libdispatch-84.5.tar.gz

[apple/libdispatch.git] / src / source.c

/*
 * Copyright (c) 2008-2009 Apple Inc. All rights reserved.
 *
 * @APPLE_APACHE_LICENSE_HEADER_START@
 * 
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 * 
 *     http://www.apache.org/licenses/LICENSE-2.0
 * 
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 * 
 * @APPLE_APACHE_LICENSE_HEADER_END@
 */

#include "internal.h"
#include "protocol.h"
#include "protocolServer.h"
#include <sys/mount.h>

#define DISPATCH_EVFILT_TIMER   (-EVFILT_SYSCOUNT - 1)
#define DISPATCH_EVFILT_CUSTOM_ADD      (-EVFILT_SYSCOUNT - 2)
#define DISPATCH_EVFILT_CUSTOM_OR       (-EVFILT_SYSCOUNT - 3)
#define DISPATCH_EVFILT_SYSCOUNT        (EVFILT_SYSCOUNT + 3)

#define DISPATCH_TIMER_INDEX_WALL 0
#define DISPATCH_TIMER_INDEX_MACH 1
static struct dispatch_kevent_s _dispatch_kevent_timer[] = {
        {
                .dk_kevent = {
                        .ident = DISPATCH_TIMER_INDEX_WALL,
                        .filter = DISPATCH_EVFILT_TIMER,
                        .udata = &_dispatch_kevent_timer[0],
                },
                .dk_sources = TAILQ_HEAD_INITIALIZER(_dispatch_kevent_timer[0].dk_sources),
        },
        {
                .dk_kevent = {
                        .ident = DISPATCH_TIMER_INDEX_MACH,
                        .filter = DISPATCH_EVFILT_TIMER,
                        .udata = &_dispatch_kevent_timer[1],
                },
                .dk_sources = TAILQ_HEAD_INITIALIZER(_dispatch_kevent_timer[1].dk_sources),
        },
};
#define DISPATCH_TIMER_COUNT (sizeof _dispatch_kevent_timer / sizeof _dispatch_kevent_timer[0])

static struct dispatch_kevent_s _dispatch_kevent_data_or = {
        .dk_kevent = {
                .filter = DISPATCH_EVFILT_CUSTOM_OR,
                .flags = EV_CLEAR,
                .udata = &_dispatch_kevent_data_or,
        },
        .dk_sources = TAILQ_HEAD_INITIALIZER(_dispatch_kevent_data_or.dk_sources),
};
static struct dispatch_kevent_s _dispatch_kevent_data_add = {
        .dk_kevent = {
                .filter = DISPATCH_EVFILT_CUSTOM_ADD,
                .udata = &_dispatch_kevent_data_add,
        },
        .dk_sources = TAILQ_HEAD_INITIALIZER(_dispatch_kevent_data_add.dk_sources),
};

#ifndef DISPATCH_NO_LEGACY
struct dispatch_source_attr_vtable_s {
        DISPATCH_VTABLE_HEADER(dispatch_source_attr_s);
};

struct dispatch_source_attr_s {
        DISPATCH_STRUCT_HEADER(dispatch_source_attr_s, dispatch_source_attr_vtable_s);
        void* finalizer_ctxt;
        dispatch_source_finalizer_function_t finalizer_func;
        void* context;
};
#endif /* DISPATCH_NO_LEGACY */

#define _dispatch_source_call_block ((void *)-1)
static void _dispatch_source_latch_and_call(dispatch_source_t ds);
static void _dispatch_source_cancel_callout(dispatch_source_t ds);
static bool _dispatch_source_probe(dispatch_source_t ds);
static void _dispatch_source_dispose(dispatch_source_t ds);
static void _dispatch_source_merge_kevent(dispatch_source_t ds, const struct kevent *ke);
static size_t _dispatch_source_debug(dispatch_source_t ds, char* buf, size_t bufsiz);
static size_t dispatch_source_debug_attr(dispatch_source_t ds, char* buf, size_t bufsiz);
static dispatch_queue_t _dispatch_source_invoke(dispatch_source_t ds);

static void _dispatch_kevent_merge(dispatch_source_t ds);
static void _dispatch_kevent_release(dispatch_source_t ds);
static void _dispatch_kevent_resume(dispatch_kevent_t dk, uint32_t new_flags, uint32_t del_flags);
static void _dispatch_kevent_machport_resume(dispatch_kevent_t dk, uint32_t new_flags, uint32_t del_flags);
static void _dispatch_kevent_machport_enable(dispatch_kevent_t dk);
static void _dispatch_kevent_machport_disable(dispatch_kevent_t dk);

static void _dispatch_drain_mach_messages(struct kevent *ke);
static void _dispatch_timer_list_update(dispatch_source_t ds);

static void
_dispatch_mach_notify_source_init(void *context __attribute__((unused)));

static const char *
_evfiltstr(short filt)
{
        switch (filt) {
#define _evfilt2(f) case (f): return #f
        _evfilt2(EVFILT_READ);
        _evfilt2(EVFILT_WRITE);
        _evfilt2(EVFILT_AIO);
        _evfilt2(EVFILT_VNODE);
        _evfilt2(EVFILT_PROC);
        _evfilt2(EVFILT_SIGNAL);
        _evfilt2(EVFILT_TIMER);
        _evfilt2(EVFILT_MACHPORT);
        _evfilt2(EVFILT_FS);
        _evfilt2(EVFILT_USER);
        _evfilt2(EVFILT_SESSION);

        _evfilt2(DISPATCH_EVFILT_TIMER);
        _evfilt2(DISPATCH_EVFILT_CUSTOM_ADD);
        _evfilt2(DISPATCH_EVFILT_CUSTOM_OR);
        default:
                return "EVFILT_missing";
        }
}

#define DSL_HASH_SIZE 256u      // must be a power of two
#define DSL_HASH(x)     ((x) & (DSL_HASH_SIZE - 1))

static TAILQ_HEAD(, dispatch_kevent_s) _dispatch_sources[DSL_HASH_SIZE];

static dispatch_kevent_t
_dispatch_kevent_find(uintptr_t ident, short filter)
{
        uintptr_t hash = DSL_HASH(filter == EVFILT_MACHPORT ? MACH_PORT_INDEX(ident) : ident);
        dispatch_kevent_t dki;

        TAILQ_FOREACH(dki, &_dispatch_sources[hash], dk_list) {
                if (dki->dk_kevent.ident == ident && dki->dk_kevent.filter == filter) {
                        break;
                }
        }
        return dki;
}

static void
_dispatch_kevent_insert(dispatch_kevent_t dk)
{
        uintptr_t ident = dk->dk_kevent.ident;
        uintptr_t hash = DSL_HASH(dk->dk_kevent.filter == EVFILT_MACHPORT ? MACH_PORT_INDEX(ident) : ident);

        TAILQ_INSERT_TAIL(&_dispatch_sources[hash], dk, dk_list);
}

void
dispatch_source_cancel(dispatch_source_t ds)
{
#if DISPATCH_DEBUG
        dispatch_debug(ds, __FUNCTION__);
#endif
        dispatch_atomic_or(&ds->ds_atomic_flags, DSF_CANCELED);
        _dispatch_wakeup(ds);
}

#ifndef DISPATCH_NO_LEGACY
void
_dispatch_source_legacy_xref_release(dispatch_source_t ds)
{
        if (ds->ds_is_legacy) {
                if (!(ds->ds_timer.flags & DISPATCH_TIMER_ONESHOT)) {
                        dispatch_source_cancel(ds);
                }

                // Clients often leave sources suspended at the last release
                dispatch_atomic_and(&ds->do_suspend_cnt, DISPATCH_OBJECT_SUSPEND_LOCK);
        } else if (slowpath(DISPATCH_OBJECT_SUSPENDED(ds))) {
                // Arguments for and against this assert are within 6705399
                DISPATCH_CLIENT_CRASH("Release of a suspended object");
        }
        _dispatch_wakeup(ds);
        _dispatch_release(ds);
}
#endif /* DISPATCH_NO_LEGACY */

long
dispatch_source_testcancel(dispatch_source_t ds)
{
        return (bool)(ds->ds_atomic_flags & DSF_CANCELED);
}


unsigned long
dispatch_source_get_mask(dispatch_source_t ds)
{
        return ds->ds_pending_data_mask;
}

uintptr_t
dispatch_source_get_handle(dispatch_source_t ds)
{
        return (int)ds->ds_ident_hack;
}

unsigned long
dispatch_source_get_data(dispatch_source_t ds)
{
        return ds->ds_data;
}

#if DISPATCH_DEBUG
void
dispatch_debug_kevents(struct kevent* kev, size_t count, const char* str)
{
        size_t i;
        for (i = 0; i < count; ++i) {
                _dispatch_log("kevent[%lu] = { ident = %p, filter = %s, flags = 0x%x, fflags = 0x%x, data = %p, udata = %p }: %s",
                        i, (void*)kev[i].ident, _evfiltstr(kev[i].filter), kev[i].flags, kev[i].fflags, (void*)kev[i].data, (void*)kev[i].udata, str);
        }
}
#endif

static size_t
_dispatch_source_kevent_debug(dispatch_source_t ds, char* buf, size_t bufsiz)
{
        size_t offset = _dispatch_source_debug(ds, buf, bufsiz);
        offset += snprintf(&buf[offset], bufsiz - offset, "filter = %s }",
                        ds->ds_dkev ? _evfiltstr(ds->ds_dkev->dk_kevent.filter) : "????");
        return offset;
}

static void
_dispatch_source_init_tail_queue_array(void *context __attribute__((unused)))
{
        unsigned int i;
        for (i = 0; i < DSL_HASH_SIZE; i++) {
                TAILQ_INIT(&_dispatch_sources[i]);
        }

        TAILQ_INSERT_TAIL(&_dispatch_sources[DSL_HASH(DISPATCH_TIMER_INDEX_WALL)], &_dispatch_kevent_timer[DISPATCH_TIMER_INDEX_WALL], dk_list);
        TAILQ_INSERT_TAIL(&_dispatch_sources[DSL_HASH(DISPATCH_TIMER_INDEX_MACH)], &_dispatch_kevent_timer[DISPATCH_TIMER_INDEX_MACH], dk_list);
        TAILQ_INSERT_TAIL(&_dispatch_sources[0], &_dispatch_kevent_data_or, dk_list);
        TAILQ_INSERT_TAIL(&_dispatch_sources[0], &_dispatch_kevent_data_add, dk_list);
}

// Find existing kevents, and merge any new flags if necessary
void
_dispatch_kevent_merge(dispatch_source_t ds)
{
        static dispatch_once_t pred;
        dispatch_kevent_t dk;
        typeof(dk->dk_kevent.fflags) new_flags;
        bool do_resume = false;

        if (ds->ds_is_installed) {
                return;
        }
        ds->ds_is_installed = true;

        dispatch_once_f(&pred, NULL, _dispatch_source_init_tail_queue_array);

        dk = _dispatch_kevent_find(ds->ds_dkev->dk_kevent.ident, ds->ds_dkev->dk_kevent.filter);
        
        if (dk) {
                // If an existing dispatch kevent is found, check to see if new flags
                // need to be added to the existing kevent
                new_flags = ~dk->dk_kevent.fflags & ds->ds_dkev->dk_kevent.fflags;
                dk->dk_kevent.fflags |= ds->ds_dkev->dk_kevent.fflags;
                free(ds->ds_dkev);
                ds->ds_dkev = dk;
                do_resume = new_flags;
        } else {
                dk = ds->ds_dkev;
                _dispatch_kevent_insert(dk);
                new_flags = dk->dk_kevent.fflags;
                do_resume = true;
        }

        TAILQ_INSERT_TAIL(&dk->dk_sources, ds, ds_list);

        // Re-register the kevent with the kernel if new flags were added
        // by the dispatch kevent
        if (do_resume) {
                dk->dk_kevent.flags |= EV_ADD;
                _dispatch_kevent_resume(ds->ds_dkev, new_flags, 0);
                ds->ds_is_armed = true;
        }
}


void
_dispatch_kevent_resume(dispatch_kevent_t dk, uint32_t new_flags, uint32_t del_flags)
{
        switch (dk->dk_kevent.filter) {
        case DISPATCH_EVFILT_TIMER:
        case DISPATCH_EVFILT_CUSTOM_ADD:
        case DISPATCH_EVFILT_CUSTOM_OR:
                // these types not registered with kevent
                return;
        case EVFILT_MACHPORT:
                _dispatch_kevent_machport_resume(dk, new_flags, del_flags);
                break;
        case EVFILT_PROC:
                if (dk->dk_kevent.flags & EV_ONESHOT) {
                        return;
                }
                // fall through
        default:
                _dispatch_update_kq(&dk->dk_kevent);
                if (dk->dk_kevent.flags & EV_DISPATCH) {
                        dk->dk_kevent.flags &= ~EV_ADD;
                }
                break;
        }
}

dispatch_queue_t
_dispatch_source_invoke(dispatch_source_t ds)
{
        // This function performs all source actions. Each action is responsible
        // for verifying that it takes place on the appropriate queue. If the
        // current queue is not the correct queue for this action, the correct queue
        // will be returned and the invoke will be re-driven on that queue.

        // The order of tests here in invoke and in probe should be consistent.
        
        dispatch_queue_t dq = _dispatch_queue_get_current();

        if (!ds->ds_is_installed) {
                // The source needs to be installed on the manager queue.
                if (dq != &_dispatch_mgr_q) {
                        return &_dispatch_mgr_q;
                }
                _dispatch_kevent_merge(ds);
        } else if ((ds->ds_atomic_flags & DSF_CANCELED) || (ds->do_xref_cnt == 0)) {
                // The source has been cancelled and needs to be uninstalled from the
                // manager queue. After uninstallation, the cancellation handler needs
                // to be delivered to the target queue.
                if (ds->ds_dkev) {
                        if (dq != &_dispatch_mgr_q) {
                                return &_dispatch_mgr_q;
                        }
                        _dispatch_kevent_release(ds);
                        return ds->do_targetq;
                } else if (ds->ds_cancel_handler) {
                        if (dq != ds->do_targetq) {
                                return ds->do_targetq;
                        }
                }       
                _dispatch_source_cancel_callout(ds);
        } else if (ds->ds_pending_data) {
                // The source has pending data to deliver via the event handler callback
                // on the target queue. Some sources need to be rearmed on the manager
                // queue after event delivery.
                if (dq != ds->do_targetq) {
                        return ds->do_targetq;
                }
                _dispatch_source_latch_and_call(ds);
                if (ds->ds_needs_rearm) {
                        return &_dispatch_mgr_q;
                }
        } else if (ds->ds_needs_rearm && !ds->ds_is_armed) {
                // The source needs to be rearmed on the manager queue.
                if (dq != &_dispatch_mgr_q) {
                        return &_dispatch_mgr_q;
                }
                _dispatch_kevent_resume(ds->ds_dkev, 0, 0);
                ds->ds_is_armed = true;
        }

        return NULL;
}

bool
_dispatch_source_probe(dispatch_source_t ds)
{
        // This function determines whether the source needs to be invoked.
        // The order of tests here in probe and in invoke should be consistent.

        if (!ds->ds_is_installed) {
                // The source needs to be installed on the manager queue.
                return true;
        } else if ((ds->ds_atomic_flags & DSF_CANCELED) || (ds->do_xref_cnt == 0)) {
                // The source needs to be uninstalled from the manager queue, or the
                // cancellation handler needs to be delivered to the target queue.
                // Note: cancellation assumes installation.
                if (ds->ds_dkev || ds->ds_cancel_handler) {
                        return true;
                }
        } else if (ds->ds_pending_data) {
                // The source has pending data to deliver to the target queue.
                return true;
        } else if (ds->ds_needs_rearm && !ds->ds_is_armed) {
                // The source needs to be rearmed on the manager queue.
                return true;
        }
        // Nothing to do.
        return false;
}

void
_dispatch_source_dispose(dispatch_source_t ds)
{
        _dispatch_queue_dispose((dispatch_queue_t)ds);
}

static void
_dispatch_kevent_debugger2(void *context, dispatch_source_t unused __attribute__((unused)))
{
        struct sockaddr sa;
        socklen_t sa_len = sizeof(sa);
        int c, fd = (int)(long)context;
        unsigned int i;
        dispatch_kevent_t dk;
        dispatch_source_t ds;
        FILE *debug_stream;

        c = accept(fd, &sa, &sa_len);
        if (c == -1) {
                if (errno != EAGAIN) {
                        dispatch_assume_zero(errno);
                }
                return;
        }
#if 0
        int r = fcntl(c, F_SETFL, 0);   // disable non-blocking IO
        if (r == -1) {
                dispatch_assume_zero(errno);
        }
#endif
        debug_stream = fdopen(c, "a");
        if (!dispatch_assume(debug_stream)) {
                close(c);
                return;
        }

        fprintf(debug_stream, "HTTP/1.0 200 OK\r\n");
        fprintf(debug_stream, "Content-type: text/html\r\n");
        fprintf(debug_stream, "Pragma: nocache\r\n");
        fprintf(debug_stream, "\r\n");
        fprintf(debug_stream, "<html>\n<head><title>PID %u</title></head>\n<body>\n<ul>\n", getpid());

        //fprintf(debug_stream, "<tr><td>DK</td><td>DK</td><td>DK</td><td>DK</td><td>DK</td><td>DK</td><td>DK</td></tr>\n");

        for (i = 0; i < DSL_HASH_SIZE; i++) {
                if (TAILQ_EMPTY(&_dispatch_sources[i])) {
                        continue;
                }
                TAILQ_FOREACH(dk, &_dispatch_sources[i], dk_list) {
                        fprintf(debug_stream, "\t<br><li>DK %p ident %lu filter %s flags 0x%hx fflags 0x%x data 0x%lx udata %p\n",
                                        dk, dk->dk_kevent.ident, _evfiltstr(dk->dk_kevent.filter), dk->dk_kevent.flags,
                                        dk->dk_kevent.fflags, dk->dk_kevent.data, dk->dk_kevent.udata);
                        fprintf(debug_stream, "\t\t<ul>\n");
                        TAILQ_FOREACH(ds, &dk->dk_sources, ds_list) {
                                fprintf(debug_stream, "\t\t\t<li>DS %p refcnt 0x%x suspend 0x%x data 0x%lx mask 0x%lx flags 0x%x</li>\n",
                                                ds, ds->do_ref_cnt, ds->do_suspend_cnt, ds->ds_pending_data, ds->ds_pending_data_mask,
                                                ds->ds_atomic_flags);
                                if (ds->do_suspend_cnt == DISPATCH_OBJECT_SUSPEND_LOCK) {
                                        dispatch_queue_t dq = ds->do_targetq;
                                        fprintf(debug_stream, "\t\t<br>DQ: %p refcnt 0x%x suspend 0x%x label: %s\n", dq, dq->do_ref_cnt, dq->do_suspend_cnt, dq->dq_label);
                                }
                        }
                        fprintf(debug_stream, "\t\t</ul>\n");
                        fprintf(debug_stream, "\t</li>\n");
                }
        }
        fprintf(debug_stream, "</ul>\n</body>\n</html>\n");
        fflush(debug_stream);
        fclose(debug_stream);
}

static void
_dispatch_kevent_debugger(void *context __attribute__((unused)))
{
        union {
                struct sockaddr_in sa_in;
                struct sockaddr sa;
        } sa_u = {
                .sa_in = {
                        .sin_family = AF_INET,
                        .sin_addr = { htonl(INADDR_LOOPBACK), },
                },
        };
        dispatch_source_t ds;
        const char *valstr;
        int val, r, fd, sock_opt = 1;
        socklen_t slen = sizeof(sa_u);

        if (issetugid()) {
                return;
        }
        valstr = getenv("LIBDISPATCH_DEBUGGER");
        if (!valstr) {
                return;
        }
        val = atoi(valstr);
        if (val == 2) {
                sa_u.sa_in.sin_addr.s_addr = 0;
        }
        fd = socket(PF_INET, SOCK_STREAM, 0);
        if (fd == -1) {
                dispatch_assume_zero(errno);
                return;
        }
        r = setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, (void *)&sock_opt, (socklen_t) sizeof sock_opt);
        if (r == -1) {
                dispatch_assume_zero(errno);
                goto out_bad;
        }
#if 0
        r = fcntl(fd, F_SETFL, O_NONBLOCK);
        if (r == -1) {
                dispatch_assume_zero(errno);
                goto out_bad;
        }
#endif
        r = bind(fd, &sa_u.sa, sizeof(sa_u));
        if (r == -1) {
                dispatch_assume_zero(errno);
                goto out_bad;
        }
        r = listen(fd, SOMAXCONN);
        if (r == -1) {
                dispatch_assume_zero(errno);
                goto out_bad;
        }
        r = getsockname(fd, &sa_u.sa, &slen);
        if (r == -1) {
                dispatch_assume_zero(errno);
                goto out_bad;
        }
        ds = dispatch_source_read_create_f(fd, NULL, &_dispatch_mgr_q, (void *)(long)fd, _dispatch_kevent_debugger2);
        if (dispatch_assume(ds)) {
                _dispatch_log("LIBDISPATCH: debug port: %hu", ntohs(sa_u.sa_in.sin_port));
                return;
        }
out_bad:
        close(fd);
}

void
_dispatch_source_drain_kevent(struct kevent *ke)
{
        static dispatch_once_t pred;
        dispatch_kevent_t dk = ke->udata;
        dispatch_source_t dsi;

        dispatch_once_f(&pred, NULL, _dispatch_kevent_debugger);

        dispatch_debug_kevents(ke, 1, __func__);

        if (ke->filter == EVFILT_MACHPORT) {
                return _dispatch_drain_mach_messages(ke);
        }
        dispatch_assert(dk);

        if (ke->flags & EV_ONESHOT) {
                dk->dk_kevent.flags |= EV_ONESHOT;
        }

        TAILQ_FOREACH(dsi, &dk->dk_sources, ds_list) {
                _dispatch_source_merge_kevent(dsi, ke);
        }
}

static void
_dispatch_kevent_dispose(dispatch_kevent_t dk)
{
        uintptr_t key;

        switch (dk->dk_kevent.filter) {
        case DISPATCH_EVFILT_TIMER:
        case DISPATCH_EVFILT_CUSTOM_ADD:
        case DISPATCH_EVFILT_CUSTOM_OR:
                // these sources live on statically allocated lists
                return;
        case EVFILT_MACHPORT:
                _dispatch_kevent_machport_resume(dk, 0, dk->dk_kevent.fflags);
                break;
        case EVFILT_PROC:
                if (dk->dk_kevent.flags & EV_ONESHOT) {
                        break;  // implicitly deleted
                }
                // fall through
        default:
                if (~dk->dk_kevent.flags & EV_DELETE) {
                        dk->dk_kevent.flags |= EV_DELETE;
                        _dispatch_update_kq(&dk->dk_kevent);
                }
                break;
        }

        if (dk->dk_kevent.filter == EVFILT_MACHPORT) {
                key = MACH_PORT_INDEX(dk->dk_kevent.ident);
        } else {
                key = dk->dk_kevent.ident;
        }

        TAILQ_REMOVE(&_dispatch_sources[DSL_HASH(key)], dk, dk_list);
        free(dk);
}

void
_dispatch_kevent_release(dispatch_source_t ds)
{
        dispatch_kevent_t dk = ds->ds_dkev;
        dispatch_source_t dsi;
        uint32_t del_flags, fflags = 0;

        ds->ds_dkev = NULL;

        TAILQ_REMOVE(&dk->dk_sources, ds, ds_list);

        if (TAILQ_EMPTY(&dk->dk_sources)) {
                _dispatch_kevent_dispose(dk);
        } else {
                TAILQ_FOREACH(dsi, &dk->dk_sources, ds_list) {
                        fflags |= (uint32_t)dsi->ds_pending_data_mask;
                }
                del_flags = (uint32_t)ds->ds_pending_data_mask & ~fflags;
                if (del_flags) {
                        dk->dk_kevent.flags |= EV_ADD;
                        dk->dk_kevent.fflags = fflags;
                        _dispatch_kevent_resume(dk, 0, del_flags);
                }
        }

        ds->ds_is_armed = false;
        ds->ds_needs_rearm = false;     // re-arm is pointless and bad now
        _dispatch_release(ds);  // the retain is done at creation time
}

void
_dispatch_source_merge_kevent(dispatch_source_t ds, const struct kevent *ke)
{
        struct kevent fake;

        if ((ds->ds_atomic_flags & DSF_CANCELED) || (ds->do_xref_cnt == 0)) {
                return;
        }

        // EVFILT_PROC may fail with ESRCH when the process exists but is a zombie.
        // We simulate an exit event in this case. <rdar://problem/5067725>
        if (ke->flags & EV_ERROR) {
                if (ke->filter == EVFILT_PROC && ke->data == ESRCH) {
                        fake = *ke;
                        fake.flags &= ~EV_ERROR;
                        fake.fflags = NOTE_EXIT;
                        fake.data = 0;
                        ke = &fake;
                } else {
                        // log the unexpected error
                        dispatch_assume_zero(ke->data);
                        return;
                }
        }

        if (ds->ds_is_level) {
                // ke->data is signed and "negative available data" makes no sense
                // zero bytes happens when EV_EOF is set
                // 10A268 does not fail this assert with EVFILT_READ and a 10 GB file
                dispatch_assert(ke->data >= 0l);
                ds->ds_pending_data = ~ke->data;
        } else if (ds->ds_is_adder) {
                dispatch_atomic_add(&ds->ds_pending_data, ke->data);
        } else {
                dispatch_atomic_or(&ds->ds_pending_data, ke->fflags & ds->ds_pending_data_mask);
        }

        // EV_DISPATCH and EV_ONESHOT sources are no longer armed after delivery
        if (ds->ds_needs_rearm) {
                ds->ds_is_armed = false;
        }

        _dispatch_wakeup(ds);
}

void
_dispatch_source_latch_and_call(dispatch_source_t ds)
{
        unsigned long prev;

        if ((ds->ds_atomic_flags & DSF_CANCELED) || (ds->do_xref_cnt == 0)) {
                return;
        }
        prev = dispatch_atomic_xchg(&ds->ds_pending_data, 0);
        if (ds->ds_is_level) {
                ds->ds_data = ~prev;
        } else {
                ds->ds_data = prev;
        }
        if (dispatch_assume(prev)) {
                if (ds->ds_handler_func) {
                        ds->ds_handler_func(ds->ds_handler_ctxt, ds);
                }
        }
}

void
_dispatch_source_cancel_callout(dispatch_source_t ds)
{
        ds->ds_pending_data_mask = 0;
        ds->ds_pending_data = 0;
        ds->ds_data = 0;

#ifdef __BLOCKS__
        if (ds->ds_handler_is_block) {
                Block_release(ds->ds_handler_ctxt);
                ds->ds_handler_is_block = false;
                ds->ds_handler_func = NULL;
                ds->ds_handler_ctxt = NULL;
        }
#endif

        if (!ds->ds_cancel_handler) {
                return;
        }
        if (ds->ds_cancel_is_block) {
#ifdef __BLOCKS__
                dispatch_block_t b = ds->ds_cancel_handler;
                if (ds->ds_atomic_flags & DSF_CANCELED) {
                        b();
                }
                Block_release(ds->ds_cancel_handler);
                ds->ds_cancel_is_block = false;
#endif
        } else {
                dispatch_function_t f = ds->ds_cancel_handler;
                if (ds->ds_atomic_flags & DSF_CANCELED) {
                        f(ds->do_ctxt);
                }
        }
        ds->ds_cancel_handler = NULL;
}

const struct dispatch_source_vtable_s _dispatch_source_kevent_vtable = {
        .do_type = DISPATCH_SOURCE_KEVENT_TYPE,
        .do_kind = "kevent-source",
        .do_invoke = _dispatch_source_invoke,
        .do_dispose = _dispatch_source_dispose,
        .do_probe = _dispatch_source_probe,
        .do_debug = _dispatch_source_kevent_debug,
};

void
dispatch_source_merge_data(dispatch_source_t ds, unsigned long val)
{       
        struct kevent kev = {
                .fflags = (typeof(kev.fflags))val,
                .data = val,
        };

        dispatch_assert(ds->ds_dkev->dk_kevent.filter == DISPATCH_EVFILT_CUSTOM_ADD ||
                                        ds->ds_dkev->dk_kevent.filter == DISPATCH_EVFILT_CUSTOM_OR);

        _dispatch_source_merge_kevent(ds, &kev);
}

size_t
dispatch_source_debug_attr(dispatch_source_t ds, char* buf, size_t bufsiz)
{
        dispatch_queue_t target = ds->do_targetq;
        return snprintf(buf, bufsiz,
                        "target = %s[%p], pending_data = 0x%lx, pending_data_mask = 0x%lx, ",
                        target ? target->dq_label : "", target,
                        ds->ds_pending_data, ds->ds_pending_data_mask);
}

size_t
_dispatch_source_debug(dispatch_source_t ds, char* buf, size_t bufsiz)
{
        size_t offset = 0;
        offset += snprintf(&buf[offset], bufsiz - offset, "%s[%p] = { ", dx_kind(ds), ds);
        offset += dispatch_object_debug_attr(ds, &buf[offset], bufsiz - offset);
        offset += dispatch_source_debug_attr(ds, &buf[offset], bufsiz - offset);
        return offset;
}

#ifndef DISPATCH_NO_LEGACY
static void
dispatch_source_attr_dispose(dispatch_source_attr_t attr)
{
        // release the finalizer block if necessary
        dispatch_source_attr_set_finalizer(attr, NULL);
        _dispatch_dispose(attr);
}

static const struct dispatch_source_attr_vtable_s dispatch_source_attr_vtable = {
        .do_type = DISPATCH_SOURCE_ATTR_TYPE,
        .do_kind = "source-attr",
        .do_dispose = dispatch_source_attr_dispose,
};

dispatch_source_attr_t
dispatch_source_attr_create(void)
{
        dispatch_source_attr_t rval = calloc(1, sizeof(struct dispatch_source_attr_s));

        if (rval) {
                rval->do_vtable = &dispatch_source_attr_vtable;
                rval->do_next = DISPATCH_OBJECT_LISTLESS;
                rval->do_targetq = dispatch_get_global_queue(0, 0);
                rval->do_ref_cnt = 1;
                rval->do_xref_cnt = 1;
        }

        return rval;
}

void
dispatch_source_attr_set_finalizer_f(dispatch_source_attr_t attr,
        void *context, dispatch_source_finalizer_function_t finalizer)
{
#ifdef __BLOCKS__
        if (attr->finalizer_func == (void*)_dispatch_call_block_and_release2) {
                Block_release(attr->finalizer_ctxt);
        }
#endif

        attr->finalizer_ctxt = context;
        attr->finalizer_func = finalizer;
}

#ifdef __BLOCKS__
long
dispatch_source_attr_set_finalizer(dispatch_source_attr_t attr,
        dispatch_source_finalizer_t finalizer)
{
        void *ctxt;
        dispatch_source_finalizer_function_t func;

        if (finalizer) {
                if (!(ctxt = Block_copy(finalizer))) {
                        return 1;
                }
                func = (void *)_dispatch_call_block_and_release2;
        } else {
                ctxt = NULL;
                func = NULL;
        }

        dispatch_source_attr_set_finalizer_f(attr, ctxt, func);

        return 0;
}

dispatch_source_finalizer_t
dispatch_source_attr_get_finalizer(dispatch_source_attr_t attr)
{
        if (attr->finalizer_func == (void*)_dispatch_call_block_and_release2) {
                return (dispatch_source_finalizer_t)attr->finalizer_ctxt;
        } else if (attr->finalizer_func == NULL) {
                return NULL;
        } else {
                abort(); // finalizer is not a block...
        }
}
#endif

void
dispatch_source_attr_set_context(dispatch_source_attr_t attr, void *context)
{
        attr->context = context;
}

dispatch_source_attr_t
dispatch_source_attr_copy(dispatch_source_attr_t proto)
{
        dispatch_source_attr_t rval = NULL;

        if (proto && (rval = malloc(sizeof(struct dispatch_source_attr_s)))) {
                memcpy(rval, proto, sizeof(struct dispatch_source_attr_s));
#ifdef __BLOCKS__
                if (rval->finalizer_func == (void*)_dispatch_call_block_and_release2) {
                        rval->finalizer_ctxt = Block_copy(rval->finalizer_ctxt);
                }
#endif
        } else if (!proto) {
                rval = dispatch_source_attr_create();
        }
        return rval;
}
#endif /* DISPATCH_NO_LEGACY */


struct dispatch_source_type_s {
        struct kevent ke;
        uint64_t mask;
};

const struct dispatch_source_type_s _dispatch_source_type_timer = {
        .ke = {
                .filter = DISPATCH_EVFILT_TIMER,
        },
        .mask = DISPATCH_TIMER_INTERVAL|DISPATCH_TIMER_ONESHOT|DISPATCH_TIMER_ABSOLUTE|DISPATCH_TIMER_WALL_CLOCK,
};

const struct dispatch_source_type_s _dispatch_source_type_read = {
        .ke = {
                .filter = EVFILT_READ,
                .flags = EV_DISPATCH,
        },
};

const struct dispatch_source_type_s _dispatch_source_type_write = {
        .ke = {
                .filter = EVFILT_WRITE,
                .flags = EV_DISPATCH,
        },
};

const struct dispatch_source_type_s _dispatch_source_type_proc = {
        .ke = {
                .filter = EVFILT_PROC,
                .flags = EV_CLEAR,
        },
        .mask = NOTE_EXIT|NOTE_FORK|NOTE_EXEC|NOTE_SIGNAL|NOTE_REAP,
};

const struct dispatch_source_type_s _dispatch_source_type_signal = {
        .ke = {
                .filter = EVFILT_SIGNAL,
        },
};

const struct dispatch_source_type_s _dispatch_source_type_vnode = {
        .ke = {
                .filter = EVFILT_VNODE,
                .flags = EV_CLEAR,
        },
        .mask = NOTE_DELETE|NOTE_WRITE|NOTE_EXTEND|NOTE_ATTRIB|NOTE_LINK|NOTE_RENAME|NOTE_REVOKE|NOTE_NONE,
};

const struct dispatch_source_type_s _dispatch_source_type_vfs = {
        .ke = {
                .filter = EVFILT_FS,
                .flags = EV_CLEAR,
        },
        .mask = VQ_NOTRESP|VQ_NEEDAUTH|VQ_LOWDISK|VQ_MOUNT|VQ_UNMOUNT|VQ_DEAD|VQ_ASSIST|VQ_NOTRESPLOCK|VQ_UPDATE|VQ_VERYLOWDISK,
};

const struct dispatch_source_type_s _dispatch_source_type_mach_send = {
        .ke = {
                .filter = EVFILT_MACHPORT,
                .flags = EV_DISPATCH,
                .fflags = DISPATCH_MACHPORT_DEAD,
        },
        .mask = DISPATCH_MACH_SEND_DEAD,
};

const struct dispatch_source_type_s _dispatch_source_type_mach_recv = {
        .ke = {
                .filter = EVFILT_MACHPORT,
                .flags = EV_DISPATCH,
                .fflags = DISPATCH_MACHPORT_RECV,
        },
};

const struct dispatch_source_type_s _dispatch_source_type_data_add = {
        .ke = {
                .filter = DISPATCH_EVFILT_CUSTOM_ADD,
        },
};

const struct dispatch_source_type_s _dispatch_source_type_data_or = {
        .ke = {
                .filter = DISPATCH_EVFILT_CUSTOM_OR,
                .flags = EV_CLEAR,
                .fflags = ~0,
        },
};

dispatch_source_t
dispatch_source_create(dispatch_source_type_t type,
        uintptr_t handle,
        unsigned long mask,
        dispatch_queue_t q)
{
        const struct kevent *proto_kev = &type->ke;
        dispatch_source_t ds = NULL;
        dispatch_kevent_t dk = NULL;

        // input validation
        if (type == NULL || (mask & ~type->mask)) {
                goto out_bad;
        }

        switch (type->ke.filter) {
        case EVFILT_SIGNAL:
                if (handle >= NSIG) {
                        goto out_bad;
                }
                break;
        case EVFILT_FS:
        case DISPATCH_EVFILT_CUSTOM_ADD:
        case DISPATCH_EVFILT_CUSTOM_OR:
        case DISPATCH_EVFILT_TIMER:
                if (handle) {
                        goto out_bad;
                }
                break;
        default:
                break;
        }
        
        ds = calloc(1ul, sizeof(struct dispatch_source_s));
        if (slowpath(!ds)) {
                goto out_bad;
        }
        dk = calloc(1ul, sizeof(struct dispatch_kevent_s));
        if (slowpath(!dk)) {
                goto out_bad;
        }

        dk->dk_kevent = *proto_kev;
        dk->dk_kevent.ident = handle;
        dk->dk_kevent.flags |= EV_ADD|EV_ENABLE;
        dk->dk_kevent.fflags |= (uint32_t)mask;
        dk->dk_kevent.udata = dk;
        TAILQ_INIT(&dk->dk_sources);

        // Initialize as a queue first, then override some settings below.
        _dispatch_queue_init((dispatch_queue_t)ds);
        strlcpy(ds->dq_label, "source", sizeof(ds->dq_label));

        // Dispatch Object
        ds->do_vtable = &_dispatch_source_kevent_vtable;
        ds->do_ref_cnt++; // the reference the manger queue holds
        ds->do_suspend_cnt = DISPATCH_OBJECT_SUSPEND_INTERVAL;
        // do_targetq will be retained below, past point of no-return
        ds->do_targetq = q;

        // Dispatch Source
        ds->ds_ident_hack = dk->dk_kevent.ident;
        ds->ds_dkev = dk;
        ds->ds_pending_data_mask = dk->dk_kevent.fflags;
        if ((EV_DISPATCH|EV_ONESHOT) & proto_kev->flags) {
                if (proto_kev->filter != EVFILT_MACHPORT) {
                        ds->ds_is_level = true;
                }
                ds->ds_needs_rearm = true;
        } else if (!(EV_CLEAR & proto_kev->flags)) {
                // we cheat and use EV_CLEAR to mean a "flag thingy"
                ds->ds_is_adder = true;
        }
        
        // If its a timer source, it needs to be re-armed
        if (type->ke.filter == DISPATCH_EVFILT_TIMER) {
                ds->ds_needs_rearm = true;
        }
        
        dispatch_assert(!(ds->ds_is_level && ds->ds_is_adder));
#if DISPATCH_DEBUG
        dispatch_debug(ds, __FUNCTION__);
#endif

        // Some sources require special processing
        if (type == DISPATCH_SOURCE_TYPE_MACH_SEND) {
                static dispatch_once_t pred;
                dispatch_once_f(&pred, NULL, _dispatch_mach_notify_source_init);
        } else if (type == DISPATCH_SOURCE_TYPE_TIMER) {
                ds->ds_timer.flags = mask;
        }

        _dispatch_retain(ds->do_targetq);
        return ds;
        
out_bad:
        free(ds);
        free(dk);
        return NULL;
}

// 6618342 Contact the team that owns the Instrument DTrace probe before renaming this symbol
static void
_dispatch_source_set_event_handler2(void *context)
{
        struct Block_layout *bl = context;

        dispatch_source_t ds = (dispatch_source_t)_dispatch_queue_get_current();
        dispatch_assert(ds->do_vtable == &_dispatch_source_kevent_vtable);
        
        if (ds->ds_handler_is_block && ds->ds_handler_ctxt) {
                Block_release(ds->ds_handler_ctxt);
        }
        ds->ds_handler_func = bl ? (void *)bl->invoke : NULL;
        ds->ds_handler_ctxt = bl;
        ds->ds_handler_is_block = true;
}

void
dispatch_source_set_event_handler(dispatch_source_t ds, dispatch_block_t handler)
{
        dispatch_assert(!ds->ds_is_legacy);
        handler = _dispatch_Block_copy(handler);
        dispatch_barrier_async_f((dispatch_queue_t)ds,
                handler, _dispatch_source_set_event_handler2);
}

static void
_dispatch_source_set_event_handler_f(void *context)
{
        dispatch_source_t ds = (dispatch_source_t)_dispatch_queue_get_current();
        dispatch_assert(ds->do_vtable == &_dispatch_source_kevent_vtable);
        
        if (ds->ds_handler_is_block && ds->ds_handler_ctxt) {
                Block_release(ds->ds_handler_ctxt);
        }
        ds->ds_handler_func = context;
        ds->ds_handler_ctxt = ds->do_ctxt;
        ds->ds_handler_is_block = false;
}

void
dispatch_source_set_event_handler_f(dispatch_source_t ds,
        dispatch_function_t handler)
{
        dispatch_assert(!ds->ds_is_legacy);
        dispatch_barrier_async_f((dispatch_queue_t)ds,
                handler, _dispatch_source_set_event_handler_f);
}

// 6618342 Contact the team that owns the Instrument DTrace probe before renaming this symbol
static void
_dispatch_source_set_cancel_handler2(void *context)
{
        dispatch_source_t ds = (dispatch_source_t)_dispatch_queue_get_current();
        dispatch_assert(ds->do_vtable == &_dispatch_source_kevent_vtable);
        
        if (ds->ds_cancel_is_block && ds->ds_cancel_handler) {
                Block_release(ds->ds_cancel_handler);
        }
        ds->ds_cancel_handler = context;
        ds->ds_cancel_is_block = true;
}

void
dispatch_source_set_cancel_handler(dispatch_source_t ds,
        dispatch_block_t handler)
{
        dispatch_assert(!ds->ds_is_legacy);
        handler = _dispatch_Block_copy(handler);
        dispatch_barrier_async_f((dispatch_queue_t)ds,
                                                         handler, _dispatch_source_set_cancel_handler2);
}

static void
_dispatch_source_set_cancel_handler_f(void *context)
{
        dispatch_source_t ds = (dispatch_source_t)_dispatch_queue_get_current();
        dispatch_assert(ds->do_vtable == &_dispatch_source_kevent_vtable);
        
        if (ds->ds_cancel_is_block && ds->ds_cancel_handler) {
                Block_release(ds->ds_cancel_handler);
        }
        ds->ds_cancel_handler = context;
        ds->ds_cancel_is_block = false;
}

void
dispatch_source_set_cancel_handler_f(dispatch_source_t ds,
        dispatch_function_t handler)
{
        dispatch_assert(!ds->ds_is_legacy);
        dispatch_barrier_async_f((dispatch_queue_t)ds,
                                                         handler, _dispatch_source_set_cancel_handler_f);
}

#ifndef DISPATCH_NO_LEGACY
// 6618342 Contact the team that owns the Instrument DTrace probe before renaming this symbol
dispatch_source_t
_dispatch_source_create2(dispatch_source_t ds,
        dispatch_source_attr_t attr,
        void *context,
        dispatch_source_handler_function_t handler)
{
        if (ds == NULL || handler == NULL) {
                return NULL;
        }

        ds->ds_is_legacy = true;

        ds->ds_handler_func = handler;
        ds->ds_handler_ctxt = context;
                
        if (attr && attr != DISPATCH_SOURCE_CREATE_SUSPENDED) {
                ds->dq_finalizer_ctxt = attr->finalizer_ctxt;
                ds->dq_finalizer_func = (typeof(ds->dq_finalizer_func))attr->finalizer_func;
                ds->do_ctxt = attr->context;
        }
#ifdef __BLOCKS__
        if (ds->dq_finalizer_func == (void*)_dispatch_call_block_and_release2) {
                ds->dq_finalizer_ctxt = Block_copy(ds->dq_finalizer_ctxt);
                if (!ds->dq_finalizer_ctxt) {
                        goto out_bad;
                }
        }
        if (handler == _dispatch_source_call_block) {
                struct Block_layout *bl = ds->ds_handler_ctxt = Block_copy(context);
                if (!ds->ds_handler_ctxt) {
                        if (ds->dq_finalizer_func == (void*)_dispatch_call_block_and_release2) {
                                Block_release(ds->dq_finalizer_ctxt);
                        }
                        goto out_bad;
                }
                ds->ds_handler_func = (void *)bl->invoke;
                ds->ds_handler_is_block = true;
        }

        // all legacy sources get a cancellation event on the normal event handler.
        dispatch_source_handler_function_t func = ds->ds_handler_func;
        dispatch_source_handler_t block = ds->ds_handler_ctxt;
        void *ctxt = ds->ds_handler_ctxt;
        bool handler_is_block = ds->ds_handler_is_block;
        
        ds->ds_cancel_is_block = true;
        if (handler_is_block) {
                ds->ds_cancel_handler = _dispatch_Block_copy(^{
                        block(ds);
                });
        } else {
                ds->ds_cancel_handler = _dispatch_Block_copy(^{
                        func(ctxt, ds);
                });
        }
#endif
        if (attr != DISPATCH_SOURCE_CREATE_SUSPENDED) {
                dispatch_resume(ds);
        }

        return ds;

out_bad:
        free(ds);
        return NULL;
}

long
dispatch_source_get_error(dispatch_source_t ds, long *err_out)
{
        // 6863892 don't report ECANCELED until kevent is unregistered
        if ((ds->ds_atomic_flags & DSF_CANCELED) && !ds->ds_dkev) {
                if (err_out) {
                        *err_out = ECANCELED;
                }
                return DISPATCH_ERROR_DOMAIN_POSIX;
        } else {
                return DISPATCH_ERROR_DOMAIN_NO_ERROR;
        }
}
#endif /* DISPATCH_NO_LEGACY */

// Updates the ordered list of timers based on next fire date for changes to ds.
// Should only be called from the context of _dispatch_mgr_q.
void
_dispatch_timer_list_update(dispatch_source_t ds)
{
        dispatch_source_t dsi = NULL;
        int idx;
        
        dispatch_assert(_dispatch_queue_get_current() == &_dispatch_mgr_q);

        // do not reschedule timers unregistered with _dispatch_kevent_release()
        if (!ds->ds_dkev) {
                return;
        }

        // Ensure the source is on the global kevent lists before it is removed and
        // readded below.
        _dispatch_kevent_merge(ds);
        
        TAILQ_REMOVE(&ds->ds_dkev->dk_sources, ds, ds_list);

        // change the list if the clock type has changed
        if (ds->ds_timer.flags & DISPATCH_TIMER_WALL_CLOCK) {
                idx = DISPATCH_TIMER_INDEX_WALL;
        } else {
                idx = DISPATCH_TIMER_INDEX_MACH;
        }
        ds->ds_dkev = &_dispatch_kevent_timer[idx];

        if (ds->ds_timer.target) {
                TAILQ_FOREACH(dsi, &ds->ds_dkev->dk_sources, ds_list) {
                        if (dsi->ds_timer.target == 0 || ds->ds_timer.target < dsi->ds_timer.target) {
                                break;
                        }
                }
        }
        
        if (dsi) {
                TAILQ_INSERT_BEFORE(dsi, ds, ds_list);
        } else {
                TAILQ_INSERT_TAIL(&ds->ds_dkev->dk_sources, ds, ds_list);
        }
}

static void
_dispatch_run_timers2(unsigned int timer)
{
        dispatch_source_t ds;
        uint64_t now, missed;

        if (timer == DISPATCH_TIMER_INDEX_MACH) {
                now = mach_absolute_time();
        } else {
                now = _dispatch_get_nanoseconds();
        }

        while ((ds = TAILQ_FIRST(&_dispatch_kevent_timer[timer].dk_sources))) {
                // We may find timers on the wrong list due to a pending update from
                // dispatch_source_set_timer. Force an update of the list in that case.
                if (timer != ds->ds_ident_hack) {
                        _dispatch_timer_list_update(ds);
                        continue;
                }
                if (!ds->ds_timer.target) {
                        // no configured timers on the list
                        break;
                }
                if (ds->ds_timer.target > now) {
                        // Done running timers for now.
                        break;
                }

                if (ds->ds_timer.flags & (DISPATCH_TIMER_ONESHOT|DISPATCH_TIMER_ABSOLUTE)) {
                        dispatch_atomic_inc(&ds->ds_pending_data);
                        ds->ds_timer.target = 0;
                } else {
                        // Calculate number of missed intervals.
                        missed = (now - ds->ds_timer.target) / ds->ds_timer.interval;
                        dispatch_atomic_add(&ds->ds_pending_data, missed + 1);
                        ds->ds_timer.target += (missed + 1) * ds->ds_timer.interval;
                }

                _dispatch_timer_list_update(ds);
                _dispatch_wakeup(ds);
        }
}

void
_dispatch_run_timers(void)
{
        unsigned int i;
        for (i = 0; i < DISPATCH_TIMER_COUNT; i++) {
                _dispatch_run_timers2(i);
        }
}

#if defined(__i386__) || defined(__x86_64__)
// these architectures always return mach_absolute_time() in nanoseconds
#define _dispatch_convert_mach2nano(x) (x)
#define _dispatch_convert_nano2mach(x) (x)
#else
static mach_timebase_info_data_t tbi;
static dispatch_once_t tbi_pred;

static void
_dispatch_convert_init(void *context __attribute__((unused)))
{
        dispatch_assume_zero(mach_timebase_info(&tbi));
}

static uint64_t
_dispatch_convert_mach2nano(uint64_t val)
{
#ifdef __LP64__
        __uint128_t tmp;
#else
        long double tmp;
#endif

        dispatch_once_f(&tbi_pred, NULL, _dispatch_convert_init);

        tmp = val;
        tmp *= tbi.numer;
        tmp /= tbi.denom;

        return tmp;
}

static uint64_t
_dispatch_convert_nano2mach(uint64_t val)
{
#ifdef __LP64__
        __uint128_t tmp;
#else
        long double tmp;
#endif

        dispatch_once_f(&tbi_pred, NULL, _dispatch_convert_init);

        tmp = val;
        tmp *= tbi.denom;
        tmp /= tbi.numer;

        return tmp;
}
#endif

// approx 1 year (60s * 60m * 24h * 365d)
#define FOREVER_SEC 3153600l
#define FOREVER_NSEC 31536000000000000ull

struct timespec *
_dispatch_get_next_timer_fire(struct timespec *howsoon)
{
        // <rdar://problem/6459649>
        // kevent(2) does not allow large timeouts, so we use a long timeout
        // instead (approximately 1 year).
        dispatch_source_t ds = NULL;
        unsigned int timer;
        uint64_t now, delta_tmp, delta = UINT64_MAX;

        // We are looking for the first unsuspended timer which has its target
        // time set. Given timers are kept in order, if we hit an timer that's
        // unset there's no point in continuing down the list.
        for (timer = 0; timer < DISPATCH_TIMER_COUNT; timer++) {
                TAILQ_FOREACH(ds, &_dispatch_kevent_timer[timer].dk_sources, ds_list) {
                        if (!ds->ds_timer.target) {
                                break;
                        }
                        if (DISPATCH_OBJECT_SUSPENDED(ds)) {
                                ds->ds_is_armed = false;
                        } else {
                                break;
                        }
                }

                if (!ds || !ds->ds_timer.target) {
                        continue;
                }
                                
                if (ds->ds_timer.flags & DISPATCH_TIMER_WALL_CLOCK) {
                        now = _dispatch_get_nanoseconds();
                } else {
                        now = mach_absolute_time();
                }
                if (ds->ds_timer.target <= now) {
                        howsoon->tv_sec = 0;
                        howsoon->tv_nsec = 0;
                        return howsoon;
                }

                // the subtraction cannot go negative because the previous "if"
                // verified that the target is greater than now.
                delta_tmp = ds->ds_timer.target - now;
                if (!(ds->ds_timer.flags & DISPATCH_TIMER_WALL_CLOCK)) {
                        delta_tmp = _dispatch_convert_mach2nano(delta_tmp);
                }
                if (delta_tmp < delta) {
                        delta = delta_tmp;
                }
        }
        if (slowpath(delta > FOREVER_NSEC)) {
                return NULL;
        } else {
                howsoon->tv_sec = (time_t)(delta / NSEC_PER_SEC);
                howsoon->tv_nsec = (long)(delta % NSEC_PER_SEC);
        }
        return howsoon;
}

struct dispatch_set_timer_params {
        dispatch_source_t ds;
        uintptr_t ident;
        struct dispatch_timer_source_s values;
};

// To be called from the context of the _dispatch_mgr_q
static void
_dispatch_source_set_timer2(void *context)
{
        struct dispatch_set_timer_params *params = context;
        dispatch_source_t ds = params->ds;
        ds->ds_ident_hack = params->ident;
        ds->ds_timer = params->values;
        _dispatch_timer_list_update(ds);
        dispatch_resume(ds);
        dispatch_release(ds);
        free(params);
}

void
dispatch_source_set_timer(dispatch_source_t ds,
        dispatch_time_t start,
        uint64_t interval,
        uint64_t leeway)
{
        struct dispatch_set_timer_params *params;
        
        // we use zero internally to mean disabled
        if (interval == 0) {
                interval = 1;
        } else if ((int64_t)interval < 0) {
                // 6866347 - make sure nanoseconds won't overflow
                interval = INT64_MAX;
        }

        // Suspend the source so that it doesn't fire with pending changes
        // The use of suspend/resume requires the external retain/release
        dispatch_retain(ds);
        dispatch_suspend(ds);
        
        if (start == DISPATCH_TIME_NOW) {
                start = mach_absolute_time();
        } else if (start == DISPATCH_TIME_FOREVER) {
                start = INT64_MAX;
        }

        while (!(params = malloc(sizeof(struct dispatch_set_timer_params)))) {
                sleep(1);
        }

        params->ds = ds;
        params->values.flags = ds->ds_timer.flags;

        if ((int64_t)start < 0) {
                // wall clock
                params->ident = DISPATCH_TIMER_INDEX_WALL;
                params->values.start = -((int64_t)start);
                params->values.target = -((int64_t)start);
                params->values.interval = interval;
                params->values.leeway = leeway;
                params->values.flags |= DISPATCH_TIMER_WALL_CLOCK;
        } else {
                // mach clock
                params->ident = DISPATCH_TIMER_INDEX_MACH;
                params->values.start = start;
                params->values.target = start;
                params->values.interval = _dispatch_convert_nano2mach(interval);
                params->values.leeway = _dispatch_convert_nano2mach(leeway);
                params->values.flags &= ~DISPATCH_TIMER_WALL_CLOCK;
        }

        dispatch_barrier_async_f(&_dispatch_mgr_q, params, _dispatch_source_set_timer2);
}

#ifndef DISPATCH_NO_LEGACY
// LEGACY
long
dispatch_source_timer_set_time(dispatch_source_t ds, uint64_t nanoseconds, uint64_t leeway)
{
        dispatch_time_t start;
        if (nanoseconds == 0) {
                nanoseconds = 1;
        }
        if (ds->ds_timer.flags == (DISPATCH_TIMER_ABSOLUTE|DISPATCH_TIMER_WALL_CLOCK)) {
                static const struct timespec t0;
                start = dispatch_walltime(&t0, nanoseconds);
        } else if (ds->ds_timer.flags & DISPATCH_TIMER_WALL_CLOCK) {
                start = dispatch_walltime(DISPATCH_TIME_NOW, nanoseconds);
        } else {
                start = dispatch_time(DISPATCH_TIME_NOW, nanoseconds);
        }
        if (ds->ds_timer.flags & (DISPATCH_TIMER_ABSOLUTE|DISPATCH_TIMER_ONESHOT)) {
                // 6866347 - make sure nanoseconds won't overflow
                nanoseconds = INT64_MAX; // non-repeating (~292 years)
        }
        dispatch_source_set_timer(ds, start, nanoseconds, leeway);
        return 0;
}

// LEGACY
uint64_t
dispatch_event_get_nanoseconds(dispatch_source_t ds)
{
        if (ds->ds_timer.flags & DISPATCH_TIMER_WALL_CLOCK) {
                return ds->ds_timer.interval;
        } else {
                return _dispatch_convert_mach2nano(ds->ds_timer.interval);
        }
}
#endif /* DISPATCH_NO_LEGACY */

static dispatch_source_t _dispatch_mach_notify_source;
static mach_port_t _dispatch_port_set;
static mach_port_t _dispatch_event_port;

#define _DISPATCH_IS_POWER_OF_TWO(v)    (!(v & (v - 1)) && v)
#define _DISPATCH_HASH(x, y)    (_DISPATCH_IS_POWER_OF_TWO(y) ? (MACH_PORT_INDEX(x) & ((y) - 1)) : (MACH_PORT_INDEX(x) % (y)))

#define _DISPATCH_MACHPORT_HASH_SIZE 32
#define _DISPATCH_MACHPORT_HASH(x)    _DISPATCH_HASH((x), _DISPATCH_MACHPORT_HASH_SIZE)

static void _dispatch_port_set_init(void *);
static mach_port_t _dispatch_get_port_set(void);

void
_dispatch_drain_mach_messages(struct kevent *ke)
{
        dispatch_source_t dsi;
        dispatch_kevent_t dk;
        struct kevent ke2;

        if (!dispatch_assume(ke->data)) {
                return;
        }
        dk = _dispatch_kevent_find(ke->data, EVFILT_MACHPORT);
        if (!dispatch_assume(dk)) {
                return;
        }
        _dispatch_kevent_machport_disable(dk);  // emulate EV_DISPATCH

        EV_SET(&ke2, ke->data, EVFILT_MACHPORT, EV_ADD|EV_ENABLE|EV_DISPATCH, DISPATCH_MACHPORT_RECV, 0, dk);

        TAILQ_FOREACH(dsi, &dk->dk_sources, ds_list) {
                _dispatch_source_merge_kevent(dsi, &ke2);
        }
}

void
_dispatch_port_set_init(void *context __attribute__((unused)))
{
        struct kevent kev = {
                .filter = EVFILT_MACHPORT,
                .flags = EV_ADD,
        };
        kern_return_t kr;

        kr = mach_port_allocate(mach_task_self(), MACH_PORT_RIGHT_PORT_SET, &_dispatch_port_set);
        DISPATCH_VERIFY_MIG(kr);
        dispatch_assume_zero(kr);
        kr = mach_port_allocate(mach_task_self(), MACH_PORT_RIGHT_RECEIVE, &_dispatch_event_port);
        DISPATCH_VERIFY_MIG(kr);
        dispatch_assume_zero(kr);
        kr = mach_port_move_member(mach_task_self(), _dispatch_event_port, _dispatch_port_set);
        DISPATCH_VERIFY_MIG(kr);
        dispatch_assume_zero(kr);

        kev.ident = _dispatch_port_set;

        _dispatch_update_kq(&kev);
}

mach_port_t
_dispatch_get_port_set(void)
{
        static dispatch_once_t pred;

        dispatch_once_f(&pred, NULL, _dispatch_port_set_init);

        return _dispatch_port_set;
}

void
_dispatch_kevent_machport_resume(dispatch_kevent_t dk, uint32_t new_flags, uint32_t del_flags)
{
        mach_port_t previous, port = (mach_port_t)dk->dk_kevent.ident;
        kern_return_t kr;

        if ((new_flags & DISPATCH_MACHPORT_RECV) || (!new_flags && !del_flags && dk->dk_kevent.fflags & DISPATCH_MACHPORT_RECV)) {
                _dispatch_kevent_machport_enable(dk);
        }
        if (new_flags & DISPATCH_MACHPORT_DEAD) {
                kr = mach_port_request_notification(mach_task_self(), port, MACH_NOTIFY_DEAD_NAME, 1,
                                _dispatch_event_port, MACH_MSG_TYPE_MAKE_SEND_ONCE, &previous);
                DISPATCH_VERIFY_MIG(kr);

        
                switch(kr) {
                        case KERN_INVALID_NAME:
                        case KERN_INVALID_RIGHT:
                                // Supress errors 
                                break;
                        default:
                        // Else, we dont expect any errors from mach. Log any errors if we do
                        if (dispatch_assume_zero(kr)) {
                                // log the error
                        } else if (dispatch_assume_zero(previous)) {
                                // Another subsystem has beat libdispatch to requesting the Mach
                                // dead-name notification on this port. We should technically cache the
                                // previous port and message it when the kernel messages our port. Or
                                // we can just say screw those subsystems and drop the previous port.
                                // They should adopt libdispatch :-P
                                kr = mach_port_deallocate(mach_task_self(), previous);
                                DISPATCH_VERIFY_MIG(kr);
                                dispatch_assume_zero(kr);
                        }
                }
        }

        if (del_flags & DISPATCH_MACHPORT_RECV) {
                _dispatch_kevent_machport_disable(dk);
        }
        if (del_flags & DISPATCH_MACHPORT_DEAD) {
                kr = mach_port_request_notification(mach_task_self(), (mach_port_t)dk->dk_kevent.ident,
                                MACH_NOTIFY_DEAD_NAME, 1, MACH_PORT_NULL, MACH_MSG_TYPE_MAKE_SEND_ONCE, &previous);
                DISPATCH_VERIFY_MIG(kr);

                switch (kr) {
                        case KERN_INVALID_NAME:
                        case KERN_INVALID_RIGHT:
                        case KERN_INVALID_ARGUMENT:
                                break;
                        default:
                        if (dispatch_assume_zero(kr)) {
                                // log the error
                        } else if (previous) {
                                // the kernel has not consumed the right yet
                                dispatch_assume_zero(_dispatch_send_consume_send_once_right(previous));
                        }
                }
        }
}

void
_dispatch_kevent_machport_enable(dispatch_kevent_t dk)
{
        mach_port_t mp = (mach_port_t)dk->dk_kevent.ident;
        kern_return_t kr;

        kr = mach_port_move_member(mach_task_self(), mp, _dispatch_get_port_set());
        DISPATCH_VERIFY_MIG(kr);
        switch (kr) {
        case KERN_INVALID_NAME:
#if DISPATCH_DEBUG
                _dispatch_log("Corruption: Mach receive right 0x%x destroyed prematurely", mp);
#endif
                break;
        default:
                dispatch_assume_zero(kr);
        }
}

void
_dispatch_kevent_machport_disable(dispatch_kevent_t dk)
{
        mach_port_t mp = (mach_port_t)dk->dk_kevent.ident;
        kern_return_t kr;

        kr = mach_port_move_member(mach_task_self(), mp, 0);
        DISPATCH_VERIFY_MIG(kr);
        switch (kr) {
        case KERN_INVALID_RIGHT:
        case KERN_INVALID_NAME:
#if DISPATCH_DEBUG
                _dispatch_log("Corruption: Mach receive right 0x%x destroyed prematurely", mp);
#endif
                break;
        case 0:
                break;
        default:
                dispatch_assume_zero(kr);
                break;
        }
}

#define _DISPATCH_MIN_MSG_SZ (8ul * 1024ul - MAX_TRAILER_SIZE)
#ifndef DISPATCH_NO_LEGACY
dispatch_source_t
dispatch_source_mig_create(mach_port_t mport, size_t max_msg_size, dispatch_source_attr_t attr,
        dispatch_queue_t dq, dispatch_mig_callback_t mig_callback)
{
        if (max_msg_size < _DISPATCH_MIN_MSG_SZ) {
                max_msg_size = _DISPATCH_MIN_MSG_SZ;
        }
        return dispatch_source_machport_create(mport, DISPATCH_MACHPORT_RECV, attr, dq,
        ^(dispatch_source_t ds) {
                if (!dispatch_source_get_error(ds, NULL)) {
                        if (dq->dq_width != 1) {
                                dispatch_retain(ds);    // this is a shim -- use the external retain
                                dispatch_async(dq, ^{
                                        dispatch_mig_server(ds, max_msg_size, mig_callback);
                                        dispatch_release(ds);   // this is a shim -- use the external release
                                });
                        } else {
                                dispatch_mig_server(ds, max_msg_size, mig_callback);
                        }
                }
        });     
}
#endif /* DISPATCH_NO_LEGACY */

static void
_dispatch_mach_notify_source_init(void *context __attribute__((unused)))
{
        size_t maxsz = sizeof(union __RequestUnion___dispatch_send_libdispatch_internal_protocol_subsystem);

        if (sizeof(union __ReplyUnion___dispatch_libdispatch_internal_protocol_subsystem) > maxsz) {
                maxsz = sizeof(union __ReplyUnion___dispatch_libdispatch_internal_protocol_subsystem);
        }

        _dispatch_get_port_set();

        _dispatch_mach_notify_source = dispatch_source_mig_create(_dispatch_event_port,
                        maxsz, NULL, &_dispatch_mgr_q, libdispatch_internal_protocol_server);

        dispatch_assert(_dispatch_mach_notify_source);
}

kern_return_t
_dispatch_mach_notify_port_deleted(mach_port_t notify __attribute__((unused)), mach_port_name_t name)
{
        dispatch_source_t dsi;
        dispatch_kevent_t dk;
        struct kevent kev;

#if DISPATCH_DEBUG
        _dispatch_log("Corruption: Mach send/send-once/dead-name right 0x%x deleted prematurely", name);
#endif

        dk = _dispatch_kevent_find(name, EVFILT_MACHPORT);
        if (!dk) {
                goto out;
        }

        EV_SET(&kev, name, EVFILT_MACHPORT, EV_ADD|EV_ENABLE|EV_DISPATCH|EV_EOF, DISPATCH_MACHPORT_DELETED, 0, dk);

        TAILQ_FOREACH(dsi, &dk->dk_sources, ds_list) {
                _dispatch_source_merge_kevent(dsi, &kev);
                // this can never happen again
                // this must happen after the merge
                // this may be racy in the future, but we don't provide a 'setter' API for the mask yet
                dsi->ds_pending_data_mask &= ~DISPATCH_MACHPORT_DELETED;
        }

        // no more sources have this flag
        dk->dk_kevent.fflags &= ~DISPATCH_MACHPORT_DELETED;

out:
        return KERN_SUCCESS;
}

kern_return_t
_dispatch_mach_notify_port_destroyed(mach_port_t notify __attribute__((unused)), mach_port_t name)
{
        kern_return_t kr;
        // this function should never be called
        dispatch_assume_zero(name);
        kr = mach_port_mod_refs(mach_task_self(), name, MACH_PORT_RIGHT_RECEIVE, -1);
        DISPATCH_VERIFY_MIG(kr);
        dispatch_assume_zero(kr);
        return KERN_SUCCESS;
}

kern_return_t
_dispatch_mach_notify_no_senders(mach_port_t notify, mach_port_mscount_t mscnt __attribute__((unused)))
{
        // this function should never be called
        dispatch_assume_zero(notify);
        return KERN_SUCCESS;
}

kern_return_t
_dispatch_mach_notify_send_once(mach_port_t notify __attribute__((unused)))
{
        // we only register for dead-name notifications
        // some code deallocated our send-once right without consuming it
#if DISPATCH_DEBUG
        _dispatch_log("Corruption: An app/library deleted a libdispatch dead-name notification");
#endif
        return KERN_SUCCESS;
}

kern_return_t
_dispatch_mach_notify_dead_name(mach_port_t notify __attribute__((unused)), mach_port_name_t name)
{
        dispatch_source_t dsi;
        dispatch_kevent_t dk;
        struct kevent kev;
        kern_return_t kr;

        dk = _dispatch_kevent_find(name, EVFILT_MACHPORT);
        if (!dk) {
                goto out;
        }

        EV_SET(&kev, name, EVFILT_MACHPORT, EV_ADD|EV_ENABLE|EV_DISPATCH|EV_EOF, DISPATCH_MACHPORT_DEAD, 0, dk);

        TAILQ_FOREACH(dsi, &dk->dk_sources, ds_list) {
                _dispatch_source_merge_kevent(dsi, &kev);
                // this can never happen again
                // this must happen after the merge
                // this may be racy in the future, but we don't provide a 'setter' API for the mask yet
                dsi->ds_pending_data_mask &= ~DISPATCH_MACHPORT_DEAD;
        }

        // no more sources have this flag
        dk->dk_kevent.fflags &= ~DISPATCH_MACHPORT_DEAD;

out:
        // the act of receiving a dead name notification allocates a dead-name right that must be deallocated
        kr = mach_port_deallocate(mach_task_self(), name);
        DISPATCH_VERIFY_MIG(kr);
        //dispatch_assume_zero(kr);

        return KERN_SUCCESS;
}

kern_return_t
_dispatch_wakeup_main_thread(mach_port_t mp __attribute__((unused)))
{
        // dummy function just to pop out the main thread out of mach_msg()
        return 0;
}

kern_return_t
_dispatch_consume_send_once_right(mach_port_t mp __attribute__((unused)))
{
        // dummy function to consume a send-once right
        return 0;
}

mach_msg_return_t
dispatch_mig_server(dispatch_source_t ds, size_t maxmsgsz, dispatch_mig_callback_t callback)
{
        mach_msg_options_t options = MACH_RCV_MSG | MACH_RCV_TIMEOUT
                | MACH_RCV_TRAILER_ELEMENTS(MACH_RCV_TRAILER_CTX)
                | MACH_RCV_TRAILER_TYPE(MACH_MSG_TRAILER_FORMAT_0);
        mach_msg_options_t tmp_options = options;
        mig_reply_error_t *bufTemp, *bufRequest, *bufReply;
        mach_msg_return_t kr = 0;
        unsigned int cnt = 1000;        // do not stall out serial queues
        int demux_success;

        maxmsgsz += MAX_TRAILER_SIZE;

        // XXX FIXME -- allocate these elsewhere
        bufRequest = alloca(maxmsgsz);
        bufReply = alloca(maxmsgsz);
        bufReply->Head.msgh_size = 0;   // make CLANG happy

        // XXX FIXME -- change this to not starve out the target queue
        for (;;) {
                if (DISPATCH_OBJECT_SUSPENDED(ds) || (--cnt == 0)) {
                        options &= ~MACH_RCV_MSG;
                        tmp_options &= ~MACH_RCV_MSG;

                        if (!(tmp_options & MACH_SEND_MSG)) {
                                break;
                        }
                }

                kr = mach_msg(&bufReply->Head, tmp_options, bufReply->Head.msgh_size,
                                (mach_msg_size_t)maxmsgsz, (mach_port_t)ds->ds_ident_hack, 0, 0);

                tmp_options = options;

                if (slowpath(kr)) {
                        switch (kr) {
                        case MACH_SEND_INVALID_DEST:
                        case MACH_SEND_TIMED_OUT:
                                if (bufReply->Head.msgh_bits & MACH_MSGH_BITS_COMPLEX) {
                                        mach_msg_destroy(&bufReply->Head);
                                }
                                break;
                        case MACH_RCV_TIMED_OUT:
                        case MACH_RCV_INVALID_NAME:
                                break;
                        default:
                                dispatch_assume_zero(kr);
                                break;
                        }
                        break;
                }

                if (!(tmp_options & MACH_RCV_MSG)) {
                        break;
                }

                bufTemp = bufRequest;
                bufRequest = bufReply;
                bufReply = bufTemp;

                demux_success = callback(&bufRequest->Head, &bufReply->Head);

                if (!demux_success) {
                        // destroy the request - but not the reply port
                        bufRequest->Head.msgh_remote_port = 0;
                        mach_msg_destroy(&bufRequest->Head);
                } else if (!(bufReply->Head.msgh_bits & MACH_MSGH_BITS_COMPLEX)) {
                        // if MACH_MSGH_BITS_COMPLEX is _not_ set, then bufReply->RetCode is present
                        if (slowpath(bufReply->RetCode)) {
                                if (bufReply->RetCode == MIG_NO_REPLY) {
                                        continue;
                                }

                                // destroy the request - but not the reply port
                                bufRequest->Head.msgh_remote_port = 0;
                                mach_msg_destroy(&bufRequest->Head);
                        }
                }

                if (bufReply->Head.msgh_remote_port) {
                        tmp_options |= MACH_SEND_MSG;
                        if (MACH_MSGH_BITS_REMOTE(bufReply->Head.msgh_bits) != MACH_MSG_TYPE_MOVE_SEND_ONCE) {
                                tmp_options |= MACH_SEND_TIMEOUT;
                        }
                }
        }

        return kr;
}