X-Git-Url: https://git.saurik.com/apple/xnu.git/blobdiff_plain/6d2010ae8f7a6078e10b361c6962983bab233e0f..d9a64523371fa019c4575bb400cbbc3a50ac9903:/osfmk/kern/timer_call.c diff --git a/osfmk/kern/timer_call.c b/osfmk/kern/timer_call.c index 83eb0e43a..86bd1a8df 100644 --- a/osfmk/kern/timer_call.c +++ b/osfmk/kern/timer_call.c @@ -32,15 +32,17 @@ #include #include +#include #include -#include #include #include #include +#include +#include #include -#if CONFIG_DTRACE && (DEVELOPMENT || DEBUG ) +#if CONFIG_DTRACE #include #endif @@ -58,47 +60,209 @@ #define DBG(x...) #endif +#if TIMER_TRACE +#define TIMER_KDEBUG_TRACE KERNEL_DEBUG_CONSTANT_IST +#else +#define TIMER_KDEBUG_TRACE(x...) +#endif + + lck_grp_t timer_call_lck_grp; lck_attr_t timer_call_lck_attr; lck_grp_attr_t timer_call_lck_grp_attr; +lck_grp_t timer_longterm_lck_grp; +lck_attr_t timer_longterm_lck_attr; +lck_grp_attr_t timer_longterm_lck_grp_attr; -#define timer_call_lock_spin(queue) \ +/* Timer queue lock must be acquired with interrupts disabled (under splclock()) */ +#if __SMP__ +#define timer_queue_lock_spin(queue) \ lck_mtx_lock_spin_always(&queue->lock_data) -#define timer_call_unlock(queue) \ +#define timer_queue_unlock(queue) \ lck_mtx_unlock_always(&queue->lock_data) - +#else +#define timer_queue_lock_spin(queue) (void)1 +#define timer_queue_unlock(queue) (void)1 +#endif #define QUEUE(x) ((queue_t)(x)) #define MPQUEUE(x) ((mpqueue_head_t *)(x)) #define TIMER_CALL(x) ((timer_call_t)(x)) +#define TCE(x) (&(x->call_entry)) +/* + * The longterm timer object is a global structure holding all timers + * beyond the short-term, local timer queue threshold. The boot processor + * is responsible for moving each timer to its local timer queue + * if and when that timer becomes due within the threshold. + */ + +/* Sentinel for "no time set": */ +#define TIMER_LONGTERM_NONE EndOfAllTime +/* The default threadhold is the delta above which a timer is "long-term" */ +#if defined(__x86_64__) +#define TIMER_LONGTERM_THRESHOLD (1ULL * NSEC_PER_SEC) /* 1 sec */ +#else +#define TIMER_LONGTERM_THRESHOLD TIMER_LONGTERM_NONE /* disabled */ +#endif -static boolean_t timer_call_enter_internal(timer_call_t call, timer_call_param_t param1, uint64_t deadline, uint32_t flags); +/* + * The scan_limit throttles processing of the longterm queue. + * If the scan time exceeds this limit, we terminate, unlock + * and defer for scan_interval. This prevents unbounded holding of + * timer queue locks with interrupts masked. + */ +#define TIMER_LONGTERM_SCAN_LIMIT (100ULL * NSEC_PER_USEC) /* 100 us */ +#define TIMER_LONGTERM_SCAN_INTERVAL (100ULL * NSEC_PER_USEC) /* 100 us */ +/* Sentinel for "scan limit exceeded": */ +#define TIMER_LONGTERM_SCAN_AGAIN 0 + +typedef struct { + uint64_t interval; /* longterm timer interval */ + uint64_t margin; /* fudge factor (10% of interval */ + uint64_t deadline; /* first/soonest longterm deadline */ + uint64_t preempted; /* sooner timer has pre-empted */ + timer_call_t call; /* first/soonest longterm timer call */ + uint64_t deadline_set; /* next timer set */ + timer_call_data_t timer; /* timer used by threshold management */ + /* Stats: */ + uint64_t scans; /* num threshold timer scans */ + uint64_t preempts; /* num threshold reductions */ + uint64_t latency; /* average threshold latency */ + uint64_t latency_min; /* minimum threshold latency */ + uint64_t latency_max; /* maximum threshold latency */ +} threshold_t; + +typedef struct { + mpqueue_head_t queue; /* longterm timer list */ + uint64_t enqueues; /* num timers queued */ + uint64_t dequeues; /* num timers dequeued */ + uint64_t escalates; /* num timers becoming shortterm */ + uint64_t scan_time; /* last time the list was scanned */ + threshold_t threshold; /* longterm timer threshold */ + uint64_t scan_limit; /* maximum scan time */ + uint64_t scan_interval; /* interval between LT "escalation" scans */ + uint64_t scan_pauses; /* num scans exceeding time limit */ +} timer_longterm_t; + +timer_longterm_t timer_longterm = { + .scan_limit = TIMER_LONGTERM_SCAN_LIMIT, + .scan_interval = TIMER_LONGTERM_SCAN_INTERVAL, + }; + +static mpqueue_head_t *timer_longterm_queue = NULL; + +static void timer_longterm_init(void); +static void timer_longterm_callout( + timer_call_param_t p0, + timer_call_param_t p1); +extern void timer_longterm_scan( + timer_longterm_t *tlp, + uint64_t now); +static void timer_longterm_update( + timer_longterm_t *tlp); +static void timer_longterm_update_locked( + timer_longterm_t *tlp); +static mpqueue_head_t * timer_longterm_enqueue_unlocked( + timer_call_t call, + uint64_t now, + uint64_t deadline, + mpqueue_head_t ** old_queue, + uint64_t soft_deadline, + uint64_t ttd, + timer_call_param_t param1, + uint32_t callout_flags); +static void timer_longterm_dequeued_locked( + timer_call_t call); + +uint64_t past_deadline_timers; +uint64_t past_deadline_deltas; +uint64_t past_deadline_longest; +uint64_t past_deadline_shortest = ~0ULL; +enum {PAST_DEADLINE_TIMER_ADJUSTMENT_NS = 10 * 1000}; + +uint64_t past_deadline_timer_adjustment; + +static boolean_t timer_call_enter_internal(timer_call_t call, timer_call_param_t param1, uint64_t deadline, uint64_t leeway, uint32_t flags, boolean_t ratelimited); boolean_t mach_timer_coalescing_enabled = TRUE; mpqueue_head_t *timer_call_enqueue_deadline_unlocked( timer_call_t call, mpqueue_head_t *queue, - uint64_t deadline); + uint64_t deadline, + uint64_t soft_deadline, + uint64_t ttd, + timer_call_param_t param1, + uint32_t flags); mpqueue_head_t *timer_call_dequeue_unlocked( timer_call_t call); +timer_coalescing_priority_params_t tcoal_prio_params; + +#if TCOAL_PRIO_STATS +int32_t nc_tcl, rt_tcl, bg_tcl, kt_tcl, fp_tcl, ts_tcl, qos_tcl; +#define TCOAL_PRIO_STAT(x) (x++) +#else +#define TCOAL_PRIO_STAT(x) +#endif + +static void +timer_call_init_abstime(void) +{ + int i; + uint64_t result; + timer_coalescing_priority_params_ns_t * tcoal_prio_params_init = timer_call_get_priority_params(); + nanoseconds_to_absolutetime(PAST_DEADLINE_TIMER_ADJUSTMENT_NS, &past_deadline_timer_adjustment); + nanoseconds_to_absolutetime(tcoal_prio_params_init->idle_entry_timer_processing_hdeadline_threshold_ns, &result); + tcoal_prio_params.idle_entry_timer_processing_hdeadline_threshold_abstime = (uint32_t)result; + nanoseconds_to_absolutetime(tcoal_prio_params_init->interrupt_timer_coalescing_ilat_threshold_ns, &result); + tcoal_prio_params.interrupt_timer_coalescing_ilat_threshold_abstime = (uint32_t)result; + nanoseconds_to_absolutetime(tcoal_prio_params_init->timer_resort_threshold_ns, &result); + tcoal_prio_params.timer_resort_threshold_abstime = (uint32_t)result; + tcoal_prio_params.timer_coalesce_rt_shift = tcoal_prio_params_init->timer_coalesce_rt_shift; + tcoal_prio_params.timer_coalesce_bg_shift = tcoal_prio_params_init->timer_coalesce_bg_shift; + tcoal_prio_params.timer_coalesce_kt_shift = tcoal_prio_params_init->timer_coalesce_kt_shift; + tcoal_prio_params.timer_coalesce_fp_shift = tcoal_prio_params_init->timer_coalesce_fp_shift; + tcoal_prio_params.timer_coalesce_ts_shift = tcoal_prio_params_init->timer_coalesce_ts_shift; + + nanoseconds_to_absolutetime(tcoal_prio_params_init->timer_coalesce_rt_ns_max, + &tcoal_prio_params.timer_coalesce_rt_abstime_max); + nanoseconds_to_absolutetime(tcoal_prio_params_init->timer_coalesce_bg_ns_max, + &tcoal_prio_params.timer_coalesce_bg_abstime_max); + nanoseconds_to_absolutetime(tcoal_prio_params_init->timer_coalesce_kt_ns_max, + &tcoal_prio_params.timer_coalesce_kt_abstime_max); + nanoseconds_to_absolutetime(tcoal_prio_params_init->timer_coalesce_fp_ns_max, + &tcoal_prio_params.timer_coalesce_fp_abstime_max); + nanoseconds_to_absolutetime(tcoal_prio_params_init->timer_coalesce_ts_ns_max, + &tcoal_prio_params.timer_coalesce_ts_abstime_max); + + for (i = 0; i < NUM_LATENCY_QOS_TIERS; i++) { + tcoal_prio_params.latency_qos_scale[i] = tcoal_prio_params_init->latency_qos_scale[i]; + nanoseconds_to_absolutetime(tcoal_prio_params_init->latency_qos_ns_max[i], + &tcoal_prio_params.latency_qos_abstime_max[i]); + tcoal_prio_params.latency_tier_rate_limited[i] = tcoal_prio_params_init->latency_tier_rate_limited[i]; + } +} + void -timer_call_initialize(void) +timer_call_init(void) { lck_attr_setdefault(&timer_call_lck_attr); lck_grp_attr_setdefault(&timer_call_lck_grp_attr); lck_grp_init(&timer_call_lck_grp, "timer_call", &timer_call_lck_grp_attr); + + timer_longterm_init(); + timer_call_init_abstime(); } void -timer_call_initialize_queue(mpqueue_head_t *queue) +timer_call_queue_init(mpqueue_head_t *queue) { - DBG("timer_call_initialize_queue(%p)\n", queue); + DBG("timer_call_queue_init(%p)\n", queue); mpqueue_init(queue, &timer_call_lck_grp, &timer_call_lck_attr); } @@ -110,55 +274,16 @@ timer_call_setup( timer_call_param_t param0) { DBG("timer_call_setup(%p,%p,%p)\n", call, func, param0); - call_entry_setup(CE(call), func, param0); + call_entry_setup(TCE(call), func, param0); simple_lock_init(&(call)->lock, 0); call->async_dequeue = FALSE; } - -/* - * Timer call entry locking model - * ============================== - * - * Timer call entries are linked on per-cpu timer queues which are protected - * by the queue lock and the call entry lock. The locking protocol is: - * - * 0) The canonical locking order is timer call entry followed by queue. - * - * 1) With only the entry lock held, entry.queue is valid: - * 1a) NULL: the entry is not queued, or - * 1b) non-NULL: this queue must be locked before the entry is modified. - * After locking the queue, the call.async_dequeue flag must be checked: - * 1c) TRUE: the entry was removed from the queue by another thread - * and we must NULL the entry.queue and reset this flag, or - * 1d) FALSE: (ie. queued), the entry can be manipulated. - * - * 2) If a queue lock is obtained first, the queue is stable: - * 2a) If a try-lock of a queued entry succeeds, the call can be operated on - * and dequeued. - * 2b) If a try-lock fails, it indicates that another thread is attempting - * to change the entry and move it to a different position in this queue - * or to different queue. The entry can be dequeued but it should not be - * operated upon since it is being changed. Furthermore, we don't null - * the entry.queue pointer (protected by the entry lock we don't own). - * Instead, we set the async_dequeue flag -- see (1c). - */ - -/* - * Inlines timer_call_entry_dequeue() and timer_call_entry_enqueue_deadline() - * cast between pointer types (mpqueue_head_t *) and (queue_t) so that - * we can use the call_entry_dequeue() and call_entry_enqueue_deadline() - * methods to operate on timer_call structs as if they are call_entry structs. - * These structures are identical except for their queue head pointer fields. - * - * In the debug case, we assert that the timer call locking protocol - * is being obeyed. - */ #if TIMER_ASSERT static __inline__ mpqueue_head_t * timer_call_entry_dequeue( timer_call_t entry) { - mpqueue_head_t *old_queue = MPQUEUE(CE(entry)->queue); + mpqueue_head_t *old_queue = MPQUEUE(TCE(entry)->queue); if (!hw_lock_held((hw_lock_t)&entry->lock)) panic("_call_entry_dequeue() " @@ -172,7 +297,8 @@ timer_call_entry_dequeue( panic("_call_entry_dequeue() " "queue %p is not locked\n", old_queue); - call_entry_dequeue(CE(entry)); + call_entry_dequeue(TCE(entry)); + old_queue->count--; return (old_queue); } @@ -183,7 +309,7 @@ timer_call_entry_enqueue_deadline( mpqueue_head_t *queue, uint64_t deadline) { - mpqueue_head_t *old_queue = MPQUEUE(CE(entry)->queue); + mpqueue_head_t *old_queue = MPQUEUE(TCE(entry)->queue); if (!hw_lock_held((hw_lock_t)&entry->lock)) panic("_call_entry_enqueue_deadline() " @@ -196,7 +322,18 @@ timer_call_entry_enqueue_deadline( panic("_call_entry_enqueue_deadline() " "old_queue %p != queue", old_queue); - call_entry_enqueue_deadline(CE(entry), QUEUE(queue), deadline); + call_entry_enqueue_deadline(TCE(entry), QUEUE(queue), deadline); + +/* For efficiency, track the earliest soft deadline on the queue, so that + * fuzzy decisions can be made without lock acquisitions. + */ + timer_call_t thead = (timer_call_t)queue_first(&queue->head); + + queue->earliest_soft_deadline = thead->flags & TIMER_CALL_RATELIMITED ? TCE(thead)->deadline : thead->soft_deadline; + + if (old_queue) + old_queue->count--; + queue->count++; return (old_queue); } @@ -207,7 +344,12 @@ static __inline__ mpqueue_head_t * timer_call_entry_dequeue( timer_call_t entry) { - return MPQUEUE(call_entry_dequeue(CE(entry))); + mpqueue_head_t *old_queue = MPQUEUE(TCE(entry)->queue); + + call_entry_dequeue(TCE(entry)); + old_queue->count--; + + return old_queue; } static __inline__ mpqueue_head_t * @@ -216,12 +358,53 @@ timer_call_entry_enqueue_deadline( mpqueue_head_t *queue, uint64_t deadline) { - return MPQUEUE(call_entry_enqueue_deadline(CE(entry), - QUEUE(queue), deadline)); + mpqueue_head_t *old_queue = MPQUEUE(TCE(entry)->queue); + + call_entry_enqueue_deadline(TCE(entry), QUEUE(queue), deadline); + + /* For efficiency, track the earliest soft deadline on the queue, + * so that fuzzy decisions can be made without lock acquisitions. + */ + + timer_call_t thead = (timer_call_t)queue_first(&queue->head); + queue->earliest_soft_deadline = thead->flags & TIMER_CALL_RATELIMITED ? TCE(thead)->deadline : thead->soft_deadline; + + if (old_queue) + old_queue->count--; + queue->count++; + + return old_queue; } #endif +static __inline__ void +timer_call_entry_enqueue_tail( + timer_call_t entry, + mpqueue_head_t *queue) +{ + call_entry_enqueue_tail(TCE(entry), QUEUE(queue)); + queue->count++; + return; +} + +/* + * Remove timer entry from its queue but don't change the queue pointer + * and set the async_dequeue flag. This is locking case 2b. + */ +static __inline__ void +timer_call_entry_dequeue_async( + timer_call_t entry) +{ + mpqueue_head_t *old_queue = MPQUEUE(TCE(entry)->queue); + if (old_queue) { + old_queue->count--; + (void) remque(qe(entry)); + entry->async_dequeue = TRUE; + } + return; +} + #if TIMER_ASSERT unsigned timer_call_enqueue_deadline_unlocked_async1; unsigned timer_call_enqueue_deadline_unlocked_async2; @@ -233,41 +416,59 @@ __inline__ mpqueue_head_t * timer_call_enqueue_deadline_unlocked( timer_call_t call, mpqueue_head_t *queue, - uint64_t deadline) + uint64_t deadline, + uint64_t soft_deadline, + uint64_t ttd, + timer_call_param_t param1, + uint32_t callout_flags) { - call_entry_t entry = CE(call); + call_entry_t entry = TCE(call); mpqueue_head_t *old_queue; DBG("timer_call_enqueue_deadline_unlocked(%p,%p,)\n", call, queue); simple_lock(&call->lock); + old_queue = MPQUEUE(entry->queue); + if (old_queue != NULL) { - timer_call_lock_spin(old_queue); + timer_queue_lock_spin(old_queue); if (call->async_dequeue) { - /* collision (1c): null queue pointer and reset flag */ - call->async_dequeue = FALSE; - entry->queue = NULL; + /* collision (1c): timer already dequeued, clear flag */ #if TIMER_ASSERT + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_ASYNC_DEQ | DBG_FUNC_NONE, + VM_KERNEL_UNSLIDE_OR_PERM(call), + call->async_dequeue, + VM_KERNEL_UNSLIDE_OR_PERM(TCE(call)->queue), + 0x1c, 0); timer_call_enqueue_deadline_unlocked_async1++; #endif - } else if (old_queue != queue) { - (void)remque(qe(entry)); + call->async_dequeue = FALSE; entry->queue = NULL; + } else if (old_queue != queue) { + timer_call_entry_dequeue(call); #if TIMER_ASSERT timer_call_enqueue_deadline_unlocked_async2++; #endif } + if (old_queue == timer_longterm_queue) + timer_longterm_dequeued_locked(call); if (old_queue != queue) { - timer_call_unlock(old_queue); - timer_call_lock_spin(queue); + timer_queue_unlock(old_queue); + timer_queue_lock_spin(queue); } } else { - timer_call_lock_spin(queue); + timer_queue_lock_spin(queue); } + call->soft_deadline = soft_deadline; + call->flags = callout_flags; + TCE(call)->param1 = param1; + call->ttd = ttd; + timer_call_entry_enqueue_deadline(call, queue, deadline); - timer_call_unlock(queue); + timer_queue_unlock(queue); simple_unlock(&call->lock); return (old_queue); @@ -281,75 +482,204 @@ mpqueue_head_t * timer_call_dequeue_unlocked( timer_call_t call) { - call_entry_t entry = CE(call); + call_entry_t entry = TCE(call); mpqueue_head_t *old_queue; DBG("timer_call_dequeue_unlocked(%p)\n", call); simple_lock(&call->lock); old_queue = MPQUEUE(entry->queue); +#if TIMER_ASSERT + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_ASYNC_DEQ | DBG_FUNC_NONE, + VM_KERNEL_UNSLIDE_OR_PERM(call), + call->async_dequeue, + VM_KERNEL_UNSLIDE_OR_PERM(TCE(call)->queue), + 0, 0); +#endif if (old_queue != NULL) { - timer_call_lock_spin(old_queue); + timer_queue_lock_spin(old_queue); if (call->async_dequeue) { - /* collision (1c): null queue pointer and reset flag */ - call->async_dequeue = FALSE; + /* collision (1c): timer already dequeued, clear flag */ #if TIMER_ASSERT + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_ASYNC_DEQ | DBG_FUNC_NONE, + VM_KERNEL_UNSLIDE_OR_PERM(call), + call->async_dequeue, + VM_KERNEL_UNSLIDE_OR_PERM(TCE(call)->queue), + 0x1c, 0); timer_call_dequeue_unlocked_async1++; #endif + call->async_dequeue = FALSE; + entry->queue = NULL; } else { - (void)remque(qe(entry)); -#if TIMER_ASSERT - timer_call_dequeue_unlocked_async2++; -#endif + timer_call_entry_dequeue(call); } - entry->queue = NULL; - timer_call_unlock(old_queue); + if (old_queue == timer_longterm_queue) + timer_longterm_dequeued_locked(call); + timer_queue_unlock(old_queue); } simple_unlock(&call->lock); return (old_queue); } +static uint64_t +past_deadline_timer_handle(uint64_t deadline, uint64_t ctime) +{ + uint64_t delta = (ctime - deadline); + + past_deadline_timers++; + past_deadline_deltas += delta; + if (delta > past_deadline_longest) + past_deadline_longest = deadline; + if (delta < past_deadline_shortest) + past_deadline_shortest = delta; + + return (ctime + past_deadline_timer_adjustment); +} + +/* + * Timer call entry locking model + * ============================== + * + * Timer call entries are linked on per-cpu timer queues which are protected + * by the queue lock and the call entry lock. The locking protocol is: + * + * 0) The canonical locking order is timer call entry followed by queue. + * + * 1) With only the entry lock held, entry.queue is valid: + * 1a) NULL: the entry is not queued, or + * 1b) non-NULL: this queue must be locked before the entry is modified. + * After locking the queue, the call.async_dequeue flag must be checked: + * 1c) TRUE: the entry was removed from the queue by another thread + * and we must NULL the entry.queue and reset this flag, or + * 1d) FALSE: (ie. queued), the entry can be manipulated. + * + * 2) If a queue lock is obtained first, the queue is stable: + * 2a) If a try-lock of a queued entry succeeds, the call can be operated on + * and dequeued. + * 2b) If a try-lock fails, it indicates that another thread is attempting + * to change the entry and move it to a different position in this queue + * or to different queue. The entry can be dequeued but it should not be + * operated upon since it is being changed. Furthermore, we don't null + * the entry.queue pointer (protected by the entry lock we don't own). + * Instead, we set the async_dequeue flag -- see (1c). + * 2c) Same as 2b but occurring when a longterm timer is matured. + * 3) A callout's parameters (deadline, flags, parameters, soft deadline &c.) + * should be manipulated with the appropriate timer queue lock held, + * to prevent queue traversal observations from observing inconsistent + * updates to an in-flight callout. + */ + +/* + * Inlines timer_call_entry_dequeue() and timer_call_entry_enqueue_deadline() + * cast between pointer types (mpqueue_head_t *) and (queue_t) so that + * we can use the call_entry_dequeue() and call_entry_enqueue_deadline() + * methods to operate on timer_call structs as if they are call_entry structs. + * These structures are identical except for their queue head pointer fields. + * + * In the debug case, we assert that the timer call locking protocol + * is being obeyed. + */ + static boolean_t timer_call_enter_internal( timer_call_t call, timer_call_param_t param1, uint64_t deadline, - uint32_t flags) + uint64_t leeway, + uint32_t flags, + boolean_t ratelimited) { - mpqueue_head_t *queue; + mpqueue_head_t *queue = NULL; mpqueue_head_t *old_queue; spl_t s; - uint64_t slop = 0; + uint64_t slop; + uint32_t urgency; + uint64_t sdeadline, ttd; + assert(call->call_entry.func != NULL); s = splclock(); - call->soft_deadline = deadline; - call->flags = flags; + sdeadline = deadline; + uint64_t ctime = mach_absolute_time(); + + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_ENTER | DBG_FUNC_START, + VM_KERNEL_UNSLIDE_OR_PERM(call), + VM_KERNEL_ADDRHIDE(param1), deadline, flags, 0); + + urgency = (flags & TIMER_CALL_URGENCY_MASK); + + boolean_t slop_ratelimited = FALSE; + slop = timer_call_slop(deadline, ctime, urgency, current_thread(), &slop_ratelimited); - if ((flags & TIMER_CALL_CRITICAL) == 0 && - mach_timer_coalescing_enabled) { - slop = timer_call_slop(deadline); + if ((flags & TIMER_CALL_LEEWAY) != 0 && leeway > slop) + slop = leeway; + + if (UINT64_MAX - deadline <= slop) { + deadline = UINT64_MAX; + } else { deadline += slop; } - queue = timer_queue_assign(deadline); + if (__improbable(deadline < ctime)) { + deadline = past_deadline_timer_handle(deadline, ctime); + sdeadline = deadline; + } - old_queue = timer_call_enqueue_deadline_unlocked(call, queue, deadline); + if (ratelimited || slop_ratelimited) { + flags |= TIMER_CALL_RATELIMITED; + } else { + flags &= ~TIMER_CALL_RATELIMITED; + } - CE(call)->param1 = param1; + ttd = sdeadline - ctime; +#if CONFIG_DTRACE + DTRACE_TMR7(callout__create, timer_call_func_t, TCE(call)->func, + timer_call_param_t, TCE(call)->param0, uint32_t, flags, + (deadline - sdeadline), + (ttd >> 32), (unsigned) (ttd & 0xFFFFFFFF), call); +#endif + + /* Program timer callout parameters under the appropriate per-CPU or + * longterm queue lock. The callout may have been previously enqueued + * and in-flight on this or another timer queue. + */ + if (!ratelimited && !slop_ratelimited) { + queue = timer_longterm_enqueue_unlocked(call, ctime, deadline, &old_queue, sdeadline, ttd, param1, flags); + } + + if (queue == NULL) { + queue = timer_queue_assign(deadline); + old_queue = timer_call_enqueue_deadline_unlocked(call, queue, deadline, sdeadline, ttd, param1, flags); + } + +#if TIMER_TRACE + TCE(call)->entry_time = ctime; +#endif + + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_ENTER | DBG_FUNC_END, + VM_KERNEL_UNSLIDE_OR_PERM(call), + (old_queue != NULL), deadline, queue->count, 0); splx(s); return (old_queue != NULL); } +/* + * timer_call_*() + * return boolean indicating whether the call was previously queued. + */ boolean_t timer_call_enter( timer_call_t call, uint64_t deadline, uint32_t flags) { - return timer_call_enter_internal(call, NULL, deadline, flags); + return timer_call_enter_internal(call, NULL, deadline, 0, flags, FALSE); } boolean_t @@ -359,7 +689,94 @@ timer_call_enter1( uint64_t deadline, uint32_t flags) { - return timer_call_enter_internal(call, param1, deadline, flags); + return timer_call_enter_internal(call, param1, deadline, 0, flags, FALSE); +} + +boolean_t +timer_call_enter_with_leeway( + timer_call_t call, + timer_call_param_t param1, + uint64_t deadline, + uint64_t leeway, + uint32_t flags, + boolean_t ratelimited) +{ + return timer_call_enter_internal(call, param1, deadline, leeway, flags, ratelimited); +} + +boolean_t +timer_call_quantum_timer_enter( + timer_call_t call, + timer_call_param_t param1, + uint64_t deadline, + uint64_t ctime) +{ + assert(call->call_entry.func != NULL); + assert(ml_get_interrupts_enabled() == FALSE); + + uint32_t flags = TIMER_CALL_SYS_CRITICAL | TIMER_CALL_LOCAL; + + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, DECR_TIMER_ENTER | DBG_FUNC_START, + VM_KERNEL_UNSLIDE_OR_PERM(call), + VM_KERNEL_ADDRHIDE(param1), deadline, + flags, 0); + + if (__improbable(deadline < ctime)) { + deadline = past_deadline_timer_handle(deadline, ctime); + } + + uint64_t ttd = deadline - ctime; +#if CONFIG_DTRACE + DTRACE_TMR7(callout__create, timer_call_func_t, TCE(call)->func, + timer_call_param_t, TCE(call)->param0, uint32_t, flags, 0, + (ttd >> 32), (unsigned) (ttd & 0xFFFFFFFF), call); +#endif + + quantum_timer_set_deadline(deadline); + TCE(call)->deadline = deadline; + TCE(call)->param1 = param1; + call->ttd = ttd; + call->flags = flags; + +#if TIMER_TRACE + TCE(call)->entry_time = ctime; +#endif + + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, DECR_TIMER_ENTER | DBG_FUNC_END, + VM_KERNEL_UNSLIDE_OR_PERM(call), + 1, deadline, 0, 0); + + return true; +} + + +boolean_t +timer_call_quantum_timer_cancel( + timer_call_t call) +{ + assert(ml_get_interrupts_enabled() == FALSE); + + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_CANCEL | DBG_FUNC_START, + VM_KERNEL_UNSLIDE_OR_PERM(call), TCE(call)->deadline, + 0, call->flags, 0); + + TCE(call)->deadline = 0; + quantum_timer_set_deadline(0); + + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_CANCEL | DBG_FUNC_END, + VM_KERNEL_UNSLIDE_OR_PERM(call), 0, + TCE(call)->deadline - mach_absolute_time(), + TCE(call)->deadline - TCE(call)->entry_time, 0); + +#if CONFIG_DTRACE + DTRACE_TMR6(callout__cancel, timer_call_func_t, TCE(call)->func, + timer_call_param_t, TCE(call)->param0, uint32_t, call->flags, 0, + (call->ttd >> 32), (unsigned) (call->ttd & 0xFFFFFFFF)); +#endif + + return true; } boolean_t @@ -371,22 +788,46 @@ timer_call_cancel( s = splclock(); + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_CANCEL | DBG_FUNC_START, + VM_KERNEL_UNSLIDE_OR_PERM(call), + TCE(call)->deadline, call->soft_deadline, call->flags, 0); + old_queue = timer_call_dequeue_unlocked(call); if (old_queue != NULL) { - timer_call_lock_spin(old_queue); - if (!queue_empty(&old_queue->head)) - timer_queue_cancel(old_queue, CE(call)->deadline, CE(queue_first(&old_queue->head))->deadline); - else - timer_queue_cancel(old_queue, CE(call)->deadline, UINT64_MAX); - timer_call_unlock(old_queue); + timer_queue_lock_spin(old_queue); + if (!queue_empty(&old_queue->head)) { + timer_queue_cancel(old_queue, TCE(call)->deadline, CE(queue_first(&old_queue->head))->deadline); + timer_call_t thead = (timer_call_t)queue_first(&old_queue->head); + old_queue->earliest_soft_deadline = thead->flags & TIMER_CALL_RATELIMITED ? TCE(thead)->deadline : thead->soft_deadline; + } + else { + timer_queue_cancel(old_queue, TCE(call)->deadline, UINT64_MAX); + old_queue->earliest_soft_deadline = UINT64_MAX; + } + timer_queue_unlock(old_queue); } + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_CANCEL | DBG_FUNC_END, + VM_KERNEL_UNSLIDE_OR_PERM(call), + VM_KERNEL_UNSLIDE_OR_PERM(old_queue), + TCE(call)->deadline - mach_absolute_time(), + TCE(call)->deadline - TCE(call)->entry_time, 0); splx(s); +#if CONFIG_DTRACE + DTRACE_TMR6(callout__cancel, timer_call_func_t, TCE(call)->func, + timer_call_param_t, TCE(call)->param0, uint32_t, call->flags, 0, + (call->ttd >> 32), (unsigned) (call->ttd & 0xFFFFFFFF)); +#endif + return (old_queue != NULL); } -uint32_t timer_queue_shutdown_lock_skips; +static uint32_t timer_queue_shutdown_lock_skips; +static uint32_t timer_queue_shutdown_discarded; + void timer_queue_shutdown( mpqueue_head_t *queue) @@ -395,13 +836,15 @@ timer_queue_shutdown( mpqueue_head_t *new_queue; spl_t s; + DBG("timer_queue_shutdown(%p)\n", queue); s = splclock(); /* Note comma operator in while expression re-locking each iteration */ - while (timer_call_lock_spin(queue), !queue_empty(&queue->head)) { + while ((void)timer_queue_lock_spin(queue), !queue_empty(&queue->head)) { call = TIMER_CALL(queue_first(&queue->head)); + if (!simple_lock_try(&call->lock)) { /* * case (2b) lock order inversion, dequeue and skip @@ -409,111 +852,240 @@ timer_queue_shutdown( * but set the async_dequeue field. */ timer_queue_shutdown_lock_skips++; - (void) remque(qe(call)); - call->async_dequeue = TRUE; - timer_call_unlock(queue); + timer_call_entry_dequeue_async(call); +#if TIMER_ASSERT + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_ASYNC_DEQ | DBG_FUNC_NONE, + VM_KERNEL_UNSLIDE_OR_PERM(call), + call->async_dequeue, + VM_KERNEL_UNSLIDE_OR_PERM(TCE(call)->queue), + 0x2b, 0); +#endif + timer_queue_unlock(queue); continue; } + boolean_t call_local = ((call->flags & TIMER_CALL_LOCAL) != 0); + /* remove entry from old queue */ timer_call_entry_dequeue(call); - timer_call_unlock(queue); - - /* and queue it on new */ - new_queue = timer_queue_assign(CE(call)->deadline); - timer_call_lock_spin(new_queue); - timer_call_entry_enqueue_deadline( - call, new_queue, CE(call)->deadline); - timer_call_unlock(new_queue); + timer_queue_unlock(queue); + + if (call_local == FALSE) { + /* and queue it on new, discarding LOCAL timers */ + new_queue = timer_queue_assign(TCE(call)->deadline); + timer_queue_lock_spin(new_queue); + timer_call_entry_enqueue_deadline( + call, new_queue, TCE(call)->deadline); + timer_queue_unlock(new_queue); + } else { + timer_queue_shutdown_discarded++; + } + assert(call_local == FALSE); simple_unlock(&call->lock); } - timer_call_unlock(queue); + timer_queue_unlock(queue); splx(s); } -uint32_t timer_queue_expire_lock_skips; -uint64_t -timer_queue_expire( - mpqueue_head_t *queue, + +void +quantum_timer_expire( uint64_t deadline) { - timer_call_t call; + processor_t processor = current_processor(); + timer_call_t call = TIMER_CALL(&(processor->quantum_timer)); + + if (__improbable(TCE(call)->deadline > deadline)) + panic("CPU quantum timer deadlin out of sync with timer call deadline"); + + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_EXPIRE | DBG_FUNC_NONE, + VM_KERNEL_UNSLIDE_OR_PERM(call), + TCE(call)->deadline, + TCE(call)->deadline, + TCE(call)->entry_time, 0); + + timer_call_func_t func = TCE(call)->func; + timer_call_param_t param0 = TCE(call)->param0; + timer_call_param_t param1 = TCE(call)->param1; + + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_CALLOUT | DBG_FUNC_START, + VM_KERNEL_UNSLIDE_OR_PERM(call), VM_KERNEL_UNSLIDE(func), + VM_KERNEL_ADDRHIDE(param0), + VM_KERNEL_ADDRHIDE(param1), + 0); + +#if CONFIG_DTRACE + DTRACE_TMR7(callout__start, timer_call_func_t, func, + timer_call_param_t, param0, unsigned, call->flags, + 0, (call->ttd >> 32), + (unsigned) (call->ttd & 0xFFFFFFFF), call); +#endif + (*func)(param0, param1); + + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_CALLOUT | DBG_FUNC_END, + VM_KERNEL_UNSLIDE_OR_PERM(call), VM_KERNEL_UNSLIDE(func), + VM_KERNEL_ADDRHIDE(param0), + VM_KERNEL_ADDRHIDE(param1), + 0); +} +static uint32_t timer_queue_expire_lock_skips; +uint64_t +timer_queue_expire_with_options( + mpqueue_head_t *queue, + uint64_t deadline, + boolean_t rescan) +{ + timer_call_t call = NULL; + uint32_t tc_iterations = 0; DBG("timer_queue_expire(%p,)\n", queue); - timer_call_lock_spin(queue); + uint64_t cur_deadline = deadline; + timer_queue_lock_spin(queue); while (!queue_empty(&queue->head)) { - call = TIMER_CALL(queue_first(&queue->head)); + /* Upon processing one or more timer calls, refresh the + * deadline to account for time elapsed in the callout + */ + if (++tc_iterations > 1) + cur_deadline = mach_absolute_time(); + + if (call == NULL) + call = TIMER_CALL(queue_first(&queue->head)); - if (call->soft_deadline <= deadline) { + if (call->soft_deadline <= cur_deadline) { timer_call_func_t func; timer_call_param_t param0, param1; + TCOAL_DEBUG(0xDDDD0000, queue->earliest_soft_deadline, call->soft_deadline, 0, 0, 0); + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_EXPIRE | DBG_FUNC_NONE, + VM_KERNEL_UNSLIDE_OR_PERM(call), + call->soft_deadline, + TCE(call)->deadline, + TCE(call)->entry_time, 0); + + if ((call->flags & TIMER_CALL_RATELIMITED) && + (TCE(call)->deadline > cur_deadline)) { + if (rescan == FALSE) + break; + } + if (!simple_lock_try(&call->lock)) { /* case (2b) lock inversion, dequeue and skip */ timer_queue_expire_lock_skips++; - (void) remque(qe(call)); - call->async_dequeue = TRUE; + timer_call_entry_dequeue_async(call); + call = NULL; continue; } timer_call_entry_dequeue(call); - func = CE(call)->func; - param0 = CE(call)->param0; - param1 = CE(call)->param1; + func = TCE(call)->func; + param0 = TCE(call)->param0; + param1 = TCE(call)->param1; simple_unlock(&call->lock); - timer_call_unlock(queue); - - KERNEL_DEBUG_CONSTANT(DECR_TIMER_CALLOUT | DBG_FUNC_START, - func, - param0, - param1, 0, 0); - -#if CONFIG_DTRACE && (DEVELOPMENT || DEBUG ) - DTRACE_TMR3(callout__start, timer_call_func_t, func, - timer_call_param_t, param0, - timer_call_param_t, param1); + timer_queue_unlock(queue); + + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_CALLOUT | DBG_FUNC_START, + VM_KERNEL_UNSLIDE_OR_PERM(call), VM_KERNEL_UNSLIDE(func), + VM_KERNEL_ADDRHIDE(param0), + VM_KERNEL_ADDRHIDE(param1), + 0); + +#if CONFIG_DTRACE + DTRACE_TMR7(callout__start, timer_call_func_t, func, + timer_call_param_t, param0, unsigned, call->flags, + 0, (call->ttd >> 32), + (unsigned) (call->ttd & 0xFFFFFFFF), call); #endif - + /* Maintain time-to-deadline in per-processor data + * structure for thread wakeup deadline statistics. + */ + uint64_t *ttdp = &(PROCESSOR_DATA(current_processor(), timer_call_ttd)); + *ttdp = call->ttd; (*func)(param0, param1); - -#if CONFIG_DTRACE && (DEVELOPMENT || DEBUG ) - DTRACE_TMR3(callout__end, timer_call_func_t, func, - timer_call_param_t, param0, - timer_call_param_t, param1); + *ttdp = 0; +#if CONFIG_DTRACE + DTRACE_TMR4(callout__end, timer_call_func_t, func, + param0, param1, call); #endif - KERNEL_DEBUG_CONSTANT(DECR_TIMER_CALLOUT | DBG_FUNC_END, - func, - param0, - param1, 0, 0); - - timer_call_lock_spin(queue); + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_CALLOUT | DBG_FUNC_END, + VM_KERNEL_UNSLIDE_OR_PERM(call), VM_KERNEL_UNSLIDE(func), + VM_KERNEL_ADDRHIDE(param0), + VM_KERNEL_ADDRHIDE(param1), + 0); + call = NULL; + timer_queue_lock_spin(queue); + } else { + if (__probable(rescan == FALSE)) { + break; + } else { + int64_t skew = TCE(call)->deadline - call->soft_deadline; + assert(TCE(call)->deadline >= call->soft_deadline); + + /* DRK: On a latency quality-of-service level change, + * re-sort potentially rate-limited timers. The platform + * layer determines which timers require + * this. In the absence of the per-callout + * synchronization requirement, a global resort could + * be more efficient. The re-sort effectively + * annuls all timer adjustments, i.e. the "soft + * deadline" is the sort key. + */ + + if (timer_resort_threshold(skew)) { + if (__probable(simple_lock_try(&call->lock))) { + timer_call_entry_dequeue(call); + timer_call_entry_enqueue_deadline(call, queue, call->soft_deadline); + simple_unlock(&call->lock); + call = NULL; + } + } + if (call) { + call = TIMER_CALL(queue_next(qe(call))); + if (queue_end(&queue->head, qe(call))) + break; + } + } } - else - break; } - if (!queue_empty(&queue->head)) - deadline = CE(call)->deadline; - else - deadline = UINT64_MAX; + if (!queue_empty(&queue->head)) { + call = TIMER_CALL(queue_first(&queue->head)); + cur_deadline = TCE(call)->deadline; + queue->earliest_soft_deadline = (call->flags & TIMER_CALL_RATELIMITED) ? TCE(call)->deadline: call->soft_deadline; + } else { + queue->earliest_soft_deadline = cur_deadline = UINT64_MAX; + } - timer_call_unlock(queue); + timer_queue_unlock(queue); - return (deadline); + return (cur_deadline); } +uint64_t +timer_queue_expire( + mpqueue_head_t *queue, + uint64_t deadline) +{ + return timer_queue_expire_with_options(queue, deadline, FALSE); +} extern int serverperfmode; -uint32_t timer_queue_migrate_lock_skips; +static uint32_t timer_queue_migrate_lock_skips; /* - * timer_queue_migrate() is called by etimer_queue_migrate() + * timer_queue_migrate() is called by timer_queue_migrate_cpu() * to move timer requests from the local processor (queue_from) * to a target processor's (queue_to). */ @@ -551,7 +1123,7 @@ timer_queue_migrate(mpqueue_head_t *queue_from, mpqueue_head_t *queue_to) * so that we need not have the target resync. */ - timer_call_lock_spin(queue_to); + timer_queue_lock_spin(queue_to); head_to = TIMER_CALL(queue_first(&queue_to->head)); if (queue_empty(&queue_to->head)) { @@ -559,7 +1131,7 @@ timer_queue_migrate(mpqueue_head_t *queue_from, mpqueue_head_t *queue_to) goto abort1; } - timer_call_lock_spin(queue_from); + timer_queue_lock_spin(queue_from); if (queue_empty(&queue_from->head)) { timers_migrated = -2; @@ -567,7 +1139,7 @@ timer_queue_migrate(mpqueue_head_t *queue_from, mpqueue_head_t *queue_to) } call = TIMER_CALL(queue_first(&queue_from->head)); - if (CE(call)->deadline < CE(head_to)->deadline) { + if (TCE(call)->deadline < TCE(head_to)->deadline) { timers_migrated = 0; goto abort2; } @@ -586,22 +1158,732 @@ timer_queue_migrate(mpqueue_head_t *queue_from, mpqueue_head_t *queue_to) call = TIMER_CALL(queue_first(&queue_from->head)); if (!simple_lock_try(&call->lock)) { /* case (2b) lock order inversion, dequeue only */ +#ifdef TIMER_ASSERT + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_ASYNC_DEQ | DBG_FUNC_NONE, + VM_KERNEL_UNSLIDE_OR_PERM(call), + VM_KERNEL_UNSLIDE_OR_PERM(TCE(call)->queue), + VM_KERNEL_UNSLIDE_OR_PERM(call->lock.interlock.lock_data), + 0x2b, 0); +#endif timer_queue_migrate_lock_skips++; - (void) remque(qe(call)); - call->async_dequeue = TRUE; + timer_call_entry_dequeue_async(call); continue; } timer_call_entry_dequeue(call); timer_call_entry_enqueue_deadline( - call, queue_to, CE(call)->deadline); + call, queue_to, TCE(call)->deadline); timers_migrated++; simple_unlock(&call->lock); } - + queue_from->earliest_soft_deadline = UINT64_MAX; abort2: - timer_call_unlock(queue_from); + timer_queue_unlock(queue_from); abort1: - timer_call_unlock(queue_to); + timer_queue_unlock(queue_to); return timers_migrated; } + +void +timer_queue_trace_cpu(int ncpu) +{ + timer_call_nosync_cpu( + ncpu, + (void(*)(void *))timer_queue_trace, + (void*) timer_queue_cpu(ncpu)); +} + +void +timer_queue_trace( + mpqueue_head_t *queue) +{ + timer_call_t call; + spl_t s; + + if (!kdebug_enable) + return; + + s = splclock(); + timer_queue_lock_spin(queue); + + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_QUEUE | DBG_FUNC_START, + queue->count, mach_absolute_time(), 0, 0, 0); + + if (!queue_empty(&queue->head)) { + call = TIMER_CALL(queue_first(&queue->head)); + do { + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_QUEUE | DBG_FUNC_NONE, + call->soft_deadline, + TCE(call)->deadline, + TCE(call)->entry_time, + VM_KERNEL_UNSLIDE(TCE(call)->func), + 0); + call = TIMER_CALL(queue_next(qe(call))); + } while (!queue_end(&queue->head, qe(call))); + } + + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_QUEUE | DBG_FUNC_END, + queue->count, mach_absolute_time(), 0, 0, 0); + + timer_queue_unlock(queue); + splx(s); +} + +void +timer_longterm_dequeued_locked(timer_call_t call) +{ + timer_longterm_t *tlp = &timer_longterm; + + tlp->dequeues++; + if (call == tlp->threshold.call) + tlp->threshold.call = NULL; +} + +/* + * Place a timer call in the longterm list + * and adjust the next timer callout deadline if the new timer is first. + */ +mpqueue_head_t * +timer_longterm_enqueue_unlocked(timer_call_t call, + uint64_t now, + uint64_t deadline, + mpqueue_head_t **old_queue, + uint64_t soft_deadline, + uint64_t ttd, + timer_call_param_t param1, + uint32_t callout_flags) +{ + timer_longterm_t *tlp = &timer_longterm; + boolean_t update_required = FALSE; + uint64_t longterm_threshold; + + longterm_threshold = now + tlp->threshold.interval; + + /* + * Return NULL without doing anything if: + * - this timer is local, or + * - the longterm mechanism is disabled, or + * - this deadline is too short. + */ + if ((callout_flags & TIMER_CALL_LOCAL) != 0 || + (tlp->threshold.interval == TIMER_LONGTERM_NONE) || + (deadline <= longterm_threshold)) + return NULL; + + /* + * Remove timer from its current queue, if any. + */ + *old_queue = timer_call_dequeue_unlocked(call); + + /* + * Lock the longterm queue, queue timer and determine + * whether an update is necessary. + */ + assert(!ml_get_interrupts_enabled()); + simple_lock(&call->lock); + timer_queue_lock_spin(timer_longterm_queue); + TCE(call)->deadline = deadline; + TCE(call)->param1 = param1; + call->ttd = ttd; + call->soft_deadline = soft_deadline; + call->flags = callout_flags; + timer_call_entry_enqueue_tail(call, timer_longterm_queue); + + tlp->enqueues++; + + /* + * We'll need to update the currently set threshold timer + * if the new deadline is sooner and no sooner update is in flight. + */ + if (deadline < tlp->threshold.deadline && + deadline < tlp->threshold.preempted) { + tlp->threshold.preempted = deadline; + tlp->threshold.call = call; + update_required = TRUE; + } + timer_queue_unlock(timer_longterm_queue); + simple_unlock(&call->lock); + + if (update_required) { + /* + * Note: this call expects that calling the master cpu + * alone does not involve locking the topo lock. + */ + timer_call_nosync_cpu( + master_cpu, + (void (*)(void *)) timer_longterm_update, + (void *)tlp); + } + + return timer_longterm_queue; +} + +/* + * Scan for timers below the longterm threshold. + * Move these to the local timer queue (of the boot processor on which the + * calling thread is running). + * Both the local (boot) queue and the longterm queue are locked. + * The scan is similar to the timer migrate sequence but is performed by + * successively examining each timer on the longterm queue: + * - if within the short-term threshold + * - enter on the local queue (unless being deleted), + * - otherwise: + * - if sooner, deadline becomes the next threshold deadline. + * The total scan time is limited to TIMER_LONGTERM_SCAN_LIMIT. Should this be + * exceeded, we abort and reschedule again so that we don't shut others from + * the timer queues. Longterm timers firing late is not critical. + */ +void +timer_longterm_scan(timer_longterm_t *tlp, + uint64_t time_start) +{ + queue_entry_t qe; + timer_call_t call; + uint64_t threshold; + uint64_t deadline; + uint64_t time_limit = time_start + tlp->scan_limit; + mpqueue_head_t *timer_master_queue; + + assert(!ml_get_interrupts_enabled()); + assert(cpu_number() == master_cpu); + + if (tlp->threshold.interval != TIMER_LONGTERM_NONE) + threshold = time_start + tlp->threshold.interval; + + tlp->threshold.deadline = TIMER_LONGTERM_NONE; + tlp->threshold.call = NULL; + + if (queue_empty(&timer_longterm_queue->head)) + return; + + timer_master_queue = timer_queue_cpu(master_cpu); + timer_queue_lock_spin(timer_master_queue); + + qe = queue_first(&timer_longterm_queue->head); + while (!queue_end(&timer_longterm_queue->head, qe)) { + call = TIMER_CALL(qe); + deadline = call->soft_deadline; + qe = queue_next(qe); + if (!simple_lock_try(&call->lock)) { + /* case (2c) lock order inversion, dequeue only */ +#ifdef TIMER_ASSERT + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_ASYNC_DEQ | DBG_FUNC_NONE, + VM_KERNEL_UNSLIDE_OR_PERM(call), + VM_KERNEL_UNSLIDE_OR_PERM(TCE(call)->queue), + VM_KERNEL_UNSLIDE_OR_PERM(call->lock.interlock.lock_data), + 0x2c, 0); +#endif + timer_call_entry_dequeue_async(call); + continue; + } + if (deadline < threshold) { + /* + * This timer needs moving (escalating) + * to the local (boot) processor's queue. + */ +#ifdef TIMER_ASSERT + if (deadline < time_start) + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_OVERDUE | DBG_FUNC_NONE, + VM_KERNEL_UNSLIDE_OR_PERM(call), + deadline, + time_start, + threshold, + 0); +#endif + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_ESCALATE | DBG_FUNC_NONE, + VM_KERNEL_UNSLIDE_OR_PERM(call), + TCE(call)->deadline, + TCE(call)->entry_time, + VM_KERNEL_UNSLIDE(TCE(call)->func), + 0); + tlp->escalates++; + timer_call_entry_dequeue(call); + timer_call_entry_enqueue_deadline( + call, timer_master_queue, TCE(call)->deadline); + /* + * A side-effect of the following call is to update + * the actual hardware deadline if required. + */ + (void) timer_queue_assign(deadline); + } else { + if (deadline < tlp->threshold.deadline) { + tlp->threshold.deadline = deadline; + tlp->threshold.call = call; + } + } + simple_unlock(&call->lock); + + /* Abort scan if we're taking too long. */ + if (mach_absolute_time() > time_limit) { + tlp->threshold.deadline = TIMER_LONGTERM_SCAN_AGAIN; + tlp->scan_pauses++; + DBG("timer_longterm_scan() paused %llu, qlen: %llu\n", + time_limit, tlp->queue.count); + break; + } + } + + timer_queue_unlock(timer_master_queue); +} + +void +timer_longterm_callout(timer_call_param_t p0, __unused timer_call_param_t p1) +{ + timer_longterm_t *tlp = (timer_longterm_t *) p0; + + timer_longterm_update(tlp); +} + +void +timer_longterm_update_locked(timer_longterm_t *tlp) +{ + uint64_t latency; + + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_UPDATE | DBG_FUNC_START, + VM_KERNEL_UNSLIDE_OR_PERM(&tlp->queue), + tlp->threshold.deadline, + tlp->threshold.preempted, + tlp->queue.count, 0); + + tlp->scan_time = mach_absolute_time(); + if (tlp->threshold.preempted != TIMER_LONGTERM_NONE) { + tlp->threshold.preempts++; + tlp->threshold.deadline = tlp->threshold.preempted; + tlp->threshold.preempted = TIMER_LONGTERM_NONE; + /* + * Note: in the unlikely event that a pre-empted timer has + * itself been cancelled, we'll simply re-scan later at the + * time of the preempted/cancelled timer. + */ + } else { + tlp->threshold.scans++; + + /* + * Maintain a moving average of our wakeup latency. + * Clamp latency to 0 and ignore above threshold interval. + */ + if (tlp->scan_time > tlp->threshold.deadline_set) + latency = tlp->scan_time - tlp->threshold.deadline_set; + else + latency = 0; + if (latency < tlp->threshold.interval) { + tlp->threshold.latency_min = + MIN(tlp->threshold.latency_min, latency); + tlp->threshold.latency_max = + MAX(tlp->threshold.latency_max, latency); + tlp->threshold.latency = + (tlp->threshold.latency*99 + latency) / 100; + } + + timer_longterm_scan(tlp, tlp->scan_time); + } + + tlp->threshold.deadline_set = tlp->threshold.deadline; + /* The next deadline timer to be set is adjusted */ + if (tlp->threshold.deadline != TIMER_LONGTERM_NONE && + tlp->threshold.deadline != TIMER_LONGTERM_SCAN_AGAIN) { + tlp->threshold.deadline_set -= tlp->threshold.margin; + tlp->threshold.deadline_set -= tlp->threshold.latency; + } + + /* Throttle next scan time */ + uint64_t scan_clamp = mach_absolute_time() + tlp->scan_interval; + if (tlp->threshold.deadline_set < scan_clamp) + tlp->threshold.deadline_set = scan_clamp; + + TIMER_KDEBUG_TRACE(KDEBUG_TRACE, + DECR_TIMER_UPDATE | DBG_FUNC_END, + VM_KERNEL_UNSLIDE_OR_PERM(&tlp->queue), + tlp->threshold.deadline, + tlp->threshold.scans, + tlp->queue.count, 0); +} + +void +timer_longterm_update(timer_longterm_t *tlp) +{ + spl_t s = splclock(); + + timer_queue_lock_spin(timer_longterm_queue); + + if (cpu_number() != master_cpu) + panic("timer_longterm_update_master() on non-boot cpu"); + + timer_longterm_update_locked(tlp); + + if (tlp->threshold.deadline != TIMER_LONGTERM_NONE) + timer_call_enter( + &tlp->threshold.timer, + tlp->threshold.deadline_set, + TIMER_CALL_LOCAL | TIMER_CALL_SYS_CRITICAL); + + timer_queue_unlock(timer_longterm_queue); + splx(s); +} + +void +timer_longterm_init(void) +{ + uint32_t longterm; + timer_longterm_t *tlp = &timer_longterm; + + DBG("timer_longterm_init() tlp: %p, queue: %p\n", tlp, &tlp->queue); + + /* + * Set the longterm timer threshold. Defaults to TIMER_LONGTERM_THRESHOLD + * or TIMER_LONGTERM_NONE (disabled) for server; + * overridden longterm boot-arg + */ + tlp->threshold.interval = serverperfmode ? TIMER_LONGTERM_NONE + : TIMER_LONGTERM_THRESHOLD; + if (PE_parse_boot_argn("longterm", &longterm, sizeof (longterm))) { + tlp->threshold.interval = (longterm == 0) ? + TIMER_LONGTERM_NONE : + longterm * NSEC_PER_MSEC; + } + if (tlp->threshold.interval != TIMER_LONGTERM_NONE) { + printf("Longterm timer threshold: %llu ms\n", + tlp->threshold.interval / NSEC_PER_MSEC); + kprintf("Longterm timer threshold: %llu ms\n", + tlp->threshold.interval / NSEC_PER_MSEC); + nanoseconds_to_absolutetime(tlp->threshold.interval, + &tlp->threshold.interval); + tlp->threshold.margin = tlp->threshold.interval / 10; + tlp->threshold.latency_min = EndOfAllTime; + tlp->threshold.latency_max = 0; + } + + tlp->threshold.preempted = TIMER_LONGTERM_NONE; + tlp->threshold.deadline = TIMER_LONGTERM_NONE; + + lck_attr_setdefault(&timer_longterm_lck_attr); + lck_grp_attr_setdefault(&timer_longterm_lck_grp_attr); + lck_grp_init(&timer_longterm_lck_grp, + "timer_longterm", &timer_longterm_lck_grp_attr); + mpqueue_init(&tlp->queue, + &timer_longterm_lck_grp, &timer_longterm_lck_attr); + + timer_call_setup(&tlp->threshold.timer, + timer_longterm_callout, (timer_call_param_t) tlp); + + timer_longterm_queue = &tlp->queue; +} + +enum { + THRESHOLD, QCOUNT, + ENQUEUES, DEQUEUES, ESCALATES, SCANS, PREEMPTS, + LATENCY, LATENCY_MIN, LATENCY_MAX, SCAN_LIMIT, SCAN_INTERVAL, PAUSES +}; +uint64_t +timer_sysctl_get(int oid) +{ + timer_longterm_t *tlp = &timer_longterm; + + switch (oid) { + case THRESHOLD: + return (tlp->threshold.interval == TIMER_LONGTERM_NONE) ? + 0 : tlp->threshold.interval / NSEC_PER_MSEC; + case QCOUNT: + return tlp->queue.count; + case ENQUEUES: + return tlp->enqueues; + case DEQUEUES: + return tlp->dequeues; + case ESCALATES: + return tlp->escalates; + case SCANS: + return tlp->threshold.scans; + case PREEMPTS: + return tlp->threshold.preempts; + case LATENCY: + return tlp->threshold.latency; + case LATENCY_MIN: + return tlp->threshold.latency_min; + case LATENCY_MAX: + return tlp->threshold.latency_max; + case SCAN_LIMIT: + return tlp->scan_limit; + case SCAN_INTERVAL: + return tlp->scan_interval; + case PAUSES: + return tlp->scan_pauses; + default: + return 0; + } +} + +/* + * timer_master_scan() is the inverse of timer_longterm_scan() + * since it un-escalates timers to the longterm queue. + */ +static void +timer_master_scan(timer_longterm_t *tlp, + uint64_t now) +{ + queue_entry_t qe; + timer_call_t call; + uint64_t threshold; + uint64_t deadline; + mpqueue_head_t *timer_master_queue; + + if (tlp->threshold.interval != TIMER_LONGTERM_NONE) + threshold = now + tlp->threshold.interval; + else + threshold = TIMER_LONGTERM_NONE; + + timer_master_queue = timer_queue_cpu(master_cpu); + timer_queue_lock_spin(timer_master_queue); + + qe = queue_first(&timer_master_queue->head); + while (!queue_end(&timer_master_queue->head, qe)) { + call = TIMER_CALL(qe); + deadline = TCE(call)->deadline; + qe = queue_next(qe); + if ((call->flags & TIMER_CALL_LOCAL) != 0) + continue; + if (!simple_lock_try(&call->lock)) { + /* case (2c) lock order inversion, dequeue only */ + timer_call_entry_dequeue_async(call); + continue; + } + if (deadline > threshold) { + /* move from master to longterm */ + timer_call_entry_dequeue(call); + timer_call_entry_enqueue_tail(call, timer_longterm_queue); + if (deadline < tlp->threshold.deadline) { + tlp->threshold.deadline = deadline; + tlp->threshold.call = call; + } + } + simple_unlock(&call->lock); + } + timer_queue_unlock(timer_master_queue); +} + +static void +timer_sysctl_set_threshold(uint64_t value) +{ + timer_longterm_t *tlp = &timer_longterm; + spl_t s = splclock(); + boolean_t threshold_increase; + + timer_queue_lock_spin(timer_longterm_queue); + + timer_call_cancel(&tlp->threshold.timer); + + /* + * Set the new threshold and note whther it's increasing. + */ + if (value == 0) { + tlp->threshold.interval = TIMER_LONGTERM_NONE; + threshold_increase = TRUE; + timer_call_cancel(&tlp->threshold.timer); + } else { + uint64_t old_interval = tlp->threshold.interval; + tlp->threshold.interval = value * NSEC_PER_MSEC; + nanoseconds_to_absolutetime(tlp->threshold.interval, + &tlp->threshold.interval); + tlp->threshold.margin = tlp->threshold.interval / 10; + if (old_interval == TIMER_LONGTERM_NONE) + threshold_increase = FALSE; + else + threshold_increase = (tlp->threshold.interval > old_interval); + } + + if (threshold_increase /* or removal */) { + /* Escalate timers from the longterm queue */ + timer_longterm_scan(tlp, mach_absolute_time()); + } else /* decrease or addition */ { + /* + * We scan the local/master queue for timers now longterm. + * To be strictly correct, we should scan all processor queues + * but timer migration results in most timers gravitating to the + * master processor in any case. + */ + timer_master_scan(tlp, mach_absolute_time()); + } + + /* Set new timer accordingly */ + tlp->threshold.deadline_set = tlp->threshold.deadline; + if (tlp->threshold.deadline != TIMER_LONGTERM_NONE) { + tlp->threshold.deadline_set -= tlp->threshold.margin; + tlp->threshold.deadline_set -= tlp->threshold.latency; + timer_call_enter( + &tlp->threshold.timer, + tlp->threshold.deadline_set, + TIMER_CALL_LOCAL | TIMER_CALL_SYS_CRITICAL); + } + + /* Reset stats */ + tlp->enqueues = 0; + tlp->dequeues = 0; + tlp->escalates = 0; + tlp->scan_pauses = 0; + tlp->threshold.scans = 0; + tlp->threshold.preempts = 0; + tlp->threshold.latency = 0; + tlp->threshold.latency_min = EndOfAllTime; + tlp->threshold.latency_max = 0; + + timer_queue_unlock(timer_longterm_queue); + splx(s); +} + +int +timer_sysctl_set(int oid, uint64_t value) +{ + switch (oid) { + case THRESHOLD: + timer_call_cpu( + master_cpu, + (void (*)(void *)) timer_sysctl_set_threshold, + (void *) value); + return KERN_SUCCESS; + case SCAN_LIMIT: + timer_longterm.scan_limit = value; + return KERN_SUCCESS; + case SCAN_INTERVAL: + timer_longterm.scan_interval = value; + return KERN_SUCCESS; + default: + return KERN_INVALID_ARGUMENT; + } +} + + +/* Select timer coalescing window based on per-task quality-of-service hints */ +static boolean_t tcoal_qos_adjust(thread_t t, int32_t *tshift, uint64_t *tmax_abstime, boolean_t *pratelimited) { + uint32_t latency_qos; + boolean_t adjusted = FALSE; + task_t ctask = t->task; + + if (ctask) { + latency_qos = proc_get_effective_thread_policy(t, TASK_POLICY_LATENCY_QOS); + + assert(latency_qos <= NUM_LATENCY_QOS_TIERS); + + if (latency_qos) { + *tshift = tcoal_prio_params.latency_qos_scale[latency_qos - 1]; + *tmax_abstime = tcoal_prio_params.latency_qos_abstime_max[latency_qos - 1]; + *pratelimited = tcoal_prio_params.latency_tier_rate_limited[latency_qos - 1]; + adjusted = TRUE; + } + } + return adjusted; +} + + +/* Adjust timer deadlines based on priority of the thread and the + * urgency value provided at timeout establishment. With this mechanism, + * timers are no longer necessarily sorted in order of soft deadline + * on a given timer queue, i.e. they may be differentially skewed. + * In the current scheme, this could lead to fewer pending timers + * processed than is technically possible when the HW deadline arrives. + */ +static void +timer_compute_leeway(thread_t cthread, int32_t urgency, int32_t *tshift, uint64_t *tmax_abstime, boolean_t *pratelimited) { + int16_t tpri = cthread->sched_pri; + if ((urgency & TIMER_CALL_USER_MASK) != 0) { + if (tpri >= BASEPRI_RTQUEUES || + urgency == TIMER_CALL_USER_CRITICAL) { + *tshift = tcoal_prio_params.timer_coalesce_rt_shift; + *tmax_abstime = tcoal_prio_params.timer_coalesce_rt_abstime_max; + TCOAL_PRIO_STAT(rt_tcl); + } else if (proc_get_effective_thread_policy(cthread, TASK_POLICY_DARWIN_BG) || + (urgency == TIMER_CALL_USER_BACKGROUND)) { + /* Determine if timer should be subjected to a lower QoS */ + if (tcoal_qos_adjust(cthread, tshift, tmax_abstime, pratelimited)) { + if (*tmax_abstime > tcoal_prio_params.timer_coalesce_bg_abstime_max) { + return; + } else { + *pratelimited = FALSE; + } + } + *tshift = tcoal_prio_params.timer_coalesce_bg_shift; + *tmax_abstime = tcoal_prio_params.timer_coalesce_bg_abstime_max; + TCOAL_PRIO_STAT(bg_tcl); + } else if (tpri >= MINPRI_KERNEL) { + *tshift = tcoal_prio_params.timer_coalesce_kt_shift; + *tmax_abstime = tcoal_prio_params.timer_coalesce_kt_abstime_max; + TCOAL_PRIO_STAT(kt_tcl); + } else if (cthread->sched_mode == TH_MODE_FIXED) { + *tshift = tcoal_prio_params.timer_coalesce_fp_shift; + *tmax_abstime = tcoal_prio_params.timer_coalesce_fp_abstime_max; + TCOAL_PRIO_STAT(fp_tcl); + } else if (tcoal_qos_adjust(cthread, tshift, tmax_abstime, pratelimited)) { + TCOAL_PRIO_STAT(qos_tcl); + } else if (cthread->sched_mode == TH_MODE_TIMESHARE) { + *tshift = tcoal_prio_params.timer_coalesce_ts_shift; + *tmax_abstime = tcoal_prio_params.timer_coalesce_ts_abstime_max; + TCOAL_PRIO_STAT(ts_tcl); + } else { + TCOAL_PRIO_STAT(nc_tcl); + } + } else if (urgency == TIMER_CALL_SYS_BACKGROUND) { + *tshift = tcoal_prio_params.timer_coalesce_bg_shift; + *tmax_abstime = tcoal_prio_params.timer_coalesce_bg_abstime_max; + TCOAL_PRIO_STAT(bg_tcl); + } else { + *tshift = tcoal_prio_params.timer_coalesce_kt_shift; + *tmax_abstime = tcoal_prio_params.timer_coalesce_kt_abstime_max; + TCOAL_PRIO_STAT(kt_tcl); + } +} + + +int timer_user_idle_level; + +uint64_t +timer_call_slop(uint64_t deadline, uint64_t now, uint32_t flags, thread_t cthread, boolean_t *pratelimited) +{ + int32_t tcs_shift = 0; + uint64_t tcs_max_abstime = 0; + uint64_t adjval; + uint32_t urgency = (flags & TIMER_CALL_URGENCY_MASK); + + if (mach_timer_coalescing_enabled && + (deadline > now) && (urgency != TIMER_CALL_SYS_CRITICAL)) { + timer_compute_leeway(cthread, urgency, &tcs_shift, &tcs_max_abstime, pratelimited); + + if (tcs_shift >= 0) + adjval = MIN((deadline - now) >> tcs_shift, tcs_max_abstime); + else + adjval = MIN((deadline - now) << (-tcs_shift), tcs_max_abstime); + /* Apply adjustments derived from "user idle level" heuristic */ + adjval += (adjval * timer_user_idle_level) >> 7; + return adjval; + } else { + return 0; + } +} + +int +timer_get_user_idle_level(void) { + return timer_user_idle_level; +} + +kern_return_t timer_set_user_idle_level(int ilevel) { + boolean_t do_reeval = FALSE; + + if ((ilevel < 0) || (ilevel > 128)) + return KERN_INVALID_ARGUMENT; + + if (ilevel < timer_user_idle_level) { + do_reeval = TRUE; + } + + timer_user_idle_level = ilevel; + + if (do_reeval) + ml_timer_evaluate(); + + return KERN_SUCCESS; +}