/*
 * Copyright (c) 2015-2020 François Tigeot <ftigeot@wolfpond.org>
 * Copyright (c) 2020 Matthew Dillon <dillon@backplane.com>
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice unmodified, this list of conditions, and the following
 *    disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */

#include <drm/drmP.h>
#include <linux/workqueue.h>

#include <sys/kthread.h>

/*
   Running behaviour, from kernel.org docs:
   - While there are work items on the workqueue the worker executes the functions
   associated with the work items one after the other.
   - When there is no work item left on the workqueue the worker becomes idle.

   There are two worker-pools,
   one for normal work items
   and the other for high priority ones, for each possible CPU
   and some extra worker-pools to serve work items queued on unbound workqueues
   - the number of these backing pools is dynamic.
 */

/* XXX: Linux functions often enable/disable irqs on the CPU they run on
 * this should be investigated */

struct workqueue_struct *system_wq;
struct workqueue_struct *system_highpri_wq;
struct workqueue_struct *system_long_wq;
struct workqueue_struct *system_unbound_wq;
struct workqueue_struct *system_power_efficient_wq;

/*
 * Linux now uses these worker pools:
 * - (per cpu) regular
 * - (per cpu) regular high priority
 * - ordered
 * - ordered high priority
 * - unbound
 * - unbound high priority
 */

static inline void
process_all_work(struct workqueue_worker *worker)
{
        struct work_struct *work;
        bool didcan;

        while (STAILQ_FIRST(&worker->ws_list_head)) {
                work = STAILQ_FIRST(&worker->ws_list_head);
                STAILQ_REMOVE_HEAD(&worker->ws_list_head, ws_entries);
                work->on_queue = false;

                /* A work shouldn't be executed concurrently on a single cpu */
                if (work->running)
                        continue;

                /* Do not run canceled works */
                if (work->canceled) {
                        /* XXX: should we allow canceled works to be reenabled ? */
                        work->canceled = false;
                        continue;
                }

                work->running = true;
                lockmgr(&worker->worker_lock, LK_RELEASE);
                work->func(work);
                lwkt_yield();
                lockmgr(&worker->worker_lock, LK_EXCLUSIVE);
                if (work->on_queue == false)
                        work->worker = NULL;
                didcan = work->canceled;
                cpu_sfence();
                work->running = false;
                if (didcan == true)
                        wakeup(work);
        }
}

static void
wq_worker_thread(void *arg)
{
        struct workqueue_worker *worker = arg;

        lockmgr(&worker->worker_lock, LK_EXCLUSIVE);
        while (1) {
                process_all_work(worker);
                lksleep(worker, &worker->worker_lock, 0, "wqidle", 0);
        }
        lockmgr(&worker->worker_lock, LK_RELEASE);
}

/*
 * Return false if work was already on a queue
 * Return true and queue it if this was not the case
 */
int
queue_work(struct workqueue_struct *wq, struct work_struct *work)
{
        struct workqueue_worker *worker;
        int ret = false;

        /* XXX: should we block instead ? */
        if (wq->is_draining)
                return false;

        if (wq->num_workers > 1)
                worker = &(*wq->workers)[mycpuid];
        else
                worker = &(*wq->workers)[0];

        lockmgr(&worker->worker_lock, LK_EXCLUSIVE);
        work->canceled = false;
        if (work->on_queue == false || work->running == false) {
                if (work->on_queue == false) {
                        STAILQ_INSERT_TAIL(&worker->ws_list_head, work,
                                           ws_entries);
                        work->on_queue = true;
                        work->worker = worker;
                        wakeup_one(worker);
                }
                ret = true;
        }
        lockmgr(&worker->worker_lock, LK_RELEASE);

        return ret;
}

static inline void
_delayed_work_fn(void *arg)
{
        struct delayed_work *dw = arg;

        queue_work(system_wq, &dw->work);
}

int
queue_delayed_work(struct workqueue_struct *wq, struct delayed_work *work,
    unsigned long delay)
{
        int pending = work->work.on_queue; // XXX: running too ?
        if (delay != 0) {
                callout_reset(&work->timer, delay, _delayed_work_fn, work);
        } else {
                _delayed_work_fn((void *)work);
        }

        return (!pending);
}

static int
init_workqueues(void *arg)
{
        system_wq = alloc_workqueue("system_wq", 0, 1);
        system_highpri_wq = alloc_workqueue("system_highpri_wq", WQ_HIGHPRI, 1);
        system_long_wq = alloc_workqueue("system_long_wq", 0, 1);
        system_unbound_wq = alloc_workqueue("system_unbound_wq", WQ_UNBOUND, 1);
        system_power_efficient_wq = alloc_workqueue("system_power_efficient_wq", 0, 1);

        return 0;
}

static int destroy_workqueues(void *arg)
{
        destroy_workqueue(system_wq);
        destroy_workqueue(system_highpri_wq);
        destroy_workqueue(system_long_wq);
        destroy_workqueue(system_unbound_wq);
        destroy_workqueue(system_power_efficient_wq);

        return 0;
}

struct workqueue_struct *
_create_workqueue_common(const char *name, int flags)
{
        struct workqueue_struct *wq;
        int priority, error;

        wq = kmalloc(sizeof(*wq), M_DRM, M_WAITOK | M_ZERO);

        if (flags & WQ_HIGHPRI)
                priority = TDPRI_INT_SUPPORT;
        else
                priority = TDPRI_KERN_DAEMON;

        if (flags & WQ_UNBOUND) {
                wq->num_workers = 1;
        } else {
                wq->num_workers = ncpus;
        }
        wq->workers = kmalloc(sizeof(struct workqueue_worker) * wq->num_workers,
                        M_DRM, M_WAITOK | M_ZERO);

        for (int i = 0;i < wq->num_workers; i++) {
                struct workqueue_worker *worker = &(*wq->workers)[i];

                lockinit(&worker->worker_lock, "lwq", 0, 0);
                STAILQ_INIT(&worker->ws_list_head);
                if (wq->num_workers > 1) {
                        error = lwkt_create(wq_worker_thread, worker,
                                    &worker->worker_thread, NULL, TDF_NOSTART, i, "%s/%d", name, i);
                } else {
                        error = lwkt_create(wq_worker_thread, worker,
                                    &worker->worker_thread, NULL, TDF_NOSTART, -1, name);
                }
                if (error) {
                        kprintf("%s: lwkt_create(%s/%d): error %d",
                            __func__, name, i, error);
                        /* XXX: destroy kernel threads and free workers[] if applicable */
                        kfree(wq);
                        return NULL;
                }
                lwkt_setpri_initial(worker->worker_thread, priority);
                lwkt_schedule(worker->worker_thread);
        }

        return wq;
}

void
destroy_workqueue(struct workqueue_struct *wq)
{
        drain_workqueue(wq);
//      wq->is_draining = true;
#if 0   /* XXX TODO */
        kill_all_threads;
        kfree(wq->wq_threads);
        kfree(wq);
#endif
}

SYSINIT(linux_workqueue_init, SI_SUB_DRIVERS, SI_ORDER_MIDDLE, init_workqueues, NULL);
SYSUNINIT(linux_workqueue_destroy, SI_SUB_DRIVERS, SI_ORDER_MIDDLE, destroy_workqueues, NULL);

bool
flush_delayed_work(struct delayed_work *dwork)
{
        callout_drain(&dwork->timer);
        return flush_work(&dwork->work);
}

/* Wait until the wq becomes empty */
void
drain_workqueue(struct workqueue_struct *wq)
{
        struct workqueue_worker *worker;

        wq->is_draining = true;

        for (int i=0;i < wq->num_workers; i++) {
                worker = &(*wq->workers)[i];

                lockmgr(&worker->worker_lock, LK_EXCLUSIVE);
                while (!STAILQ_EMPTY(&worker->ws_list_head)) {
                /* XXX: introduces latency */
                        tsleep(&drain_workqueue, 0, "wkdrain", 1);
                }
                lockmgr(&worker->worker_lock, LK_RELEASE);
        }

        /* XXX: No more work will be queued. is that right ? */
//      wq->is_draining = false;
}

bool
work_pending(struct work_struct *work)
{
        /* XXX: is on_queue the only constraint ? */
        return work->on_queue;
}

unsigned int
work_busy(struct work_struct *work)
{
        return (work->on_queue || work->running);
}

static inline void
__flush_work_func(struct work_struct *work)
{
        wakeup_one(work);
}

/* XXX introduces latency ? */
void
flush_workqueue(struct workqueue_struct *wq)
{
        struct work_struct __flush_work;

        INIT_WORK(&__flush_work, __flush_work_func);

        queue_work(wq, &__flush_work);
        while (__flush_work.on_queue || __flush_work.running) {
                tsleep(&__flush_work, 0, "flshwq", 0);
        }
}

/*
 * Wait until a work is done (has been executed)
 * Return true if this function had to wait, and false otherwise
 */
bool
flush_work(struct work_struct *work)
{
        int ret = false;

        /* XXX: probably unreliable */
        while (work->on_queue || work->running) {
                ret = true;
                /* XXX: use something more intelligent than tsleep() */
                tsleep(&flush_work, 0, "flshwrk", 1);
        }

        return ret;
}

static inline bool
_cancel_work(struct work_struct *work, bool sync_wait)
{
        struct workqueue_worker *worker;
        bool ret;

        ret = false;

        for (;;) {
                if (work->on_queue) {
                        worker = work->worker;
                        if (worker == NULL)
                                continue;
                        lockmgr(&worker->worker_lock, LK_EXCLUSIVE);
                        if (worker != work->worker || work->on_queue == false) {
                                lockmgr(&worker->worker_lock, LK_RELEASE);
                                continue;
                        }
                        STAILQ_REMOVE(&worker->ws_list_head, work,
                                      work_struct, ws_entries);
                        work->on_queue = false;
                        ret = true;
                        lockmgr(&worker->worker_lock, LK_RELEASE);
                }
                if (work->running == false)
                        break;

                worker = work->worker;
                if (worker == NULL)
                        continue;
                lockmgr(&worker->worker_lock, LK_EXCLUSIVE);
                if (worker != work->worker || work->running == false) {
                        lockmgr(&worker->worker_lock, LK_RELEASE);
                        continue;
                }
                work->canceled = true;
                ret = true;
                if (sync_wait == false) {
                        lockmgr(&worker->worker_lock, LK_RELEASE);
                        break;
                }
                /* XXX this races */
                lksleep(work, &worker->worker_lock, 0, "wqcan", 1);
                lockmgr(&worker->worker_lock, LK_RELEASE);
                /* retest */
        }

        return ret;
}

/*
 * If work was queued, remove it from the queue and return true.
 * If work was not queued, return false.
 * In any case, wait for work to complete or be removed from the workqueue,
 * callers may free associated data structures after this call.
 */
bool
cancel_work_sync(struct work_struct *work)
{
        return _cancel_work(work, true);
}

/* Return false if work wasn't pending
 * Return true if work was pending and canceled */
bool
cancel_delayed_work(struct delayed_work *dwork)
{
        struct work_struct *work = &dwork->work;

        work->canceled = true;
        callout_cancel(&dwork->timer);

        return _cancel_work(work, false);
}

bool
cancel_delayed_work_sync(struct delayed_work *dwork)
{
        struct work_struct *work = &dwork->work;

        work->canceled = true;
        callout_cancel(&dwork->timer);

        return _cancel_work(work, true);
}

bool
delayed_work_pending(struct delayed_work *dw)
{
        /* XXX: possibly wrong if the timer hasn't yet fired */
        return work_pending(&dw->work);
}

void
destroy_work_on_stack(struct work_struct *work)
{
}

void
destroy_delayed_work_on_stack(struct delayed_work *work)
{
}