ret = 99;
    err = thread_join(t, &ret, ZX_TIME_INFINITE);
    printf("thread_join returns err %d, retval %d (should be 0 and 55)\n", err, ret);
}

struct lock_pair_t {
    spin_lock_t first = SPIN_LOCK_INITIAL_VALUE;
    spin_lock_t second = SPIN_LOCK_INITIAL_VALUE;
};

// Acquires lock on "second" and holds it until it sees that "first" is released.
static int hold_and_release(void* arg) {
    lock_pair_t* pair = reinterpret_cast<lock_pair_t*>(arg);
    ASSERT(pair != nullptr);
    spin_lock_saved_state_t state;
    spin_lock_irqsave(&pair->second, state);
    while (spin_lock_holder_cpu(&pair->first) != UINT_MAX) {
        thread_yield();
    }
    spin_unlock_irqrestore(&pair->second, state);
    return 0;
}

static void spinlock_test(void) {
    spin_lock_saved_state_t state;
    spin_lock_t lock;

    spin_lock_init(&lock);

    // Verify basic functionality (single core).
    printf("testing spinlock:\n");
    ASSERT(!spin_lock_held(&lock));
    ASSERT(!arch_ints_disabled());
    spin_lock_irqsave(&lock, state);
    ASSERT(arch_ints_disabled());
    ASSERT(spin_lock_held(&lock));
    ASSERT(spin_lock_holder_cpu(&lock) == arch_curr_cpu_num());
    spin_unlock_irqrestore(&lock, state);
    ASSERT(!spin_lock_held(&lock));
    ASSERT(!arch_ints_disabled());

    // Verify slightly more advanced functionality that requires multiple cores.
    cpu_mask_t online = mp_get_online_mask();
    if (!online || ispow2(online)) {
        printf("skipping rest of spinlock_test, not enough online cpus\n");
        return;
    }

    // Hold the first lock, then create a thread and wait for it to acquire the lock.
    lock_pair_t pair;
    spin_lock_irqsave(&pair.first, state);
    thread_t* holder_thread = thread_create("hold_and_release", &hold_and_release, &pair,
                                            DEFAULT_PRIORITY);
    ASSERT(holder_thread != nullptr);
    thread_resume(holder_thread);
    while (spin_lock_holder_cpu(&pair.second) == UINT_MAX) {
        thread_yield();
    }

    // See that from our perspective "second" is not held.
    ASSERT(!spin_lock_held(&pair.second));
    spin_unlock_irqrestore(&pair.first, state);
    thread_join(holder_thread, NULL, ZX_TIME_INFINITE);

    printf("seems to work\n");
}

static void sleeper_thread_exit(enum thread_user_state_change new_state, thread_t* arg) {
    TRACEF("arg %p\n", arg);
}

static int sleeper_kill_thread(void* arg) {
    thread_sleep_relative(ZX_MSEC(100));

    zx_time_t t = current_time();
    zx_status_t err = thread_sleep_interruptable(t + ZX_SEC(5));
    zx_duration_t duration = (current_time() - t) / ZX_MSEC(1);
    TRACEF("thread_sleep_interruptable returns %d after %" PRIi64 " msecs\n", err, duration);

    return 0;
}

static void waiter_thread_exit(enum thread_user_state_change new_state, thread_t* arg) {
    TRACEF("arg %p\n", arg);
}

static int waiter_kill_thread_infinite_wait(void* arg) {
    event_t* e = (event_t*)arg;

    thread_sleep_relative(ZX_MSEC(100));

    zx_time_t t = current_time();
    zx_status_t err = event_wait_deadline(e, ZX_TIME_INFINITE, true);
    zx_duration_t duration = (current_time() - t) / ZX_MSEC(1);
    TRACEF("event_wait_deadline returns %d after %" PRIi64 " msecs\n", err, duration);

    return 0;
}

static int waiter_kill_thread(void* arg) {
    event_t* e = (event_t*)arg;

    thread_sleep_relative(ZX_MSEC(100));

    zx_time_t t = current_time();
    zx_status_t err = event_wait_deadline(e, t + ZX_SEC(5), true);
    zx_duration_t duration = (current_time() - t) / ZX_MSEC(1);
    TRACEF("event_wait_deadline with deadline returns %d after %" PRIi64 " msecs\n", err, duration);

    return 0;
}

static void kill_tests(void) {
    thread_t* t;

    printf("starting sleeper thread, then killing it while it sleeps.\n");
    t = thread_create("sleeper", sleeper_kill_thread, 0, LOW_PRIORITY);
    thread_set_user_callback(t, &sleeper_thread_exit);
    thread_resume(t);
    thread_sleep_relative(ZX_MSEC(200));
    thread_kill(t);
    thread_join(t, NULL, ZX_TIME_INFINITE);

    printf("starting sleeper thread, then killing it before it wakes up.\n");
    t = thread_create("sleeper", sleeper_kill_thread, 0, LOW_PRIORITY);
    thread_set_user_callback(t, &sleeper_thread_exit);
    thread_resume(t);
    thread_kill(t);
    thread_join(t, NULL, ZX_TIME_INFINITE);

    printf("starting sleeper thread, then killing it before it is unsuspended.\n");
    t = thread_create("sleeper", sleeper_kill_thread, 0, LOW_PRIORITY);
    thread_set_user_callback(t, &sleeper_thread_exit);
    thread_kill(t); // kill it before it is resumed
    thread_resume(t);
    thread_join(t, NULL, ZX_TIME_INFINITE);

    event_t e;

    printf("starting waiter thread that waits forever, then killing it while it blocks.\n");
    event_init(&e, false, 0);
    t = thread_create("waiter", waiter_kill_thread_infinite_wait, &e, LOW_PRIORITY);
    thread_set_user_callback(t, &waiter_thread_exit);
    thread_resume(t);
    thread_sleep_relative(ZX_MSEC(200));
    thread_kill(t);
    thread_join(t, NULL, ZX_TIME_INFINITE);
    event_destroy(&e);

    printf("starting waiter thread that waits forever, then killing it before it wakes up.\n");
    event_init(&e, false, 0);
    t = thread_create("waiter", waiter_kill_thread_infinite_wait, &e, LOW_PRIORITY);
    thread_set_user_callback(t, &waiter_thread_exit);
    thread_resume(t);
    thread_kill(t);
    thread_join(t, NULL, ZX_TIME_INFINITE);
    event_destroy(&e);

    printf("starting waiter thread that waits some time, then killing it while it blocks.\n");
    event_init(&e, false, 0);
    t = thread_create("waiter", waiter_kill_thread, &e, LOW_PRIORITY);
    thread_set_user_callback(t, &waiter_thread_exit);
    thread_resume(t);
    thread_sleep_relative(ZX_MSEC(200));
    thread_kill(t);
    thread_join(t, NULL, ZX_TIME_INFINITE);
    event_destroy(&e);

    printf("starting waiter thread that waits some time, then killing it before it wakes up.\n");
    event_init(&e, false, 0);
    t = thread_create("waiter", waiter_kill_thread, &e, LOW_PRIORITY);
    thread_set_user_callback(t, &waiter_thread_exit);
    thread_resume(t);
    thread_kill(t);
    thread_join(t, NULL, ZX_TIME_INFINITE);
    event_destroy(&e);
}

struct affinity_test_state {
    thread_t* threads[16] = {};
    volatile bool shutdown = false;
};

template <typename T>
static void spin_while(zx_time_t t, T func) {
    zx_time_t start = current_time();

    while ((current_time() - start) < t) {
        func();
    }
}

static int affinity_test_thread(void* arg) {
    thread_t* t = get_current_thread();
    affinity_test_state* state = static_cast<affinity_test_state*>(arg);

    printf("top of affinity tester %p\n", t);

    while (!state->shutdown) {
        int which = rand() % static_cast<int>(fbl::count_of(state->threads));
        switch (rand() % 5) {
        case 0: // set affinity
            //printf("%p set aff %p\n", t, state->threads[which]);
            thread_set_cpu_affinity(state->threads[which], (cpu_mask_t)rand());
            break;
        case 1: // sleep for a bit
            //printf("%p sleep\n", t);
            thread_sleep_relative(ZX_USEC(rand() % 100));
            break;
        case 2: // spin for a bit
            //printf("%p spin\n", t);
            spin((uint32_t)rand() % 100);
            //printf("%p spin done\n", t);
            break;
        case 3: // yield
            //printf("%p yield\n", t);
            spin_while(ZX_USEC((uint32_t)rand() % 100), thread_yield);
            //printf("%p yield done\n", t);
            break;
        case 4: // reschedule
            //printf("%p reschedule\n", t);
            spin_while(ZX_USEC((uint32_t)rand() % 100), thread_reschedule);
            //printf("%p reschedule done\n", t);
            break;
        }
    }

    printf("affinity tester %p exiting\n", t);

    return 0;
}

// start a bunch of threads that randomly set the affinity of the other threads
// to random masks while doing various work.
// a successful pass is one where it completes the run without tripping over any asserts
// in the scheduler code.
__NO_INLINE static void affinity_test() {
    printf("starting thread affinity test\n");

    cpu_mask_t online = mp_get_online_mask();
    if (!online || ispow2(online)) {
        printf("aborting test, not enough online cpus\n");
        return;
    }

    affinity_test_state state;

    for (auto& t : state.threads) {
        t = thread_create("affinity_tester", &affinity_test_thread, &state,
                          LOW_PRIORITY);
    }

    for (auto& t : state.threads) {
        thread_resume(t);
    }

    static const int duration = 30;
    printf("running tests for %i seconds\n", duration);
    for (int i = 0; i < duration; i++) {
        thread_sleep_relative(ZX_SEC(1));
        printf("%d sec elapsed\n", i + 1);
    }
    state.shutdown = true;
    thread_sleep_relative(ZX_SEC(1));

    for (auto& t : state.threads) {
        printf("joining thread %p\n", t);
        thread_join(t, nullptr, ZX_TIME_INFINITE);
    }

    printf("done with affinity test\n");
}

#define TLS_TEST_TAGV   ((void*)0x666)

static void tls_test_callback(void *tls) {
    ASSERT(tls == TLS_TEST_TAGV);
    atomic_add(&atomic_count, 1);
}

static int tls_test_thread(void* arg) {
    tls_set(0u, TLS_TEST_TAGV);
    tls_set_callback(0u, &tls_test_callback);
    tls_set(1u, TLS_TEST_TAGV);
    tls_set_callback(1u, &tls_test_callback);
    return 0;
}

static void tls_tests() {
    printf("starting tls tests\n");
    atomic_count = 0;

    thread_t* t = thread_create("tls-test", tls_test_thread, 0, LOW_PRIORITY);
    thread_resume(t);
    thread_sleep_relative(ZX_MSEC(200));
    thread_join(t, nullptr, ZX_TIME_INFINITE);

    ASSERT(atomic_count == 2);
    atomic_count = 0;

    printf("done with tls tests\n");
}

static int prio_test_thread(void* arg) {
    thread_t* t = get_current_thread();
    ASSERT(t->base_priority == LOW_PRIORITY);

    auto ev = (event_t*)arg;
    event_signal(ev, false);

    // Busy loop until our priority changes.
    volatile int* v_pri = &t->base_priority;
    int count = 0;
    for (;;) {
        if (*v_pri == DEFAULT_PRIORITY) {
            break;
        }
        ++count;
    }

    event_signal(ev, false);

    // And then when it changes again.
    for (;;) {
        if (*v_pri == HIGH_PRIORITY) {
            break;
        }
        ++count;
    }

    return count;
}

__NO_INLINE static void priority_test() {
    printf("starting priority tests\n");

    thread_t* t = get_current_thread();
    int base_priority = t->base_priority;

    if (base_priority != DEFAULT_PRIORITY) {
        printf("unexpected initial state, aborting test\n");
        return;
    }

    thread_set_priority(t, DEFAULT_PRIORITY + 2);
    thread_sleep_relative(ZX_MSEC(1));
    ASSERT(t->base_priority == DEFAULT_PRIORITY + 2);

    thread_set_priority(t, DEFAULT_PRIORITY - 2);
    thread_sleep_relative(ZX_MSEC(1));
    ASSERT(t->base_priority == DEFAULT_PRIORITY - 2);

    cpu_mask_t online = mp_get_online_mask();
    if (!online || ispow2(online)) {
        printf("skipping rest, not enough online cpus\n");
        return;
    }

    event_t ev;
    event_init(&ev, false, EVENT_FLAG_AUTOUNSIGNAL);

    thread_t* nt = thread_create(
        "prio-test", prio_test_thread, &ev, LOW_PRIORITY);

    cpu_num_t curr = arch_curr_cpu_num();
    cpu_num_t other;
    if (mp_is_cpu_online(curr + 1)) {
        other = curr + 1;
    } else if (mp_is_cpu_online(curr -1)) {
        other = curr - 1;
    } else {
        ASSERT(false);
    }

    thread_set_cpu_affinity(nt, cpu_num_to_mask(other));
    thread_resume(nt);

    zx_status_t status = event_wait_deadline(&ev, ZX_TIME_INFINITE, true);
    ASSERT(status == ZX_OK);
    thread_set_priority(nt, DEFAULT_PRIORITY);

    status = event_wait_deadline(&ev, ZX_TIME_INFINITE, true);
    ASSERT(status == ZX_OK);
    thread_set_priority(nt, HIGH_PRIORITY);

    int count = 0;
    thread_join(nt, &count, ZX_TIME_INFINITE);
    printf("%d loops\n", count);

    printf("done with priority tests\n");
}


int thread_tests(int, const cmd_args*, uint32_t) {
    kill_tests();

    mutex_test();
    event_test();
    mutex_inherit_test();

    spinlock_test();
    atomic_test();

    thread_sleep_relative(ZX_MSEC(200));
    context_switch_test();

    preempt_test();

    join_test();

    affinity_test();

    tls_tests();

    priority_test();

    return 0;
}

static int spinner_thread(void* arg) {
    for (;;)
        ;

    return 0;
}

int spinner(int argc, const cmd_args* argv, uint32_t) {
    if (argc < 2) {
        printf("not enough args\n");
        printf("usage: %s <priority> <rt>\n", argv[0].str);
        return -1;
    }

    thread_t* t = thread_create("spinner", spinner_thread, NULL, (int)argv[1].u);
    if (!t)
        return ZX_ERR_NO_MEMORY;

    if (argc >= 3 && !strcmp(argv[2].str, "rt")) {
        thread_set_real_time(t);
    }
    thread_detach_and_resume(t);

    return 0;
}