]> git.saurik.com Git - apple/xnu.git/blobdiff - osfmk/kern/sched_prim.c
xnu-2782.1.97.tar.gz
[apple/xnu.git] / osfmk / kern / sched_prim.c
index 11bad560aaa1f0ac63f0edd6d600e2982179961e..3a88e11fb4fcf2d25f9ea6eefe81b6314aa3731f 100644 (file)
 #include <machine/machine_cpu.h>
 #include <machine/machlimits.h>
 
+#ifdef CONFIG_MACH_APPROXIMATE_TIME
+#include <machine/commpage.h>
+#endif
+
 #include <kern/kern_types.h>
 #include <kern/clock.h>
 #include <kern/counters.h>
 #include <kern/cpu_number.h>
 #include <kern/cpu_data.h>
 #include <kern/debug.h>
-#include <kern/lock.h>
 #include <kern/macro_help.h>
 #include <kern/machine.h>
 #include <kern/misc_protos.h>
@@ -91,6 +94,7 @@
 #include <kern/queue.h>
 #include <kern/sched.h>
 #include <kern/sched_prim.h>
+#include <kern/sfi.h>
 #include <kern/syscall_subr.h>
 #include <kern/task.h>
 #include <kern/thread.h>
 
 #include <kern/pms.h>
 
+#if defined(CONFIG_TELEMETRY) && defined(CONFIG_SCHED_TIMESHARE_CORE)
+#include <kern/telemetry.h>
+#endif
+
 struct rt_queue        rt_runq;
 #define RT_RUNQ                ((processor_t)-1)
 decl_simple_lock_data(static,rt_lock);
 
-#if defined(CONFIG_SCHED_TRADITIONAL) || defined(CONFIG_SCHED_PROTO) || defined(CONFIG_SCHED_GRRR) || defined(CONFIG_SCHED_FIXEDPRIORITY)
+#if defined(CONFIG_SCHED_FAIRSHARE_CORE)
 static struct fairshare_queue  fs_runq;
 #define FS_RUNQ                ((processor_t)-2)
 decl_simple_lock_data(static,fs_lock);
-#endif
+#endif /* CONFIG_SCHED_FAIRSHARE_CORE */
 
 #define                DEFAULT_PREEMPTION_RATE         100             /* (1/s) */
 int                    default_preemption_rate = DEFAULT_PREEMPTION_RATE;
@@ -138,7 +146,7 @@ uint64_t    max_poll_computation;
 uint64_t       max_unsafe_computation;
 uint64_t       sched_safe_duration;
 
-#if defined(CONFIG_SCHED_TRADITIONAL)
+#if defined(CONFIG_SCHED_TIMESHARE_CORE)
 
 uint32_t       std_quantum;
 uint32_t       min_std_quantum;
@@ -147,7 +155,7 @@ uint32_t    bg_quantum;
 uint32_t       std_quantum_us;
 uint32_t       bg_quantum_us;
 
-#endif /* CONFIG_SCHED_TRADITIONAL */
+#endif /* CONFIG_SCHED_TIMESHARE_CORE */
 
 uint32_t       thread_depress_time;
 uint32_t       default_timeshare_computation;
@@ -156,10 +164,13 @@ uint32_t  default_timeshare_constraint;
 uint32_t       max_rt_quantum;
 uint32_t       min_rt_quantum;
 
-#if defined(CONFIG_SCHED_TRADITIONAL)
+#if defined(CONFIG_SCHED_TIMESHARE_CORE)
 
 unsigned       sched_tick;
 uint32_t       sched_tick_interval;
+#if defined(CONFIG_TELEMETRY)
+uint32_t       sched_telemetry_interval;
+#endif /* CONFIG_TELEMETRY */
 
 uint32_t       sched_pri_shift = INT8_MAX;
 uint32_t       sched_background_pri_shift = INT8_MAX;
@@ -169,7 +180,9 @@ uint32_t    sched_use_combined_fgbg_decay = 0;
 
 uint32_t       sched_decay_usage_age_factor = 1; /* accelerate 5/8^n usage aging */
 
-static boolean_t sched_traditional_use_pset_runqueue = FALSE;
+/* Allow foreground to decay past default to resolve inversions */
+#define DEFAULT_DECAY_BAND_LIMIT ((BASEPRI_FOREGROUND - BASEPRI_DEFAULT) + 2)
+int            sched_pri_decay_band_limit = DEFAULT_DECAY_BAND_LIMIT;
 
 /* Defaults for timer deadline profiling */
 #define TIMER_DEADLINE_TRACKING_BIN_1_DEFAULT 2000000 /* Timers with deadlines <=
@@ -182,6 +195,12 @@ uint64_t timer_deadline_tracking_bin_2;
 
 thread_t sched_maintenance_thread;
 
+#endif /* CONFIG_SCHED_TIMESHARE_CORE */
+
+#if defined(CONFIG_SCHED_TRADITIONAL)
+
+static boolean_t sched_traditional_use_pset_runqueue = FALSE;
+
 __attribute__((always_inline))
 static inline run_queue_t runq_for_processor(processor_t processor)
 {
@@ -228,16 +247,17 @@ uint32_t  sched_load_average, sched_mach_factor;
 
 /* Forwards */
 
-#if defined(CONFIG_SCHED_TRADITIONAL)
+#if defined(CONFIG_SCHED_TIMESHARE_CORE)
 
 static void load_shift_init(void);
 static void preempt_pri_init(void);
 
-#endif /* CONFIG_SCHED_TRADITIONAL */
+#endif /* CONFIG_SCHED_TIMESHARE_CORE */
 
 static thread_t        thread_select(
                                        thread_t                        thread,
-                                       processor_t                     processor);
+                                       processor_t                     processor,
+                                       ast_t                           reason);
 
 #if CONFIG_SCHED_IDLE_IN_PLACE
 static thread_t        thread_select_idle(
@@ -251,7 +271,8 @@ thread_t    processor_idle(
 
 ast_t
 csw_check_locked(      processor_t             processor,
-                                       processor_set_t pset);
+                                       processor_set_t pset,
+                                       ast_t                   check_reason);
 
 #if defined(CONFIG_SCHED_TRADITIONAL)
 
@@ -285,8 +306,6 @@ processor_queue_remove(
 
 static boolean_t       processor_queue_empty(processor_t               processor);
 
-static boolean_t       priority_is_urgent(int priority);
-
 static ast_t           processor_csw_check(processor_t processor);
 
 static boolean_t       processor_queue_has_priority(processor_t                processor,
@@ -302,16 +321,13 @@ static boolean_t  sched_traditional_with_pset_runqueue_processor_queue_empty(proc
 static uint64_t     sched_traditional_processor_runq_stats_count_sum(processor_t   processor);
 
 static uint64_t                sched_traditional_with_pset_runqueue_processor_runq_stats_count_sum(processor_t   processor);
-#endif
-       
 
-#if defined(CONFIG_SCHED_TRADITIONAL)
+static int      sched_traditional_processor_bound_count(processor_t processor);
 
-static void
-sched_traditional_init(void);
+#endif
 
-static void
-sched_traditional_timebase_init(void);
+
+#if defined(CONFIG_SCHED_TRADITIONAL)
 
 static void
 sched_traditional_processor_init(processor_t processor);
@@ -334,22 +350,15 @@ static void
 sched_timer_deadline_tracking_init(void);
 
 #if defined(CONFIG_SCHED_TRADITIONAL)
-static void
-sched_traditional_maintenance_continue(void);
-
-static uint32_t
-sched_traditional_initial_quantum_size(thread_t thread);
 
 static sched_mode_t
 sched_traditional_initial_thread_sched_mode(task_t parent_task);
 
-static boolean_t
-sched_traditional_supports_timeshare_mode(void);
-
 static thread_t
 sched_traditional_choose_thread(
-                                                               processor_t             processor,
-                                                               int                             priority);
+                                processor_t     processor,
+                                int             priority,
+                       __unused ast_t           reason);
 
 #endif
 
@@ -360,13 +369,10 @@ extern int debug_task;
 #define TLOG(a, fmt, args...) do {} while (0)
 #endif
 
-#if    DEBUG
-static
+__assert_only static
 boolean_t      thread_runnable(
                                thread_t                thread);
 
-#endif /*DEBUG*/
-
 /*
  *     State machine
  *
@@ -393,84 +399,86 @@ boolean_t thread_runnable(
  *
  */
 
-#if defined(CONFIG_SCHED_TRADITIONAL)
+#if defined(CONFIG_SCHED_TIMESHARE_CORE)
 int8_t         sched_load_shifts[NRQS];
 int            sched_preempt_pri[NRQBM];
-#endif
+#endif /* CONFIG_SCHED_TIMESHARE_CORE */
 
 
 #if defined(CONFIG_SCHED_TRADITIONAL)
 
 const struct sched_dispatch_table sched_traditional_dispatch = {
-       sched_traditional_init,
-       sched_traditional_timebase_init,
-       sched_traditional_processor_init,
-       sched_traditional_pset_init,
-       sched_traditional_maintenance_continue,
-       sched_traditional_choose_thread,
-       steal_thread,
-       compute_priority,
-       choose_processor,
-       processor_enqueue,
-       processor_queue_shutdown,
-       processor_queue_remove,
-       processor_queue_empty,
-       priority_is_urgent,
-       processor_csw_check,
-       processor_queue_has_priority,
-       sched_traditional_initial_quantum_size,
-       sched_traditional_initial_thread_sched_mode,
-       sched_traditional_supports_timeshare_mode,
-       can_update_priority,
-       update_priority,
-       lightweight_update_priority,
-       sched_traditional_quantum_expire,
-       should_current_thread_rechoose_processor,
-       sched_traditional_processor_runq_count,
-       sched_traditional_processor_runq_stats_count_sum,
-       sched_traditional_fairshare_init,
-       sched_traditional_fairshare_runq_count,
-       sched_traditional_fairshare_runq_stats_count_sum,
-       sched_traditional_fairshare_enqueue,
-       sched_traditional_fairshare_dequeue,
-       sched_traditional_fairshare_queue_remove,
-       TRUE /* direct_dispatch_to_idle_processors */
+       .init                                           = sched_traditional_init,
+       .timebase_init                                  = sched_traditional_timebase_init,
+       .processor_init                                 = sched_traditional_processor_init,
+       .pset_init                                      = sched_traditional_pset_init,
+       .maintenance_continuation                       = sched_traditional_maintenance_continue,
+       .choose_thread                                  = sched_traditional_choose_thread,
+       .steal_thread                                   = steal_thread,
+       .compute_priority                               = compute_priority,
+       .choose_processor                               = choose_processor,
+       .processor_enqueue                              = processor_enqueue,
+       .processor_queue_shutdown                       = processor_queue_shutdown,
+       .processor_queue_remove                         = processor_queue_remove,
+       .processor_queue_empty                          = processor_queue_empty,
+       .priority_is_urgent                             = priority_is_urgent,
+       .processor_csw_check                            = processor_csw_check,
+       .processor_queue_has_priority                   = processor_queue_has_priority,
+       .initial_quantum_size                           = sched_traditional_initial_quantum_size,
+       .initial_thread_sched_mode                      = sched_traditional_initial_thread_sched_mode,
+       .can_update_priority                            = can_update_priority,
+       .update_priority                                = update_priority,
+       .lightweight_update_priority                    = lightweight_update_priority,
+       .quantum_expire                                 = sched_traditional_quantum_expire,
+       .should_current_thread_rechoose_processor       = should_current_thread_rechoose_processor,
+       .processor_runq_count                           = sched_traditional_processor_runq_count,
+       .processor_runq_stats_count_sum                 = sched_traditional_processor_runq_stats_count_sum,
+       .fairshare_init                                 = sched_traditional_fairshare_init,
+       .fairshare_runq_count                           = sched_traditional_fairshare_runq_count,
+       .fairshare_runq_stats_count_sum                 = sched_traditional_fairshare_runq_stats_count_sum,
+       .fairshare_enqueue                              = sched_traditional_fairshare_enqueue,
+       .fairshare_dequeue                              = sched_traditional_fairshare_dequeue,
+       .fairshare_queue_remove                         = sched_traditional_fairshare_queue_remove,
+       .processor_bound_count                          = sched_traditional_processor_bound_count,
+       .thread_update_scan                             = thread_update_scan,
+       .direct_dispatch_to_idle_processors             = TRUE,
 };
 
 const struct sched_dispatch_table sched_traditional_with_pset_runqueue_dispatch = {
-       sched_traditional_with_pset_runqueue_init,
-       sched_traditional_timebase_init,
-       sched_traditional_processor_init,
-       sched_traditional_pset_init,
-       sched_traditional_maintenance_continue,
-       sched_traditional_choose_thread,
-       steal_thread,
-       compute_priority,
-       choose_processor,
-       processor_enqueue,
-       processor_queue_shutdown,
-       processor_queue_remove,
-       sched_traditional_with_pset_runqueue_processor_queue_empty,
-       priority_is_urgent,
-       processor_csw_check,
-       processor_queue_has_priority,
-       sched_traditional_initial_quantum_size,
-       sched_traditional_initial_thread_sched_mode,
-       sched_traditional_supports_timeshare_mode,
-       can_update_priority,
-       update_priority,
-       lightweight_update_priority,
-       sched_traditional_quantum_expire,
-       should_current_thread_rechoose_processor,
-       sched_traditional_processor_runq_count,
-       sched_traditional_with_pset_runqueue_processor_runq_stats_count_sum,
-       sched_traditional_fairshare_init,
-       sched_traditional_fairshare_runq_count,
-       sched_traditional_fairshare_runq_stats_count_sum,
-       sched_traditional_fairshare_enqueue,
-       sched_traditional_fairshare_dequeue,
-       sched_traditional_fairshare_queue_remove,
-       FALSE /* direct_dispatch_to_idle_processors */
+       .init                                           = sched_traditional_with_pset_runqueue_init,
+       .timebase_init                                  = sched_traditional_timebase_init,
+       .processor_init                                 = sched_traditional_processor_init,
+       .pset_init                                      = sched_traditional_pset_init,
+       .maintenance_continuation                       = sched_traditional_maintenance_continue,
+       .choose_thread                                  = sched_traditional_choose_thread,
+       .steal_thread                                   = steal_thread,
+       .compute_priority                               = compute_priority,
+       .choose_processor                               = choose_processor,
+       .processor_enqueue                              = processor_enqueue,
+       .processor_queue_shutdown                       = processor_queue_shutdown,
+       .processor_queue_remove                         = processor_queue_remove,
+       .processor_queue_empty                          = sched_traditional_with_pset_runqueue_processor_queue_empty,
+       .priority_is_urgent                             = priority_is_urgent,
+       .processor_csw_check                            = processor_csw_check,
+       .processor_queue_has_priority                   = processor_queue_has_priority,
+       .initial_quantum_size                           = sched_traditional_initial_quantum_size,
+       .initial_thread_sched_mode                      = sched_traditional_initial_thread_sched_mode,
+       .can_update_priority                            = can_update_priority,
+       .update_priority                                = update_priority,
+       .lightweight_update_priority                    = lightweight_update_priority,
+       .quantum_expire                                 = sched_traditional_quantum_expire,
+       .should_current_thread_rechoose_processor       = should_current_thread_rechoose_processor,
+       .processor_runq_count                           = sched_traditional_processor_runq_count,
+       .processor_runq_stats_count_sum                 = sched_traditional_with_pset_runqueue_processor_runq_stats_count_sum,
+       .fairshare_init                                 = sched_traditional_fairshare_init,
+       .fairshare_runq_count                           = sched_traditional_fairshare_runq_count,
+       .fairshare_runq_stats_count_sum                 = sched_traditional_fairshare_runq_stats_count_sum,
+       .fairshare_enqueue                              = sched_traditional_fairshare_enqueue,
+       .fairshare_dequeue                              = sched_traditional_fairshare_dequeue,
+       .fairshare_queue_remove                         = sched_traditional_fairshare_queue_remove,
+       .processor_bound_count                          = sched_traditional_processor_bound_count,
+       .thread_update_scan                             = thread_update_scan,
+       .direct_dispatch_to_idle_processors             = FALSE,
 };
 
 #endif
@@ -516,6 +524,19 @@ sched_init(void)
                }
        }
 
+       
+       if (!PE_parse_boot_argn("sched_pri_decay_limit", &sched_pri_decay_band_limit, sizeof(sched_pri_decay_band_limit))) {
+               /* No boot-args, check in device tree */
+               if (!PE_get_default("kern.sched_pri_decay_limit",
+                                                       &sched_pri_decay_band_limit,
+                                                       sizeof(sched_pri_decay_band_limit))) {
+                       /* Allow decay all the way to normal limits */
+                       sched_pri_decay_band_limit = DEFAULT_DECAY_BAND_LIMIT;
+               }
+       }
+
+       kprintf("Setting scheduler priority decay band limit %d\n", sched_pri_decay_band_limit);
+
        if (strlen(sched_arg) > 0) {
                if (0) {
                        /* Allow pattern below */
@@ -524,66 +545,67 @@ sched_init(void)
                        sched_current_dispatch = &sched_traditional_dispatch;
                        _sched_enum = sched_enum_traditional;
                        strlcpy(sched_string, kSchedTraditionalString, sizeof(sched_string));
-                       kprintf("Scheduler: Runtime selection of %s\n", kSchedTraditionalString);
                } else if (0 == strcmp(sched_arg, kSchedTraditionalWithPsetRunqueueString)) {
                        sched_current_dispatch = &sched_traditional_with_pset_runqueue_dispatch;
                        _sched_enum = sched_enum_traditional_with_pset_runqueue;
                        strlcpy(sched_string, kSchedTraditionalWithPsetRunqueueString, sizeof(sched_string));
-                       kprintf("Scheduler: Runtime selection of %s\n", kSchedTraditionalWithPsetRunqueueString);
 #endif
 #if defined(CONFIG_SCHED_PROTO)
                } else if (0 == strcmp(sched_arg, kSchedProtoString)) {
                        sched_current_dispatch = &sched_proto_dispatch;
                        _sched_enum = sched_enum_proto;
                        strlcpy(sched_string, kSchedProtoString, sizeof(sched_string));
-                       kprintf("Scheduler: Runtime selection of %s\n", kSchedProtoString);
 #endif
 #if defined(CONFIG_SCHED_GRRR)
                } else if (0 == strcmp(sched_arg, kSchedGRRRString)) {
                        sched_current_dispatch = &sched_grrr_dispatch;
                        _sched_enum = sched_enum_grrr;
                        strlcpy(sched_string, kSchedGRRRString, sizeof(sched_string));
-                       kprintf("Scheduler: Runtime selection of %s\n", kSchedGRRRString);
 #endif
-#if defined(CONFIG_SCHED_FIXEDPRIORITY)
-               } else if (0 == strcmp(sched_arg, kSchedFixedPriorityString)) {
-                       sched_current_dispatch = &sched_fixedpriority_dispatch;
-                       _sched_enum = sched_enum_fixedpriority;
-                       strlcpy(sched_string, kSchedFixedPriorityString, sizeof(sched_string));
-                       kprintf("Scheduler: Runtime selection of %s\n", kSchedFixedPriorityString);
-               } else if (0 == strcmp(sched_arg, kSchedFixedPriorityWithPsetRunqueueString)) {
-                       sched_current_dispatch = &sched_fixedpriority_with_pset_runqueue_dispatch;
-                       _sched_enum = sched_enum_fixedpriority_with_pset_runqueue;
-                       strlcpy(sched_string, kSchedFixedPriorityWithPsetRunqueueString, sizeof(sched_string));
-                       kprintf("Scheduler: Runtime selection of %s\n", kSchedFixedPriorityWithPsetRunqueueString);
+#if defined(CONFIG_SCHED_MULTIQ)
+               } else if (0 == strcmp(sched_arg, kSchedMultiQString)) {
+                       sched_current_dispatch = &sched_multiq_dispatch;
+                       _sched_enum = sched_enum_multiq;
+                       strlcpy(sched_string, kSchedMultiQString, sizeof(sched_string));
+               } else if (0 == strcmp(sched_arg, kSchedDualQString)) {
+                       sched_current_dispatch = &sched_dualq_dispatch;
+                       _sched_enum = sched_enum_dualq;
+                       strlcpy(sched_string, kSchedDualQString, sizeof(sched_string));
 #endif
                } else {
+#if defined(CONFIG_SCHED_TRADITIONAL)
+                       printf("Unrecognized scheduler algorithm: %s\n", sched_arg);
+                       printf("Scheduler: Using instead: %s\n", kSchedTraditionalWithPsetRunqueueString);
+
+                       sched_current_dispatch = &sched_traditional_with_pset_runqueue_dispatch;
+                       _sched_enum = sched_enum_traditional_with_pset_runqueue;
+                       strlcpy(sched_string, kSchedTraditionalWithPsetRunqueueString, sizeof(sched_string));
+#else
                        panic("Unrecognized scheduler algorithm: %s", sched_arg);
+#endif
                }
+               kprintf("Scheduler: Runtime selection of %s\n", sched_string);
        } else {
-#if   defined(CONFIG_SCHED_TRADITIONAL)
+#if   defined(CONFIG_SCHED_MULTIQ)
+               sched_current_dispatch = &sched_multiq_dispatch;
+               _sched_enum = sched_enum_multiq;
+               strlcpy(sched_string, kSchedMultiQString, sizeof(sched_string));
+#elif defined(CONFIG_SCHED_TRADITIONAL)
                sched_current_dispatch = &sched_traditional_with_pset_runqueue_dispatch;
                _sched_enum = sched_enum_traditional_with_pset_runqueue;
                strlcpy(sched_string, kSchedTraditionalWithPsetRunqueueString, sizeof(sched_string));
-               kprintf("Scheduler: Default of %s\n", kSchedTraditionalWithPsetRunqueueString);
 #elif defined(CONFIG_SCHED_PROTO)
                sched_current_dispatch = &sched_proto_dispatch;
                _sched_enum = sched_enum_proto;
                strlcpy(sched_string, kSchedProtoString, sizeof(sched_string));
-               kprintf("Scheduler: Default of %s\n", kSchedProtoString);
 #elif defined(CONFIG_SCHED_GRRR)
                sched_current_dispatch = &sched_grrr_dispatch;
                _sched_enum = sched_enum_grrr;
                strlcpy(sched_string, kSchedGRRRString, sizeof(sched_string));
-               kprintf("Scheduler: Default of %s\n", kSchedGRRRString);
-#elif defined(CONFIG_SCHED_FIXEDPRIORITY)
-               sched_current_dispatch = &sched_fixedpriority_dispatch;
-               _sched_enum = sched_enum_fixedpriority;
-               strlcpy(sched_string, kSchedFixedPriorityString, sizeof(sched_string));
-               kprintf("Scheduler: Default of %s\n", kSchedFixedPriorityString);
 #else
 #error No default scheduler implementation
 #endif
+               kprintf("Scheduler: Default of %s\n", sched_string);
        }
        
        SCHED(init)();
@@ -608,9 +630,9 @@ sched_timebase_init(void)
        sched_realtime_timebase_init();
 }
 
-#if defined(CONFIG_SCHED_TRADITIONAL)
+#if defined(CONFIG_SCHED_TIMESHARE_CORE)
 
-static void
+void
 sched_traditional_init(void)
 {
        /*
@@ -634,7 +656,7 @@ sched_traditional_init(void)
        sched_tick = 0;
 }
 
-static void
+void
 sched_traditional_timebase_init(void)
 {
        uint64_t        abstime;
@@ -672,16 +694,27 @@ sched_traditional_timebase_init(void)
                abstime >>= 1;
        sched_fixed_shift = shift;
 
-       max_unsafe_computation = max_unsafe_quanta * std_quantum;
-       sched_safe_duration = 2 * max_unsafe_quanta * std_quantum;
+       max_unsafe_computation = ((uint64_t)max_unsafe_quanta) * std_quantum;
+       sched_safe_duration = 2 * ((uint64_t)max_unsafe_quanta) * std_quantum;
        
-       max_poll_computation = max_poll_quanta * std_quantum;
+       max_poll_computation = ((uint64_t)max_poll_quanta) * std_quantum;
        thread_depress_time = 1 * std_quantum;
        default_timeshare_computation = std_quantum / 2;
        default_timeshare_constraint = std_quantum;
 
+#if defined(CONFIG_TELEMETRY)
+       /* interval for high frequency telemetry */
+       clock_interval_to_absolutetime_interval(10, NSEC_PER_MSEC, &abstime);
+       assert((abstime >> 32) == 0 && (uint32_t)abstime != 0);
+       sched_telemetry_interval = (uint32_t)abstime;
+#endif
 }
 
+#endif /* CONFIG_SCHED_TIMESHARE_CORE */
+
+
+#if defined(CONFIG_SCHED_TRADITIONAL)
+
 static void
 sched_traditional_processor_init(processor_t processor)
 {
@@ -709,7 +742,7 @@ sched_traditional_with_pset_runqueue_init(void)
 
 #endif /* CONFIG_SCHED_TRADITIONAL */
 
-#if defined(CONFIG_SCHED_TRADITIONAL) || defined(CONFIG_SCHED_PROTO) || defined(CONFIG_SCHED_GRRR) || defined(CONFIG_SCHED_FIXEDPRIORITY)
+#if defined(CONFIG_SCHED_FAIRSHARE_CORE)
 void
 sched_traditional_fairshare_init(void)
 {
@@ -718,7 +751,7 @@ sched_traditional_fairshare_init(void)
        fs_runq.count = 0;
        queue_init(&fs_runq.queue);
 }
-#endif
+#endif /* CONFIG_SCHED_FAIRSHARE_CORE */
 
 static void
 sched_realtime_init(void)
@@ -747,7 +780,7 @@ sched_realtime_timebase_init(void)
 
 }
 
-#if defined(CONFIG_SCHED_TRADITIONAL)
+#if defined(CONFIG_SCHED_TIMESHARE_CORE)
 
 /*
  * Set up values for timeshare
@@ -814,7 +847,7 @@ preempt_pri_init(void)
                setbit(i, p);
 }
 
-#endif /* CONFIG_SCHED_TRADITIONAL */
+#endif /* CONFIG_SCHED_TIMESHARE_CORE */
 
 /*
  *     Thread wait timer expiration.
@@ -855,6 +888,7 @@ thread_unblock(
 {
        boolean_t               result = FALSE;
        thread_t                cthread = current_thread();
+       uint32_t                new_run_count;
 
        /*
         *      Set wait_result.
@@ -884,12 +918,12 @@ thread_unblock(
                /*
                 *      Update run counts.
                 */
-               sched_run_incr();
+               new_run_count = sched_run_incr(thread);
                if (thread->sched_mode == TH_MODE_TIMESHARE) {
-                       sched_share_incr();
+                       sched_share_incr(thread);
 
-                       if (thread->max_priority <= MAXPRI_THROTTLE)
-                               sched_background_incr();
+                       if (thread->sched_flags & TH_SFLAG_THROTTLED)
+                               sched_background_incr(thread);
                }
        }
        else {
@@ -907,6 +941,7 @@ thread_unblock(
                assert((thread->state & TH_IDLE) == 0);
 #endif
 
+               new_run_count = sched_run_count; /* updated in thread_select_idle() */
                result = TRUE;
        }
 
@@ -914,13 +949,16 @@ thread_unblock(
         * Calculate deadline for real-time threads.
         */
        if (thread->sched_mode == TH_MODE_REALTIME) {
-               thread->realtime.deadline = thread->realtime.constraint + mach_absolute_time();
+               uint64_t                ctime;
+
+               ctime = mach_absolute_time();
+               thread->realtime.deadline = thread->realtime.constraint + ctime;
        }
 
        /*
         * Clear old quantum, fail-safe computation, etc.
         */
-       thread->current_quantum = 0;
+       thread->quantum_remaining = 0;
        thread->computation_metered = 0;
        thread->reason = AST_NONE;
 
@@ -970,12 +1008,9 @@ thread_unblock(
                thread->callout_woke_thread = FALSE;
        }
 
-       /* Event should only be triggered if thread is not already running */
-       if (result == FALSE) {
-               KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
-                       MACHDBG_CODE(DBG_MACH_SCHED,MACH_MAKE_RUNNABLE) | DBG_FUNC_NONE,
-                       (uintptr_t)thread_tid(thread), thread->sched_pri, thread->wait_result, 0, 0);
-       }
+       KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
+               MACHDBG_CODE(DBG_MACH_SCHED,MACH_MAKE_RUNNABLE) | DBG_FUNC_NONE,
+               (uintptr_t)thread_tid(thread), thread->sched_pri, thread->wait_result, new_run_count, 0);
 
        DTRACE_SCHED2(wakeup, struct thread *, thread, struct proc *, thread->task->bsd_info);
 
@@ -1123,7 +1158,8 @@ assert_wait(
        register wait_queue_t   wq;
        register int            index;
 
-       assert(event != NO_EVENT);
+       if(event == NO_EVENT)
+               panic("assert_wait() called with NO_EVENT");
 
        KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
                MACHDBG_CODE(DBG_MACH_SCHED, MACH_WAIT)|DBG_FUNC_NONE,
@@ -1147,7 +1183,9 @@ assert_wait_timeout(
        uint64_t                        deadline;
        spl_t                           s;
 
-       assert(event != NO_EVENT);
+       if(event == NO_EVENT)
+               panic("assert_wait_timeout() called with NO_EVENT");
+
        wqueue = &wait_queues[wait_hash(event)];
 
        s = splsched();
@@ -1197,7 +1235,9 @@ assert_wait_timeout_with_leeway(
 
        clock_interval_to_absolutetime_interval(leeway, scale_factor, &slop);
 
-       assert(event != NO_EVENT);
+       if(event == NO_EVENT)
+               panic("assert_wait_timeout_with_leeway() called with NO_EVENT");
+
        wqueue = &wait_queues[wait_hash(event)];
 
        s = splsched();
@@ -1267,7 +1307,9 @@ assert_wait_deadline_with_leeway(
        wait_queue_t            wqueue;
        spl_t                           s;
 
-       assert(event != NO_EVENT);
+       if(event == NO_EVENT)
+               panic("assert_wait_deadline_with_leeway() called with NO_EVENT");
+
        wqueue = &wait_queues[wait_hash(event)];
 
        s = splsched();
@@ -1290,85 +1332,6 @@ assert_wait_deadline_with_leeway(
        return (wresult);
 }
 
-/*
- *     thread_sleep_fast_usimple_lock:
- *
- *     Cause the current thread to wait until the specified event
- *     occurs.  The specified simple_lock is unlocked before releasing
- *     the cpu and re-acquired as part of waking up.
- *
- *     This is the simple lock sleep interface for components that use a
- *     faster version of simple_lock() than is provided by usimple_lock().
- */
-__private_extern__ wait_result_t
-thread_sleep_fast_usimple_lock(
-       event_t                 event,
-       simple_lock_t           lock,
-       wait_interrupt_t        interruptible)
-{
-       wait_result_t res;
-
-       res = assert_wait(event, interruptible);
-       if (res == THREAD_WAITING) {
-               simple_unlock(lock);
-               res = thread_block(THREAD_CONTINUE_NULL);
-               simple_lock(lock);
-       }
-       return res;
-}
-
-
-/*
- *     thread_sleep_usimple_lock:
- *
- *     Cause the current thread to wait until the specified event
- *     occurs.  The specified usimple_lock is unlocked before releasing
- *     the cpu and re-acquired as part of waking up.
- *
- *     This is the simple lock sleep interface for components where
- *     simple_lock() is defined in terms of usimple_lock().
- */
-wait_result_t
-thread_sleep_usimple_lock(
-       event_t                 event,
-       usimple_lock_t          lock,
-       wait_interrupt_t        interruptible)
-{
-       wait_result_t res;
-
-       res = assert_wait(event, interruptible);
-       if (res == THREAD_WAITING) {
-               usimple_unlock(lock);
-               res = thread_block(THREAD_CONTINUE_NULL);
-               usimple_lock(lock);
-       }
-       return res;
-}
-
-/*
- *     thread_sleep_lock_write:
- *
- *     Cause the current thread to wait until the specified event
- *     occurs.  The specified (write) lock is unlocked before releasing
- *     the cpu. The (write) lock will be re-acquired before returning.
- */
-wait_result_t
-thread_sleep_lock_write(
-       event_t                 event,
-       lock_t                  *lock,
-       wait_interrupt_t interruptible)
-{
-       wait_result_t   res;
-
-       res = assert_wait(event, interruptible);
-       if (res == THREAD_WAITING) {
-               lock_write_done(lock);
-               res = thread_block(THREAD_CONTINUE_NULL);
-               lock_write(lock);
-       }
-       return res;
-}
-
 /*
  * thread_isoncpu:
  *
@@ -1378,7 +1341,7 @@ thread_sleep_lock_write(
  * thread state to be serialized in the thread PCB.
  * 
  * Thread locked, returns the same way. While locked, fields
- * like "state" and "runq" cannot change.
+ * like "state" cannot change. "runq" can change only from set to unset.
  */
 static inline boolean_t
 thread_isoncpu(thread_t thread)
@@ -1388,6 +1351,7 @@ thread_isoncpu(thread_t thread)
                return (FALSE);
 
        /* Waiting on a runqueue, not currently running */
+       /* TODO: This is invalid - it can get dequeued without thread lock, but not context switched. */
        if (thread->runq != PROCESSOR_NULL)
                return (FALSE);
 
@@ -1701,6 +1665,9 @@ thread_wakeup_prim_internal(
        register wait_queue_t   wq;
        register int                    index;
 
+       if(event == NO_EVENT)
+               panic("thread_wakeup_prim() called with NO_EVENT");
+
        index = wait_hash(event);
        wq = &wait_queues[index];
        if (one_thread)
@@ -1713,6 +1680,7 @@ thread_wakeup_prim_internal(
  *     thread_bind:
  *
  *     Force the current thread to execute on the specified processor.
+ *     Takes effect after the next thread_block().
  *
  *     Returns the previous binding.  PROCESSOR_NULL means
  *     not bound.
@@ -1730,6 +1698,9 @@ thread_bind(
        s = splsched();
        thread_lock(self);
 
+       /* <rdar://problem/15102234> */
+       assert(self->sched_pri < BASEPRI_RTQUEUES);
+
        prev = self->bound_processor;
        self->bound_processor = processor;
 
@@ -1739,6 +1710,66 @@ thread_bind(
        return (prev);
 }
 
+/* Invoked prior to idle entry to determine if, on SMT capable processors, an SMT
+ * rebalancing opportunity exists when a core is (instantaneously) idle, but
+ * other SMT-capable cores may be over-committed. TODO: some possible negatives:
+ * IPI thrash if this core does not remain idle following the load balancing ASTs
+ * Idle "thrash", when IPI issue is followed by idle entry/core power down
+ * followed by a wakeup shortly thereafter.
+ */
+
+/* Invoked with pset locked, returns with pset unlocked */
+#if (DEVELOPMENT || DEBUG)
+int sched_smt_balance = 1;
+#endif
+
+static void
+sched_SMT_balance(processor_t cprocessor, processor_set_t cpset) {
+       processor_t ast_processor = NULL;
+
+#if (DEVELOPMENT || DEBUG)
+       if (__improbable(sched_smt_balance == 0))
+               goto smt_balance_exit;
+#endif
+       
+       assert(cprocessor == current_processor());
+       if (cprocessor->is_SMT == FALSE)
+               goto smt_balance_exit;
+
+       processor_t sib_processor = cprocessor->processor_secondary ? cprocessor->processor_secondary : cprocessor->processor_primary;
+
+       /* Determine if both this processor and its sibling are idle,
+        * indicating an SMT rebalancing opportunity.
+        */
+       if (sib_processor->state != PROCESSOR_IDLE)
+               goto smt_balance_exit;
+
+       processor_t sprocessor;
+
+       sprocessor = (processor_t)queue_first(&cpset->active_queue);
+
+       while (!queue_end(&cpset->active_queue, (queue_entry_t)sprocessor)) {
+               if ((sprocessor->state == PROCESSOR_RUNNING) &&
+                   (sprocessor->processor_primary != sprocessor) &&
+                   (sprocessor->processor_primary->state == PROCESSOR_RUNNING) &&
+                   (sprocessor->current_pri < BASEPRI_RTQUEUES) &&
+                   ((cpset->pending_AST_cpu_mask & (1U << sprocessor->cpu_id)) == 0)) {
+                       assert(sprocessor != cprocessor);
+                       ast_processor = sprocessor;
+                       break;
+               }
+               sprocessor = (processor_t)queue_next((queue_entry_t)sprocessor);
+       }
+
+smt_balance_exit:
+       pset_unlock(cpset);
+
+       if (ast_processor) {
+               KERNEL_DEBUG_CONSTANT(MACHDBG_CODE(DBG_MACH_SCHED, MACH_SCHED_SMT_BALANCE), ast_processor->cpu_id, ast_processor->state, ast_processor->processor_primary->state, 0, 0);
+               cause_ast_check(ast_processor);
+       }
+}
+
 /*
  *     thread_select:
  *
@@ -1749,11 +1780,11 @@ thread_bind(
 static thread_t
 thread_select(
        thread_t                        thread,
-       processor_t                     processor)
+       processor_t                     processor,
+       ast_t                           reason)
 {
        processor_set_t         pset = processor->processor_set;
        thread_t                        new_thread = THREAD_NULL;
-       boolean_t                       inactive_state;
 
        assert(processor == current_processor());
 
@@ -1766,13 +1797,13 @@ thread_select(
                
                processor->current_pri = thread->sched_pri;
                processor->current_thmode = thread->sched_mode;
+               processor->current_sfi_class = thread->sfi_class;
 
                pset_lock(pset);
 
-               assert(pset->low_count);
-               assert(pset->low_pri);
+               assert(processor->state != PROCESSOR_OFF_LINE);
 
-               if (processor->processor_meta != PROCESSOR_META_NULL && processor->processor_meta->primary != processor) {
+               if (processor->processor_primary != processor) {
                        /*
                         * Should this secondary SMT processor attempt to find work? For pset runqueue systems,
                         * we should look for work only under the same conditions that choose_processor()
@@ -1781,13 +1812,11 @@ thread_select(
                         * An exception is that bound threads are dispatched to a processor without going through
                         * choose_processor(), so in those cases we should continue trying to dequeue work.
                         */
-                       if (!processor->runq_bound_count && !queue_empty(&pset->idle_queue) && !rt_runq.count) {
+                       if (!SCHED(processor_bound_count)(processor) && !queue_empty(&pset->idle_queue) && !rt_runq.count) {
                                goto idle;
                        }
                }
 
-               inactive_state = processor->state != PROCESSOR_SHUTDOWN && machine_processor_is_inactive(processor);
-
                simple_lock(&rt_lock);
 
                /*
@@ -1796,28 +1825,21 @@ thread_select(
                 *      bound to a different processor, nor be in the wrong
                 *      processor set.
                 */
-               if (    ((thread->state & ~TH_SUSP) == TH_RUN)                  &&
-                   (thread->sched_pri >= BASEPRI_RTQUEUES              ||
-                   processor->processor_meta == PROCESSOR_META_NULL ||
-                   processor->processor_meta->primary == processor) &&
-                   (thread->bound_processor == PROCESSOR_NULL  ||
-                   thread->bound_processor == processor)               &&
-                   (thread->affinity_set == AFFINITY_SET_NULL  ||
-                   thread->affinity_set->aset_pset == pset)) {
-                       if (thread->sched_pri >= BASEPRI_RTQUEUES &&
-                           first_timeslice(processor)) {
+               if (((thread->state & ~TH_SUSP) == TH_RUN) &&
+                   (thread->sched_pri >= BASEPRI_RTQUEUES     || processor->processor_primary == processor) &&
+                   (thread->bound_processor == PROCESSOR_NULL || thread->bound_processor == processor)      &&
+                   (thread->affinity_set == AFFINITY_SET_NULL || thread->affinity_set->aset_pset == pset)) {
+                       if (thread->sched_pri >= BASEPRI_RTQUEUES && first_timeslice(processor)) {
                                if (rt_runq.count > 0) {
-                                       register queue_t                q;
+                                       thread_t next_rt;
 
-                                       q = &rt_runq.queue;
-                                       if (((thread_t)q->next)->realtime.deadline <
-                                           processor->deadline) {
-                                               if ((((thread_t)q->next)->bound_processor == PROCESSOR_NULL) || (((thread_t)q->next)->bound_processor == processor)) {
-                                               thread = (thread_t)dequeue_head(q);
+                                       next_rt = (thread_t)queue_first(&rt_runq.queue);
+                                       if (next_rt->realtime.deadline < processor->deadline &&
+                                          (next_rt->bound_processor == PROCESSOR_NULL || next_rt->bound_processor == processor)) {
+                                               thread = (thread_t)dequeue_head(&rt_runq.queue);
                                                thread->runq = PROCESSOR_NULL;
                                                SCHED_STATS_RUNQ_CHANGE(&rt_runq.runq_stats, rt_runq.count);
                                                rt_runq.count--;
-                                               }
                                        }
                                }
 
@@ -1830,16 +1852,11 @@ thread_select(
                                return (thread);
                        }
 
-                       if (!inactive_state && (thread->sched_mode != TH_MODE_FAIRSHARE || SCHED(fairshare_runq_count)() == 0) && (rt_runq.count == 0 || BASEPRI_RTQUEUES < thread->sched_pri) &&
-                                       (new_thread = SCHED(choose_thread)(processor, thread->sched_mode == TH_MODE_FAIRSHARE ? MINPRI : thread->sched_pri)) == THREAD_NULL) {
+                       if ((thread->sched_mode != TH_MODE_FAIRSHARE || SCHED(fairshare_runq_count)() == 0) && (rt_runq.count == 0 || BASEPRI_RTQUEUES < thread->sched_pri) && (new_thread = SCHED(choose_thread)(processor, thread->sched_mode == TH_MODE_FAIRSHARE ? MINPRI : thread->sched_pri, reason)) == THREAD_NULL) {
 
                                simple_unlock(&rt_lock);
 
-                               /* I am the highest priority runnable (non-idle) thread */
-
-                               pset_pri_hint(pset, processor, processor->current_pri);
-
-                               pset_count_hint(pset, processor, SCHED(processor_runq_count)(processor));
+                               /* This thread is still the highest priority runnable (non-idle) thread */
 
                                processor->deadline = UINT64_MAX;
 
@@ -1851,15 +1868,9 @@ thread_select(
 
                if (new_thread != THREAD_NULL ||
                                (SCHED(processor_queue_has_priority)(processor, rt_runq.count == 0 ? IDLEPRI : BASEPRI_RTQUEUES, TRUE) &&
-                                        (new_thread = SCHED(choose_thread)(processor, MINPRI)) != THREAD_NULL)) {
+                                        (new_thread = SCHED(choose_thread)(processor, MINPRI, reason)) != THREAD_NULL)) {
                                simple_unlock(&rt_lock);
 
-                               if (!inactive_state) {
-                                       pset_pri_hint(pset, processor, new_thread->sched_pri);
-
-                                       pset_count_hint(pset, processor, SCHED(processor_runq_count)(processor));
-                               }
-
                                processor->deadline = UINT64_MAX;
                                pset_unlock(pset);
 
@@ -1867,9 +1878,11 @@ thread_select(
                }
 
                if (rt_runq.count > 0) {
-                       thread = (thread_t)dequeue_head(&rt_runq.queue);
+                       thread_t next_rt = (thread_t)queue_first(&rt_runq.queue);
+
+                       if (__probable((next_rt->bound_processor == NULL || (next_rt->bound_processor == processor)))) {
+                               thread = (thread_t)dequeue_head(&rt_runq.queue);
 
-                       if (__probable((thread->bound_processor == NULL || (thread->bound_processor == processor)))) {
                                thread->runq = PROCESSOR_NULL;
                                SCHED_STATS_RUNQ_CHANGE(&rt_runq.runq_stats, rt_runq.count);
                                rt_runq.count--;
@@ -1880,8 +1893,6 @@ thread_select(
                                pset_unlock(pset);
 
                                return (thread);
-                       } else {
-                               enqueue_head(&rt_runq.queue, (queue_entry_t)thread);
                        }
                }
 
@@ -1900,24 +1911,6 @@ thread_select(
                        
                processor->deadline = UINT64_MAX;
 
-               /*
-                *      Set processor inactive based on
-                *      indication from the platform code.
-                */
-               if (inactive_state) {
-                       if (processor->state == PROCESSOR_RUNNING)
-                               remqueue((queue_entry_t)processor);
-                       else
-                       if (processor->state == PROCESSOR_IDLE)
-                               remqueue((queue_entry_t)processor);
-
-                       processor->state = PROCESSOR_INACTIVE;
-
-                       pset_unlock(pset);
-
-                       return (processor->idle_thread);
-               }
-
                /*
                 *      No runnable threads, attempt to steal
                 *      from other processors.
@@ -1945,24 +1938,24 @@ thread_select(
                        remqueue((queue_entry_t)processor);
                        processor->state = PROCESSOR_IDLE;
 
-                       if (processor->processor_meta == PROCESSOR_META_NULL || processor->processor_meta->primary == processor) {
+                       if (processor->processor_primary == processor) {
                                enqueue_head(&pset->idle_queue, (queue_entry_t)processor);
-                               pset_pri_init_hint(pset, processor);
-                               pset_count_init_hint(pset, processor);
                        }
                        else {
-                               enqueue_head(&processor->processor_meta->idle_queue, (queue_entry_t)processor);
-                                       pset_unlock(pset);
-                                       return (processor->idle_thread);
+                               enqueue_head(&pset->idle_secondary_queue, (queue_entry_t)processor);
                        }
                }
 
-               pset_unlock(pset);
+               /* Invoked with pset locked, returns with pset unlocked */
+               sched_SMT_balance(processor, pset);
 
 #if CONFIG_SCHED_IDLE_IN_PLACE
                /*
                 *      Choose idle thread if fast idle is not possible.
                 */
+               if (processor->processor_primary != processor)
+                       return (processor->idle_thread);
+
                if ((thread->state & (TH_IDLE|TH_TERMINATE|TH_SUSP)) || !(thread->state & TH_WAIT) || thread->wake_active || thread->sched_pri >= BASEPRI_RTQUEUES)
                        return (processor->idle_thread);
 
@@ -2007,16 +2000,17 @@ thread_select_idle(
        int                     urgency;
 
        if (thread->sched_mode == TH_MODE_TIMESHARE) {
-               if (thread->max_priority <= MAXPRI_THROTTLE)
-                       sched_background_decr();
+               if (thread->sched_flags & TH_SFLAG_THROTTLED)
+                       sched_background_decr(thread);
 
-               sched_share_decr();
+               sched_share_decr(thread);
        }
-       sched_run_decr();
+       sched_run_decr(thread);
 
        thread->state |= TH_IDLE;
        processor->current_pri = IDLEPRI;
        processor->current_thmode = TH_MODE_NONE;
+       processor->current_sfi_class = SFI_CLASS_KERNEL;
 
        /* Reload precise timing global policy to thread-local policy */
        thread->precise_user_kernel_time = use_precise_user_kernel_time(thread);
@@ -2027,6 +2021,11 @@ thread_select_idle(
         *      Switch execution timing to processor idle thread.
         */
        processor->last_dispatch = mach_absolute_time();
+
+#ifdef CONFIG_MACH_APPROXIMATE_TIME
+       commpage_update_mach_approximate_time(processor->last_dispatch);
+#endif
+
        thread->last_run_time = processor->last_dispatch;
        thread_timer_event(processor->last_dispatch, &processor->idle_thread->system_timer);
        PROCESSOR_DATA(processor, kernel_timer) = &processor->idle_thread->system_timer;
@@ -2064,10 +2063,8 @@ thread_select_idle(
                PROCESSOR_DATA(processor, kernel_timer) = &thread->system_timer;
 
                thread_quantum_init(thread);
-               thread->last_quantum_refill_time = processor->last_dispatch;
-
-               processor->quantum_end = processor->last_dispatch + thread->current_quantum;
-               timer_call_enter1(&processor->quantum_timer, thread, processor->quantum_end, TIMER_CALL_SYS_CRITICAL);
+               processor->quantum_end = processor->last_dispatch + thread->quantum_remaining;
+               timer_call_enter1(&processor->quantum_timer, thread, processor->quantum_end, TIMER_CALL_SYS_CRITICAL | TIMER_CALL_LOCAL);
                processor->timeslice = 1;
 
                thread->computation_epoch = processor->last_dispatch;
@@ -2086,12 +2083,12 @@ thread_select_idle(
 
        thread_tell_urgency(urgency, arg1, arg2, new_thread);
 
-       sched_run_incr();
+       sched_run_incr(thread);
        if (thread->sched_mode == TH_MODE_TIMESHARE) {
-               sched_share_incr();
+               sched_share_incr(thread);
 
-               if (thread->max_priority <= MAXPRI_THROTTLE)
-                       sched_background_incr();
+               if (thread->sched_flags & TH_SFLAG_THROTTLED)
+                       sched_background_incr(thread);
        }
 
        return (new_thread);
@@ -2101,12 +2098,13 @@ thread_select_idle(
 #if defined(CONFIG_SCHED_TRADITIONAL) 
 static thread_t
 sched_traditional_choose_thread(
-                                                               processor_t             processor,
-                                                               int                             priority)
+                                processor_t     processor,
+                                int             priority,
+                       __unused ast_t           reason)
 {
        thread_t thread;
        
-       thread = choose_thread(processor, runq_for_processor(processor), priority);
+       thread = choose_thread_from_runq(processor, runq_for_processor(processor), priority);
        if (thread != THREAD_NULL) {
                runq_consider_decr_bound_count(processor, thread);
        }
@@ -2116,10 +2114,10 @@ sched_traditional_choose_thread(
 
 #endif /* defined(CONFIG_SCHED_TRADITIONAL)  */
 
-#if defined(CONFIG_SCHED_TRADITIONAL) || defined(CONFIG_SCHED_FIXEDPRIORITY)
+#if defined(CONFIG_SCHED_TRADITIONAL)
 
 /*
- *     choose_thread:
+ *     choose_thread_from_runq:
  *
  *     Locate a thread to execute from the processor run queue
  *     and return it.  Only choose a thread with greater or equal
@@ -2129,7 +2127,7 @@ sched_traditional_choose_thread(
  *     on failure.
  */
 thread_t
-choose_thread(
+choose_thread_from_runq(
        processor_t             processor,
        run_queue_t             rq,
        int                             priority)
@@ -2170,7 +2168,7 @@ choose_thread(
        return (THREAD_NULL);
 }
 
-#endif /* defined(CONFIG_SCHED_TRADITIONAL) || defined(CONFIG_SCHED_FIXEDPRIORITY) */
+#endif /* defined(CONFIG_SCHED_TRADITIONAL) */
 
 /*
  *     Perform a context switch and start executing the new thread.
@@ -2180,32 +2178,6 @@ choose_thread(
  *     Called at splsched.
  */
 
-#define funnel_release_check(thread, debug)                            \
-MACRO_BEGIN                                                                                            \
-       if ((thread)->funnel_state & TH_FN_OWNED) {                     \
-               (thread)->funnel_state = TH_FN_REFUNNEL;                \
-               KERNEL_DEBUG(0x603242c | DBG_FUNC_NONE,                 \
-                       (thread)->funnel_lock, (debug), 0, 0, 0);       \
-               funnel_unlock((thread)->funnel_lock);                   \
-       }                                                                                                       \
-MACRO_END
-
-#define funnel_refunnel_check(thread, debug)                           \
-MACRO_BEGIN                                                                                                    \
-       if ((thread)->funnel_state & TH_FN_REFUNNEL) {                  \
-               kern_return_t   result = (thread)->wait_result;         \
-                                                                                                                       \
-               (thread)->funnel_state = 0;                                                     \
-               KERNEL_DEBUG(0x6032428 | DBG_FUNC_NONE,                         \
-                       (thread)->funnel_lock, (debug), 0, 0, 0);               \
-               funnel_lock((thread)->funnel_lock);                                     \
-               KERNEL_DEBUG(0x6032430 | DBG_FUNC_NONE,                         \
-                       (thread)->funnel_lock, (debug), 0, 0, 0);               \
-               (thread)->funnel_state = TH_FN_OWNED;                           \
-               (thread)->wait_result = result;                                         \
-       }                                                                                                               \
-MACRO_END
-
 /*
  * thread_invoke
  *
@@ -2224,6 +2196,10 @@ thread_invoke(
        processor_t             processor;
        uint64_t                ctime = mach_absolute_time();
 
+#ifdef CONFIG_MACH_APPROXIMATE_TIME
+       commpage_update_mach_approximate_time(ctime);
+#endif
+
        if (__improbable(get_preemption_level() != 0)) {
                int pl = get_preemption_level();
                panic("thread_invoke: preemption_level %d, possible cause: %s",
@@ -2232,10 +2208,11 @@ thread_invoke(
        }
 
        assert(self == current_thread());
+       assert(self->runq == PROCESSOR_NULL);
 
-#if defined(CONFIG_SCHED_TRADITIONAL)
+#if defined(CONFIG_SCHED_TIMESHARE_CORE)
        sched_traditional_consider_maintenance(ctime);
-#endif /* CONFIG_SCHED_TRADITIONAL */  
+#endif /* CONFIG_SCHED_TIMESHARE_CORE */       
        
        /*
         * Mark thread interruptible.
@@ -2243,13 +2220,16 @@ thread_invoke(
        thread_lock(thread);
        thread->state &= ~TH_UNINT;
 
-#if DEBUG
        assert(thread_runnable(thread));
-#endif
+       assert(thread->bound_processor == PROCESSOR_NULL || thread->bound_processor == current_processor());
+       assert(thread->runq == PROCESSOR_NULL);
 
        /* Reload precise timing global policy to thread-local policy */
        thread->precise_user_kernel_time = use_precise_user_kernel_time(thread);
        
+       /* Update SFI class based on other factors */
+       thread->sfi_class = sfi_thread_classify(thread);
+
        /*
         * Allow time constraint threads to hang onto
         * a stack.
@@ -2277,6 +2257,7 @@ thread_invoke(
                        processor->active_thread = thread;
                        processor->current_pri = thread->sched_pri;
                        processor->current_thmode = thread->sched_mode;
+                       processor->current_sfi_class = thread->sfi_class;
                        if (thread->last_processor != processor && thread->last_processor != NULL) {
                                if (thread->last_processor->processor_set != processor->processor_set)
                                        thread->ps_switch++;
@@ -2328,7 +2309,6 @@ thread_invoke(
 
                        counter(c_thread_invoke_hits++);
 
-                       funnel_refunnel_check(thread, 2);
                        (void) spllo();
 
                        assert(continuation);
@@ -2347,7 +2327,6 @@ thread_invoke(
 
                        self->continuation = self->parameter = NULL;
 
-                       funnel_refunnel_check(self, 3);
                        (void) spllo();
 
                        call_continuation(continuation, parameter, self->wait_result);
@@ -2387,6 +2366,7 @@ need_stack:
        processor->active_thread = thread;
        processor->current_pri = thread->sched_pri;
        processor->current_thmode = thread->sched_mode;
+       processor->current_sfi_class = thread->sfi_class;
        if (thread->last_processor != processor && thread->last_processor != NULL) {
                if (thread->last_processor->processor_set != processor->processor_set)
                        thread->ps_switch++;
@@ -2417,7 +2397,6 @@ need_stack:
                                         PROCESSOR_DATA(processor, current_state));
        }
        
-
        KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
                MACHDBG_CODE(DBG_MACH_SCHED,MACH_SCHED) | DBG_FUNC_NONE,
                self->reason, (uintptr_t)thread_tid(thread), self->sched_pri, thread->sched_pri, 0);
@@ -2451,7 +2430,6 @@ need_stack:
        if (continuation) {
                self->continuation = self->parameter = NULL;
 
-               funnel_refunnel_check(self, 3);
                (void) spllo();
 
                call_continuation(continuation, parameter, self->wait_result);
@@ -2497,7 +2475,7 @@ thread_dispatch(
                                remainder = processor->quantum_end -
                                    processor->last_dispatch;
 
-                       consumed = thread->current_quantum - remainder;
+                       consumed = thread->quantum_remaining - remainder;
 
                        if ((thread->reason & AST_LEDGER) == 0) {
                                /*
@@ -2508,6 +2486,15 @@ thread_dispatch(
                                    task_ledgers.cpu_time, consumed);
                                ledger_credit(thread->t_threadledger,
                                    thread_ledgers.cpu_time, consumed);
+#ifdef CONFIG_BANK
+                               if (thread->t_bankledger) {
+                                       ledger_credit(thread->t_bankledger,
+                                               bank_ledgers.cpu_time,
+                                               (consumed - thread->t_deduct_bank_ledger_time));
+
+                               }
+                               thread->t_deduct_bank_ledger_time =0;
+#endif
                        }
 
                        wake_lock(thread);
@@ -2518,18 +2505,17 @@ thread_dispatch(
                         */
                        if (first_timeslice(processor) &&
                            processor->quantum_end > processor->last_dispatch)
-                               thread->current_quantum = (uint32_t)remainder;
+                               thread->quantum_remaining = (uint32_t)remainder;
                        else
-                               thread->current_quantum = 0;
+                               thread->quantum_remaining = 0;
 
                        if (thread->sched_mode == TH_MODE_REALTIME) {
                                /*
                                 *      Cancel the deadline if the thread has
                                 *      consumed the entire quantum.
                                 */
-                               if (thread->current_quantum == 0) {
+                               if (thread->quantum_remaining == 0) {
                                        thread->realtime.deadline = UINT64_MAX;
-                                       thread->reason |= AST_QUANTUM;
                                }
                        } else {
 #if defined(CONFIG_SCHED_TRADITIONAL)
@@ -2538,9 +2524,9 @@ thread_dispatch(
                                 *      remaining quantum as an expired quantum
                                 *      but include what's left next time.
                                 */
-                               if (thread->current_quantum < min_std_quantum) {
+                               if (thread->quantum_remaining < min_std_quantum) {
                                        thread->reason |= AST_QUANTUM;
-                                       thread->current_quantum += SCHED(initial_quantum_size)(thread);
+                                       thread->quantum_remaining += SCHED(initial_quantum_size)(thread);
                                }
 #endif
                        }
@@ -2550,9 +2536,23 @@ thread_dispatch(
                         *      take the remainder of the quantum.
                         */
                        if ((thread->reason & (AST_HANDOFF|AST_QUANTUM)) == AST_HANDOFF) {
-                               self->current_quantum = thread->current_quantum;
+                               self->quantum_remaining = thread->quantum_remaining;
                                thread->reason |= AST_QUANTUM;
-                               thread->current_quantum = 0;
+                               thread->quantum_remaining = 0;
+                       } else {
+#if defined(CONFIG_SCHED_MULTIQ)
+                               if (sched_groups_enabled && thread->sched_group == self->sched_group) {
+                                       /* TODO: Remove tracepoint */
+                                       KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
+                                           MACHDBG_CODE(DBG_MACH_SCHED, MACH_QUANTUM_HANDOFF) | DBG_FUNC_NONE,
+                                           self->reason, (uintptr_t)thread_tid(thread),
+                                           self->quantum_remaining, thread->quantum_remaining, 0);
+
+                                       self->quantum_remaining = thread->quantum_remaining;
+                                       thread->quantum_remaining = 0;
+                                       /*  TODO: Should we set AST_QUANTUM here? */
+                               }
+#endif /* defined(CONFIG_SCHED_MULTIQ) */
                        }
 
                        thread->computation_metered += (processor->last_dispatch - thread->computation_epoch);
@@ -2593,6 +2593,10 @@ thread_dispatch(
 
                                thread->reason = AST_NONE;
 
+                               KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
+                                       MACHDBG_CODE(DBG_MACH_SCHED,MACH_DISPATCH) | DBG_FUNC_NONE,
+                                       (uintptr_t)thread_tid(thread), thread->reason, thread->state, sched_run_count, 0);
+                               
                                if (thread->wake_active) {
                                        thread->wake_active = FALSE;
                                        thread_unlock(thread);
@@ -2609,6 +2613,7 @@ thread_dispatch(
                                 *      Waiting.
                                 */
                                boolean_t should_terminate = FALSE;
+                               uint32_t new_run_count;
 
                                /* Only the first call to thread_dispatch
                                 * after explicit termination should add
@@ -2623,12 +2628,22 @@ thread_dispatch(
                                thread->chosen_processor = PROCESSOR_NULL;
 
                                if (thread->sched_mode == TH_MODE_TIMESHARE) {
-                                       if (thread->max_priority <= MAXPRI_THROTTLE)
-                                               sched_background_decr();
+                                       if (thread->sched_flags & TH_SFLAG_THROTTLED)
+                                               sched_background_decr(thread);
+
+                                       sched_share_decr(thread);
+                               }
+                               new_run_count = sched_run_decr(thread);
 
-                                       sched_share_decr();
+                               if ((thread->state & (TH_WAIT | TH_TERMINATE)) == TH_WAIT) {
+                                       if (thread->reason & AST_SFI) {
+                                               thread->wait_sfi_begin_time = processor->last_dispatch;
+                                       }
                                }
-                               sched_run_decr();
+
+                               KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
+                                       MACHDBG_CODE(DBG_MACH_SCHED,MACH_DISPATCH) | DBG_FUNC_NONE,
+                                       (uintptr_t)thread_tid(thread), thread->reason, thread->state, new_run_count, 0);
 
                                (*thread->sched_call)(SCHED_CALL_BLOCK, thread);
 
@@ -2652,6 +2667,15 @@ thread_dispatch(
        if (!(self->state & TH_IDLE)) {
                uint64_t        arg1, arg2;
                int             urgency;
+               ast_t                   new_ast;
+
+               thread_lock(self);
+               new_ast = sfi_thread_needs_ast(self, NULL);
+               thread_unlock(self);
+
+               if (new_ast != AST_NONE) {
+                       ast_on(new_ast);
+               }
 
                urgency = thread_get_urgency(self, &arg1, &arg2);
 
@@ -2660,16 +2684,15 @@ thread_dispatch(
                /*
                 *      Get a new quantum if none remaining.
                 */
-               if (self->current_quantum == 0) {
+               if (self->quantum_remaining == 0) {
                        thread_quantum_init(self);
-                       self->last_quantum_refill_time = processor->last_dispatch;
                }
 
                /*
                 *      Set up quantum timer and timeslice.
                 */
-               processor->quantum_end = (processor->last_dispatch + self->current_quantum);
-               timer_call_enter1(&processor->quantum_timer, self, processor->quantum_end, TIMER_CALL_SYS_CRITICAL);
+               processor->quantum_end = processor->last_dispatch + self->quantum_remaining;
+               timer_call_enter1(&processor->quantum_timer, self, processor->quantum_end, TIMER_CALL_SYS_CRITICAL | TIMER_CALL_LOCAL);
 
                processor->timeslice = 1;
 
@@ -2683,11 +2706,6 @@ thread_dispatch(
        }
 }
 
-#include <libkern/OSDebug.h>
-
-uint32_t       kdebug_thread_block = 0;
-
-
 /*
  *     thread_block_reason:
  *
@@ -2716,9 +2734,6 @@ thread_block_reason(
 
        s = splsched();
 
-       if (!(reason & AST_PREEMPT))
-               funnel_release_check(self, 2);
-
        processor = current_processor();
 
        /* If we're explicitly yielding, force a subsequent quantum */
@@ -2731,7 +2746,7 @@ thread_block_reason(
        self->continuation = continuation;
        self->parameter = parameter;
 
-       if (__improbable(kdebug_thread_block && kdebug_enable && self->state != TH_RUN)) {              
+       if (self->state & ~(TH_RUN | TH_IDLE)) {
                KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
                        MACHDBG_CODE(DBG_MACH_SCHED,MACH_BLOCK), 
                        reason, VM_KERNEL_UNSLIDE(continuation), 0, 0, 0);
@@ -2739,11 +2754,10 @@ thread_block_reason(
 
        do {
                thread_lock(self);
-               new_thread = thread_select(self, processor);
+               new_thread = thread_select(self, processor, reason);
                thread_unlock(self);
        } while (!thread_invoke(self, new_thread, reason));
 
-       funnel_refunnel_check(self, 5);
        splx(s);
 
        return (self->wait_result);
@@ -2788,8 +2802,6 @@ thread_run(
 {
        ast_t           handoff = AST_HANDOFF;
 
-       funnel_release_check(self, 3);
-
        self->continuation = continuation;
        self->parameter = parameter;
 
@@ -2797,13 +2809,11 @@ thread_run(
                processor_t             processor = current_processor();
 
                thread_lock(self);
-               new_thread = thread_select(self, processor);
+               new_thread = thread_select(self, processor, AST_NONE);
                thread_unlock(self);
                handoff = AST_NONE;
        }
 
-       funnel_refunnel_check(self, 6);
-
        return (self->wait_result);
 }
 
@@ -2830,8 +2840,6 @@ thread_continue(
 
        self->continuation = self->parameter = NULL;
 
-       funnel_refunnel_check(self, 4);
-
        if (thread != THREAD_NULL)
                (void)spllo();
 
@@ -2844,22 +2852,27 @@ void
 thread_quantum_init(thread_t thread)
 {
        if (thread->sched_mode == TH_MODE_REALTIME) {
-               thread->current_quantum = thread->realtime.computation;
+               thread->quantum_remaining = thread->realtime.computation;
        } else {
-               thread->current_quantum = SCHED(initial_quantum_size)(thread);
+               thread->quantum_remaining = SCHED(initial_quantum_size)(thread);
        }
 }
 
-#if defined(CONFIG_SCHED_TRADITIONAL)
-static uint32_t
+#if defined(CONFIG_SCHED_TIMESHARE_CORE)
+
+uint32_t
 sched_traditional_initial_quantum_size(thread_t thread)
 {
-       if ((thread == THREAD_NULL) || thread->priority > MAXPRI_THROTTLE)
+       if ((thread == THREAD_NULL) || !(thread->sched_flags & TH_SFLAG_THROTTLED))
                return std_quantum;
        else
                return bg_quantum;
 }
 
+#endif /* CONFIG_SCHED_TIMESHARE_CORE */
+
+#if defined(CONFIG_SCHED_TRADITIONAL)
+
 static sched_mode_t
 sched_traditional_initial_thread_sched_mode(task_t parent_task)
 {
@@ -2869,12 +2882,6 @@ sched_traditional_initial_thread_sched_mode(task_t parent_task)
                return TH_MODE_TIMESHARE;
 }
 
-static boolean_t
-sched_traditional_supports_timeshare_mode(void)
-{
-       return TRUE;
-}
-
 #endif /* CONFIG_SCHED_TRADITIONAL */
 
 /*
@@ -2897,7 +2904,7 @@ run_queue_init(
                queue_init(&rq->queues[i]);
 }
 
-#if defined(CONFIG_SCHED_TRADITIONAL) || defined(CONFIG_SCHED_PROTO) || defined(CONFIG_SCHED_GRRR) || defined(CONFIG_SCHED_FIXEDPRIORITY)
+#if defined(CONFIG_SCHED_FAIRSHARE_CORE)
 int
 sched_traditional_fairshare_runq_count(void)
 {
@@ -2976,7 +2983,7 @@ sched_traditional_fairshare_queue_remove(thread_t thread)
        }       
 }
 
-#endif /* defined(CONFIG_SCHED_TRADITIONAL) || defined(CONFIG_SCHED_PROTO) || defined(CONFIG_SCHED_GRRR) || defined(CONFIG_SCHED_FIXEDPRIORITY) */
+#endif /* CONFIG_SCHED_FAIRSHARE_CORE */
 
 /*
  *     run_queue_dequeue:
@@ -3042,13 +3049,12 @@ run_queue_enqueue(
                        rq->highq = thread->sched_pri;
                        result = TRUE;
                }
-       }
-       else
+       } else {
                if (options & SCHED_TAILQ)
                        enqueue_tail(queue, (queue_entry_t)thread);
                else
                        enqueue_head(queue, (queue_entry_t)thread);
-       
+       }
        if (SCHED(priority_is_urgent)(thread->sched_pri))
                rq->urgency++;
        SCHED_STATS_RUNQ_CHANGE(&rq->runq_stats, rq->count);
@@ -3107,10 +3113,11 @@ fairshare_setrun(
 
        SCHED(fairshare_enqueue)(thread);
        
+       pset_unlock(pset);
+
        if (processor != current_processor())
                machine_signal_idle(processor);
 
-       pset_unlock(pset);
 
 }
 
@@ -3177,8 +3184,13 @@ realtime_setrun(
        processor_set_t         pset = processor->processor_set;
        ast_t                           preempt;
 
+       boolean_t do_signal_idle = FALSE, do_cause_ast = FALSE;
+
        thread->chosen_processor = processor;
 
+       /* <rdar://problem/15102234> */
+       assert(thread->bound_processor == PROCESSOR_NULL);
+
        /*
         *      Dispatch directly onto idle processor.
         */
@@ -3190,6 +3202,7 @@ realtime_setrun(
                processor->next_thread = thread;
                processor->current_pri = thread->sched_pri;
                processor->current_thmode = thread->sched_mode;
+               processor->current_sfi_class = thread->sfi_class;
                processor->deadline = thread->realtime.deadline;
                processor->state = PROCESSOR_DISPATCHING;
 
@@ -3197,11 +3210,14 @@ realtime_setrun(
                        if (!(pset->pending_AST_cpu_mask & (1U << processor->cpu_id))) {
                                /* cleared on exit from main processor_idle() loop */
                                pset->pending_AST_cpu_mask |= (1U << processor->cpu_id);
-                               machine_signal_idle(processor);
+                               do_signal_idle = TRUE;
                        }
                }
-
                pset_unlock(pset);
+
+               if (do_signal_idle) {
+                       machine_signal_idle(processor);
+               }
                return;
        }
 
@@ -3221,6 +3237,7 @@ realtime_setrun(
                        processor->next_thread = THREAD_NULL;
                        processor->current_pri = thread->sched_pri;
                        processor->current_thmode = thread->sched_mode;
+                       processor->current_sfi_class = thread->sfi_class;
                        processor->deadline = thread->realtime.deadline;
                        processor->state = PROCESSOR_DISPATCHING;
                        if (processor == current_processor()) {
@@ -3229,13 +3246,14 @@ realtime_setrun(
                                if (!(pset->pending_AST_cpu_mask & (1U << processor->cpu_id))) {
                                        /* cleared on exit from main processor_idle() loop */
                                        pset->pending_AST_cpu_mask |= (1U << processor->cpu_id);
-                                       machine_signal_idle(processor);
+                                       do_signal_idle = TRUE;
                                }
-                       }                               
+                       }
                } else if (processor->state == PROCESSOR_DISPATCHING) {
                        if ((processor->next_thread == THREAD_NULL) && ((processor->current_pri < thread->sched_pri) || (processor->deadline > thread->realtime.deadline))) {
                                processor->current_pri = thread->sched_pri;
                                processor->current_thmode = thread->sched_mode;
+                               processor->current_sfi_class = thread->sfi_class;
                                processor->deadline = thread->realtime.deadline;
                        }
                } else {
@@ -3245,7 +3263,7 @@ realtime_setrun(
                                if (!(pset->pending_AST_cpu_mask & (1U << processor->cpu_id))) {
                                        /* cleared after IPI causes csw_check() to be called */
                                        pset->pending_AST_cpu_mask |= (1U << processor->cpu_id);
-                                       cause_ast_check(processor);
+                                       do_cause_ast = TRUE;
                                }
                        }
                }
@@ -3254,16 +3272,26 @@ realtime_setrun(
        }
 
        pset_unlock(pset);
+
+       if (do_signal_idle) {
+               machine_signal_idle(processor);
+       } else if (do_cause_ast) {
+               cause_ast_check(processor);
+       }
 }
 
-#if defined(CONFIG_SCHED_TRADITIONAL)
 
-static boolean_t
+#if defined(CONFIG_SCHED_TIMESHARE_CORE)
+
+boolean_t
 priority_is_urgent(int priority)
 {
        return testbit(priority, sched_preempt_pri) ? TRUE : FALSE;
 }
 
+#endif /* CONFIG_SCHED_TIMESHARE_CORE */
+
+#if defined(CONFIG_SCHED_TRADITIONAL)
 /*
  *     processor_enqueue:
  *
@@ -3313,6 +3341,8 @@ processor_setrun(
        ast_t                           preempt;
        enum { eExitIdle, eInterruptRunning, eDoNothing } ipi_action = eDoNothing;
 
+       boolean_t do_signal_idle = FALSE, do_cause_ast = FALSE;
+
        thread->chosen_processor = processor;
 
        /*
@@ -3327,16 +3357,21 @@ processor_setrun(
                processor->next_thread = thread;
                processor->current_pri = thread->sched_pri;
                processor->current_thmode = thread->sched_mode;
+               processor->current_sfi_class = thread->sfi_class;
                processor->deadline = UINT64_MAX;
                processor->state = PROCESSOR_DISPATCHING;
 
                if (!(pset->pending_AST_cpu_mask & (1U << processor->cpu_id))) {
                        /* cleared on exit from main processor_idle() loop */
                        pset->pending_AST_cpu_mask |= (1U << processor->cpu_id);
-                       machine_signal_idle(processor);
+                       do_signal_idle = TRUE;
                }
 
                pset_unlock(pset);
+               if (do_signal_idle) {
+                       machine_signal_idle(processor);
+               }
+
                return;
        }
 
@@ -3365,6 +3400,7 @@ processor_setrun(
                        processor->next_thread = THREAD_NULL;
                        processor->current_pri = thread->sched_pri;
                        processor->current_thmode = thread->sched_mode;
+                       processor->current_sfi_class = thread->sfi_class;
                        processor->deadline = UINT64_MAX;
                        processor->state = PROCESSOR_DISPATCHING;
 
@@ -3373,6 +3409,7 @@ processor_setrun(
                        if ((processor->next_thread == THREAD_NULL) && (processor->current_pri < thread->sched_pri)) {
                                processor->current_pri = thread->sched_pri;
                                processor->current_thmode = thread->sched_mode;
+                               processor->current_sfi_class = thread->sfi_class;
                                processor->deadline = UINT64_MAX;
                        }
                } else if (     (processor->state == PROCESSOR_RUNNING          ||
@@ -3396,6 +3433,7 @@ processor_setrun(
                        processor->next_thread = THREAD_NULL;
                        processor->current_pri = thread->sched_pri;
                        processor->current_thmode = thread->sched_mode;
+                       processor->current_sfi_class = thread->sfi_class;
                        processor->deadline = UINT64_MAX;
                        processor->state = PROCESSOR_DISPATCHING;
 
@@ -3408,31 +3446,37 @@ processor_setrun(
                        break;
                case eExitIdle:
                        if (processor == current_processor()) {
-                               if (csw_check_locked(processor, pset) != AST_NONE)
+                               if (csw_check_locked(processor, pset, AST_NONE) != AST_NONE)
                                        ast_on(preempt);
                        } else {
                                if (!(pset->pending_AST_cpu_mask & (1U << processor->cpu_id))) {
                                        /* cleared on exit from main processor_idle() loop */
                                        pset->pending_AST_cpu_mask |= (1U << processor->cpu_id);
-                                       machine_signal_idle(processor);
+                                       do_signal_idle = TRUE;
                                }
                        }
                        break;
                case eInterruptRunning:
                        if (processor == current_processor()) {
-                               if (csw_check_locked(processor, pset) != AST_NONE)
+                               if (csw_check_locked(processor, pset, AST_NONE) != AST_NONE)
                                        ast_on(preempt);
                        } else {
                                if (!(pset->pending_AST_cpu_mask & (1U << processor->cpu_id))) {
                                        /* cleared after IPI causes csw_check() to be called */
                                        pset->pending_AST_cpu_mask |= (1U << processor->cpu_id);
-                                       cause_ast_check(processor);
+                                       do_cause_ast = TRUE;
                                }
                        }
                        break;
        }
 
        pset_unlock(pset);
+
+       if (do_signal_idle) {
+               machine_signal_idle(processor);
+       } else if (do_cause_ast) {
+               cause_ast_check(processor);
+       }
 }
 
 #if defined(CONFIG_SCHED_TRADITIONAL)
@@ -3510,8 +3554,7 @@ static boolean_t
 should_current_thread_rechoose_processor(processor_t                   processor)
 {
        return (processor->current_pri < BASEPRI_RTQUEUES
-                       && processor->processor_meta != PROCESSOR_META_NULL
-                       && processor->processor_meta->primary != processor);
+                       && processor->processor_primary != processor);
 }
 
 static int
@@ -3520,7 +3563,6 @@ sched_traditional_processor_runq_count(processor_t   processor)
        return runq_for_processor(processor)->count;
 }
 
-
 static uint64_t
 sched_traditional_processor_runq_stats_count_sum(processor_t   processor)
 {
@@ -3536,9 +3578,13 @@ sched_traditional_with_pset_runqueue_processor_runq_stats_count_sum(processor_t
                return 0ULL;
 }
 
-#endif /* CONFIG_SCHED_TRADITIONAL */
+static int
+sched_traditional_processor_bound_count(processor_t   processor)
+{
+       return processor->runq_bound_count;
+}
 
-#define next_pset(p)   (((p)->pset_list != PROCESSOR_SET_NULL)? (p)->pset_list: (p)->node->psets)
+#endif /* CONFIG_SCHED_TRADITIONAL */
 
 /*
  *     choose_next_pset:
@@ -3581,147 +3627,228 @@ choose_processor(
        thread_t                        thread)
 {
        processor_set_t         nset, cset = pset;
-       processor_meta_t        pmeta = PROCESSOR_META_NULL;
-       processor_t             mprocessor;
        
        /*
-        *      Prefer the hinted processor, when appropriate.
+        * Prefer the hinted processor, when appropriate.
         */
 
+       /* Fold last processor hint from secondary processor to its primary */
        if (processor != PROCESSOR_NULL) {
-               if (processor->processor_meta != PROCESSOR_META_NULL)
-                       processor = processor->processor_meta->primary;
+               processor = processor->processor_primary;
        }
 
-       mprocessor = machine_choose_processor(pset, processor);
-       if (mprocessor != PROCESSOR_NULL)
-               processor = mprocessor;
+       /*
+        * Only consult platform layer if pset is active, which
+        * it may not be in some cases when a multi-set system
+        * is going to sleep.
+        */
+       if (pset->online_processor_count) {
+               if ((processor == PROCESSOR_NULL) || (processor->processor_set == pset && processor->state == PROCESSOR_IDLE)) {
+                       processor_t mc_processor = machine_choose_processor(pset, processor);
+                       if (mc_processor != PROCESSOR_NULL)
+                               processor = mc_processor->processor_primary;
+               }
+       }
 
+       /*
+        * At this point, we may have a processor hint, and we may have
+        * an initial starting pset. If the hint is not in the pset, or
+        * if the hint is for a processor in an invalid state, discard
+        * the hint.
+        */
        if (processor != PROCESSOR_NULL) {
-               if (processor->processor_set != pset ||
-                   processor->state == PROCESSOR_INACTIVE ||
-                   processor->state == PROCESSOR_SHUTDOWN ||
-                   processor->state == PROCESSOR_OFF_LINE)
+               if (processor->processor_set != pset) {
                        processor = PROCESSOR_NULL;
-               else
-                       if (processor->state == PROCESSOR_IDLE ||
-                           ((thread->sched_pri >= BASEPRI_RTQUEUES) &&
-                           (processor->current_pri < BASEPRI_RTQUEUES)))
-                               return (processor);
+               } else {
+                       switch (processor->state) {
+                               case PROCESSOR_START:
+                               case PROCESSOR_SHUTDOWN:
+                               case PROCESSOR_OFF_LINE:
+                                       /*
+                                        * Hint is for a processor that cannot support running new threads.
+                                        */
+                                       processor = PROCESSOR_NULL;
+                                       break;
+                               case PROCESSOR_IDLE:
+                                       /*
+                                        * Hint is for an idle processor. Assume it is no worse than any other
+                                        * idle processor. The platform layer had an opportunity to provide
+                                        * the "least cost idle" processor above.
+                                        */
+                                       return (processor);
+                                       break;
+                               case PROCESSOR_RUNNING:
+                               case PROCESSOR_DISPATCHING:
+                                       /*
+                                        * Hint is for an active CPU. This fast-path allows
+                                        * realtime threads to preempt non-realtime threads
+                                        * to regain their previous executing processor.
+                                        */
+                                       if ((thread->sched_pri >= BASEPRI_RTQUEUES) &&
+                                               (processor->current_pri < BASEPRI_RTQUEUES))
+                                               return (processor);
+
+                                       /* Otherwise, use hint as part of search below */
+                                       break;
+                               default:
+                                       processor = PROCESSOR_NULL;
+                                       break;
+                       }
+               }
        }
 
        /*
-        *      Iterate through the processor sets to locate
-        *      an appropriate processor.
+        * Iterate through the processor sets to locate
+        * an appropriate processor. Seed results with
+        * a last-processor hint, if available, so that
+        * a search must find something strictly better
+        * to replace it.
+        *
+        * A primary/secondary pair of SMT processors are
+        * "unpaired" if the primary is busy but its
+        * corresponding secondary is idle (so the physical
+        * core has full use of its resources).
         */
+
+       integer_t lowest_priority = MAXPRI + 1;
+       integer_t lowest_unpaired_primary_priority = MAXPRI + 1;
+       integer_t lowest_count = INT_MAX;
+       uint64_t  furthest_deadline = 1;
+       processor_t lp_processor = PROCESSOR_NULL;
+       processor_t lp_unpaired_primary_processor = PROCESSOR_NULL;
+       processor_t lp_unpaired_secondary_processor = PROCESSOR_NULL;
+       processor_t lc_processor = PROCESSOR_NULL;
+       processor_t fd_processor = PROCESSOR_NULL;
+
+       if (processor != PROCESSOR_NULL) {
+               /* All other states should be enumerated above. */
+               assert(processor->state == PROCESSOR_RUNNING || processor->state == PROCESSOR_DISPATCHING);
+
+               lowest_priority = processor->current_pri;
+               lp_processor = processor;
+
+               if (processor->current_pri >= BASEPRI_RTQUEUES) {
+                       furthest_deadline = processor->deadline;
+                       fd_processor = processor;
+               }
+
+               lowest_count = SCHED(processor_runq_count)(processor);
+               lc_processor = processor;
+       }
+
        do {
+
                /*
-                *      Choose an idle processor.
+                * Choose an idle processor, in pset traversal order
                 */
                if (!queue_empty(&cset->idle_queue))
                        return ((processor_t)queue_first(&cset->idle_queue));
 
-               if (thread->sched_pri >= BASEPRI_RTQUEUES) {
-                       integer_t lowest_priority = MAXPRI + 1;
-                       integer_t lowest_unpaired = MAXPRI + 1;
-                       uint64_t  furthest_deadline = 1;
-                       processor_t lp_processor = PROCESSOR_NULL;
-                       processor_t lp_unpaired = PROCESSOR_NULL;
-                       processor_t fd_processor = PROCESSOR_NULL;
-
-                       lp_processor = cset->low_pri;
-                       /* Consider hinted processor */
-                       if (lp_processor != PROCESSOR_NULL &&
-                           ((lp_processor->processor_meta == PROCESSOR_META_NULL) ||
-                           ((lp_processor == lp_processor->processor_meta->primary) &&
-                           !queue_empty(&lp_processor->processor_meta->idle_queue))) &&
-                           lp_processor->state != PROCESSOR_INACTIVE &&
-                           lp_processor->state != PROCESSOR_SHUTDOWN &&
-                           lp_processor->state != PROCESSOR_OFF_LINE &&
-                           (lp_processor->current_pri < thread->sched_pri))
-                               return lp_processor;
+               /*
+                * Otherwise, enumerate active and idle processors to find candidates
+                * with lower priority/etc.
+                */
 
-                       processor = (processor_t)queue_first(&cset->active_queue);
-                       while (!queue_end(&cset->active_queue, (queue_entry_t)processor)) {
-                               /* Discover the processor executing the
-                                * thread with the lowest priority within
-                                * this pset, or the one with the furthest
-                                * deadline
-                                */
-                               integer_t cpri = processor->current_pri;
-                               if (cpri < lowest_priority) {
-                                       lowest_priority = cpri;
-                                       lp_processor = processor;
-                               }
+               processor = (processor_t)queue_first(&cset->active_queue);
+               while (!queue_end(&cset->active_queue, (queue_entry_t)processor)) {
 
-                               if ((cpri >= BASEPRI_RTQUEUES) && (processor->deadline > furthest_deadline)) {
-                                       furthest_deadline = processor->deadline;
-                                       fd_processor = processor;
-                               }
+                       integer_t cpri = processor->current_pri;
+                       if (cpri < lowest_priority) {
+                               lowest_priority = cpri;
+                               lp_processor = processor;
+                       }
 
+                       if ((cpri >= BASEPRI_RTQUEUES) && (processor->deadline > furthest_deadline)) {
+                               furthest_deadline = processor->deadline;
+                               fd_processor = processor;
+                       }
 
-                               if (processor->processor_meta != PROCESSOR_META_NULL &&
-                                   !queue_empty(&processor->processor_meta->idle_queue)) {
-                                       if (cpri < lowest_unpaired) {
-                                               lowest_unpaired = cpri;
-                                               lp_unpaired = processor;
-                                               pmeta = processor->processor_meta;
-                                       }
-                                       else
-                                               if (pmeta == PROCESSOR_META_NULL)
-                                                       pmeta = processor->processor_meta;
+                       integer_t ccount = SCHED(processor_runq_count)(processor);
+                       if (ccount < lowest_count) {
+                               lowest_count = ccount;
+                               lc_processor = processor;
+                       }
+
+                       processor = (processor_t)queue_next((queue_entry_t)processor);
+               }
+
+               /*
+                * For SMT configs, these idle secondary processors must have active primary. Otherwise
+                * the idle primary would have short-circuited the loop above
+                */
+               processor = (processor_t)queue_first(&cset->idle_secondary_queue);
+               while (!queue_end(&cset->idle_secondary_queue, (queue_entry_t)processor)) {
+                       processor_t cprimary = processor->processor_primary;
+
+                       /* If the primary processor is offline or starting up, it's not a candidate for this path */
+                       if (cprimary->state == PROCESSOR_RUNNING || cprimary->state == PROCESSOR_DISPATCHING) {
+                               integer_t primary_pri = cprimary->current_pri;
+
+                               if (primary_pri < lowest_unpaired_primary_priority) {
+                                       lowest_unpaired_primary_priority = primary_pri;
+                                       lp_unpaired_primary_processor = cprimary;
+                                       lp_unpaired_secondary_processor = processor;
                                }
-                               processor = (processor_t)queue_next((queue_entry_t)processor);
                        }
 
-                       if (thread->sched_pri > lowest_unpaired)
-                               return lp_unpaired;
+                       processor = (processor_t)queue_next((queue_entry_t)processor);
+               }
+
 
-                       if (pmeta != PROCESSOR_META_NULL)
-                               return ((processor_t)queue_first(&pmeta->idle_queue));
-                       if (thread->sched_pri > lowest_priority)
+               if (thread->sched_pri >= BASEPRI_RTQUEUES) {
+
+                       /*
+                        * For realtime threads, the most important aspect is
+                        * scheduling latency, so we attempt to assign threads
+                        * to good preemption candidates (assuming an idle primary
+                        * processor was not available above).
+                        */
+
+                       if (thread->sched_pri > lowest_unpaired_primary_priority) {
+                               /* Move to end of active queue so that the next thread doesn't also pick it */
+                               remqueue((queue_entry_t)lp_unpaired_primary_processor);
+                               enqueue_tail(&cset->active_queue, (queue_entry_t)lp_unpaired_primary_processor);
+                               return lp_unpaired_primary_processor;
+                       }
+                       if (thread->sched_pri > lowest_priority) {
+                               /* Move to end of active queue so that the next thread doesn't also pick it */
+                               remqueue((queue_entry_t)lp_processor);
+                               enqueue_tail(&cset->active_queue, (queue_entry_t)lp_processor);
                                return lp_processor;
+                       }
                        if (thread->realtime.deadline < furthest_deadline)
                                return fd_processor;
 
-                       processor = PROCESSOR_NULL;
-               }
-               else {
                        /*
-                        *      Check any hinted processors in the processor set if available.
+                        * If all primary and secondary CPUs are busy with realtime
+                        * threads with deadlines earlier than us, move on to next
+                        * pset.
                         */
-                       if (cset->low_pri != PROCESSOR_NULL && cset->low_pri->state != PROCESSOR_INACTIVE &&
-                                       cset->low_pri->state != PROCESSOR_SHUTDOWN && cset->low_pri->state != PROCESSOR_OFF_LINE &&
-                                               (processor == PROCESSOR_NULL ||
-                                                       (thread->sched_pri > BASEPRI_DEFAULT && cset->low_pri->current_pri < thread->sched_pri))) {
-                               processor = cset->low_pri;
+               }
+               else {
+
+                       if (thread->sched_pri > lowest_unpaired_primary_priority) {
+                               /* Move to end of active queue so that the next thread doesn't also pick it */
+                               remqueue((queue_entry_t)lp_unpaired_primary_processor);
+                               enqueue_tail(&cset->active_queue, (queue_entry_t)lp_unpaired_primary_processor);
+                               return lp_unpaired_primary_processor;
                        }
-                       else
-                       if (cset->low_count != PROCESSOR_NULL && cset->low_count->state != PROCESSOR_INACTIVE &&
-                                       cset->low_count->state != PROCESSOR_SHUTDOWN && cset->low_count->state != PROCESSOR_OFF_LINE &&
-                                               (processor == PROCESSOR_NULL || (thread->sched_pri <= BASEPRI_DEFAULT &&
-                                                                                                                       SCHED(processor_runq_count)(cset->low_count) < SCHED(processor_runq_count)(processor)))) {
-                               processor = cset->low_count;
+                       if (thread->sched_pri > lowest_priority) {
+                               /* Move to end of active queue so that the next thread doesn't also pick it */
+                               remqueue((queue_entry_t)lp_processor);
+                               enqueue_tail(&cset->active_queue, (queue_entry_t)lp_processor);
+                               return lp_processor;
                        }
 
                        /*
-                        *      Otherwise, choose an available processor in the set.
+                        * If all primary processor in this pset are running a higher
+                        * priority thread, move on to next pset. Only when we have
+                        * exhausted this search do we fall back to other heuristics.
                         */
-                       if (processor == PROCESSOR_NULL) {
-                               processor = (processor_t)dequeue_head(&cset->active_queue);
-                               if (processor != PROCESSOR_NULL)
-                                       enqueue_tail(&cset->active_queue, (queue_entry_t)processor);
-                       }
-
-                       if (processor != PROCESSOR_NULL && pmeta == PROCESSOR_META_NULL) {
-                               if (processor->processor_meta != PROCESSOR_META_NULL &&
-                                                                                       !queue_empty(&processor->processor_meta->idle_queue))
-                                       pmeta = processor->processor_meta;
-                       }
                }
 
                /*
-                *      Move onto the next processor set.
+                * Move onto the next processor set.
                 */
                nset = next_pset(cset);
 
@@ -3734,62 +3861,54 @@ choose_processor(
        } while (nset != pset);
 
        /*
-        *      Make sure that we pick a running processor,
-        *      and that the correct processor set is locked.
+        * Make sure that we pick a running processor,
+        * and that the correct processor set is locked.
+        * Since we may have unlock the candidate processor's
+        * pset, it may have changed state.
+        *
+        * All primary processors are running a higher priority
+        * thread, so the only options left are enqueuing on
+        * the secondary processor that would perturb the least priority
+        * primary, or the least busy primary.
         */
        do {
-               if (pmeta != PROCESSOR_META_NULL) {
-                       if (cset != pmeta->primary->processor_set) {
-                               pset_unlock(cset);
-
-                               cset = pmeta->primary->processor_set;
-                               pset_lock(cset);
-                       }
-
-                       if (!queue_empty(&pmeta->idle_queue))
-                               return ((processor_t)queue_first(&pmeta->idle_queue));
-
-                       pmeta = PROCESSOR_META_NULL;
-               }
-
-               /*
-                *      If we haven't been able to choose a processor,
-                *      pick the boot processor and return it.
-                */
-               if (processor == PROCESSOR_NULL) {
-                       processor = master_processor;
 
+               /* lowest_priority is evaluated in the main loops above */
+               if (lp_unpaired_secondary_processor != PROCESSOR_NULL) {
+                       processor = lp_unpaired_secondary_processor;
+                       lp_unpaired_secondary_processor = PROCESSOR_NULL;
+               } else if (lc_processor != PROCESSOR_NULL) {
+                       processor = lc_processor;
+                       lc_processor = PROCESSOR_NULL;
+               } else {
                        /*
-                        *      Check that the correct processor set is
-                        *      returned locked.
+                        * All processors are executing higher
+                        * priority threads, and the lowest_count
+                        * candidate was not usable
                         */
-                       if (cset != processor->processor_set) {
-                               pset_unlock(cset);
-
-                               cset = processor->processor_set;
-                               pset_lock(cset);
-                       }
-
-                       return (processor);
+                       processor = master_processor;
                }
 
                /*
-                *      Check that the processor set for the chosen
-                *      processor is locked.
+                * Check that the correct processor set is
+                * returned locked.
                 */
                if (cset != processor->processor_set) {
                        pset_unlock(cset);
-
                        cset = processor->processor_set;
                        pset_lock(cset);
                }
 
                /*
-                *      We must verify that the chosen processor is still available.
+                * We must verify that the chosen processor is still available.
+                * master_processor is an exception, since we may need to preempt
+                * a running thread on it during processor shutdown (for sleep),
+                * and that thread needs to be enqueued on its runqueue to run
+                * when the processor is restarted.
                 */
-               if (processor->state == PROCESSOR_INACTIVE ||
-                                       processor->state == PROCESSOR_SHUTDOWN || processor->state == PROCESSOR_OFF_LINE)
+               if (processor != master_processor && (processor->state == PROCESSOR_SHUTDOWN || processor->state == PROCESSOR_OFF_LINE))
                        processor = PROCESSOR_NULL;
+
        } while (processor == PROCESSOR_NULL);
 
        return (processor);
@@ -3812,9 +3931,7 @@ thread_setrun(
        processor_t                     processor;
        processor_set_t         pset;
 
-#if DEBUG
        assert(thread_runnable(thread));
-#endif
        
        /*
         *      Update priority if needed.
@@ -3822,6 +3939,8 @@ thread_setrun(
        if (SCHED(can_update_priority)(thread))
                SCHED(update_priority)(thread);
 
+       thread->sfi_class = sfi_thread_classify(thread);
+
        assert(thread->runq == PROCESSOR_NULL);
 
        if (thread->bound_processor == PROCESSOR_NULL) {
@@ -3892,6 +4011,7 @@ thread_setrun(
 
        /*
         *      Dispatch the thread on the choosen processor.
+        *      TODO: This should be based on sched_mode, not sched_pri
         */
        if (thread->sched_pri >= BASEPRI_RTQUEUES)
                realtime_setrun(processor, thread);
@@ -3985,11 +4105,12 @@ processor_queue_shutdown(
  *     Check for a preemption point in
  *     the current context.
  *
- *     Called at splsched.
+ *     Called at splsched with thread locked.
  */
 ast_t
 csw_check(
-       processor_t             processor)
+       processor_t             processor,
+       ast_t                   check_reason)
 {
        processor_set_t pset = processor->processor_set;
        ast_t                   result;
@@ -3999,7 +4120,7 @@ csw_check(
        /* If we were sent a remote AST and interrupted a running processor, acknowledge it here with pset lock held */
        pset->pending_AST_cpu_mask &= ~(1U << processor->cpu_id);
 
-       result = csw_check_locked(processor, pset);
+       result = csw_check_locked(processor, pset, check_reason);
 
        pset_unlock(pset);
 
@@ -4008,41 +4129,47 @@ csw_check(
 
 /*
  * Check for preemption at splsched with
- * pset locked
+ * pset and thread locked
  */
 ast_t
 csw_check_locked(
        processor_t             processor,
-       processor_set_t pset __unused)
+       processor_set_t pset __unused,
+       ast_t                   check_reason)
 {
-       ast_t                   result = AST_NONE;
+       ast_t                   result;
        thread_t                thread = processor->active_thread;
 
        if (first_timeslice(processor)) {
                if (rt_runq.count > 0)
-                       return (AST_PREEMPT | AST_URGENT);
+                       return (check_reason | AST_PREEMPT | AST_URGENT);
        }
        else {
                if (rt_runq.count > 0) {
                        if (BASEPRI_RTQUEUES > processor->current_pri)
-                               return (AST_PREEMPT | AST_URGENT);
+                               return (check_reason | AST_PREEMPT | AST_URGENT);
                        else
-                               return (AST_PREEMPT);
+                               return (check_reason | AST_PREEMPT);
                }
        }
 
        result = SCHED(processor_csw_check)(processor);
        if (result != AST_NONE)
-               return (result);
+               return (check_reason | result);
 
        if (SCHED(should_current_thread_rechoose_processor)(processor))
-               return (AST_PREEMPT);
+               return (check_reason | AST_PREEMPT);
        
-       if (machine_processor_is_inactive(processor))
-               return (AST_PREEMPT);
-
        if (thread->state & TH_SUSP)
-               return (AST_PREEMPT);
+               return (check_reason | AST_PREEMPT);
+
+       /*
+        * Current thread may not need to be preempted, but maybe needs
+        * an SFI wait?
+        */
+       result = sfi_thread_needs_ast(thread, NULL);
+       if (result != AST_NONE)
+               return (check_reason | result);
 
        return (AST_NONE);
 }
@@ -4059,11 +4186,31 @@ csw_check_locked(
 void
 set_sched_pri(
        thread_t                thread,
-       int                             priority)
+       int                     priority)
 {
        boolean_t               removed = thread_run_queue_remove(thread);
+       int curgency, nurgency;
+       uint64_t urgency_param1, urgency_param2;
+       thread_t cthread = current_thread();
 
+       if (thread == cthread) {
+               curgency = thread_get_urgency(thread, &urgency_param1, &urgency_param2);
+       }
+       
        thread->sched_pri = priority;
+
+       if (thread == cthread) {
+               nurgency = thread_get_urgency(thread, &urgency_param1, &urgency_param2);
+/* set_sched_pri doesn't alter RT params. We expect direct base priority/QoS
+ * class alterations from user space to occur relatively infrequently, hence
+ * those are lazily handled. QoS classes have distinct priority bands, and QoS
+ * inheritance is expected to involve priority changes.
+ */
+               if (nurgency != curgency) {
+                       thread_tell_urgency(nurgency, urgency_param1, urgency_param2, thread);
+               }
+       }
+
        if (removed)
                thread_setrun(thread, SCHED_PREEMPT | SCHED_TAILQ);
        else
@@ -4075,7 +4222,8 @@ set_sched_pri(
 
                        processor->current_pri = priority;
                        processor->current_thmode = thread->sched_mode;
-                       if ((preempt = csw_check(processor)) != AST_NONE)
+                       processor->current_sfi_class = thread->sfi_class = sfi_thread_classify(thread);
+                       if ((preempt = csw_check(processor, AST_NONE)) != AST_NONE)
                                ast_on(preempt);
                }
                else
@@ -4117,8 +4265,11 @@ run_queue_check(
 
 #if defined(CONFIG_SCHED_TRADITIONAL)
 
-/* locks the runqueue itself */
-
+/*
+ * Locks the runqueue itself.
+ *
+ * Thread must be locked.
+ */
 static boolean_t
 processor_queue_remove(
                                           processor_t                  processor,
@@ -4155,69 +4306,82 @@ processor_queue_remove(
 
 #endif /* CONFIG_SCHED_TRADITIONAL */
 
+
 /*
  *     thread_run_queue_remove:
  *
- *     Remove a thread from a current run queue and
+ *     Remove a thread from its current run queue and
  *     return TRUE if successful.
  *
  *     Thread must be locked.
+ *
+ *     If thread->runq is PROCESSOR_NULL, the thread will not re-enter the
+ *     run queues because the caller locked the thread.  Otherwise
+ *     the thread is on a run queue, but could be chosen for dispatch
+ *     and removed by another processor under a different lock, which
+ *     will set thread->runq to PROCESSOR_NULL.
+ *
+ *     Hence the thread select path must not rely on anything that could
+ *     be changed under the thread lock after calling this function,
+ *     most importantly thread->sched_pri.
  */
 boolean_t
 thread_run_queue_remove(
-       thread_t                thread)
+                        thread_t        thread)
 {
-       processor_t             processor = thread->runq;
+       boolean_t removed = FALSE;
+       processor_t processor = thread->runq;
 
-       /*
-        *      If processor is PROCESSOR_NULL, the thread will stay out of the
-        *      run queues because the caller locked the thread.  Otherwise
-        *      the thread is on a run queue, but could be chosen for dispatch
-        *      and removed.
-        */
-       if (processor != PROCESSOR_NULL) {
-               queue_t                 q;
+       if ((thread->state & (TH_RUN|TH_WAIT)) == TH_WAIT) {
+               /* Thread isn't runnable */
+               assert(thread->runq == PROCESSOR_NULL);
+               return FALSE;
+       }
 
+       if (processor == PROCESSOR_NULL) {
                /*
-                *      The processor run queues are locked by the
-                *      processor set.  Real-time priorities use a
-                *      global queue with a dedicated lock.
+                * The thread is either not on the runq,
+                * or is in the midst of being removed from the runq.
+                *
+                * runq is set to NULL under the pset lock, not the thread
+                * lock, so the thread may still be in the process of being dequeued
+                * from the runq. It will wait in invoke for the thread lock to be
+                * dropped.
                 */
-               if (thread->sched_mode == TH_MODE_FAIRSHARE) {
-                       return SCHED(fairshare_queue_remove)(thread);
-               }
-               
-               if (thread->sched_pri < BASEPRI_RTQUEUES) {
-                       return SCHED(processor_queue_remove)(processor, thread);
-               }
 
-               simple_lock(&rt_lock);
-               q = &rt_runq.queue;
+               return FALSE;
+       }
 
-               if (processor == thread->runq) {
-                       /*
-                        *      Thread is on a run queue and we have a lock on
-                        *      that run queue.
-                        */
-                       remqueue((queue_entry_t)thread);
-                       SCHED_STATS_RUNQ_CHANGE(&rt_runq.runq_stats, rt_runq.count);
-                       rt_runq.count--;
+       if (thread->sched_mode == TH_MODE_FAIRSHARE) {
+               return SCHED(fairshare_queue_remove)(thread);
+       }
+       
+       if (thread->sched_pri < BASEPRI_RTQUEUES) {
+               return SCHED(processor_queue_remove)(processor, thread);
+       }
 
-                       thread->runq = PROCESSOR_NULL;
-               }
-               else {
-                       /*
-                        *      The thread left the run queue before we could
-                        *      lock the run queue.
-                        */
-                       assert(thread->runq == PROCESSOR_NULL);
-                       processor = PROCESSOR_NULL;
-               }
+       simple_lock(&rt_lock);
 
-               simple_unlock(&rt_lock);
+       if (thread->runq != PROCESSOR_NULL) {
+               /*
+                *      Thread is on a run queue and we have a lock on
+                *      that run queue.
+                */
+
+               assert(thread->runq == RT_RUNQ);
+
+               remqueue((queue_entry_t)thread);
+               SCHED_STATS_RUNQ_CHANGE(&rt_runq.runq_stats, rt_runq.count);
+               rt_runq.count--;
+
+               thread->runq = PROCESSOR_NULL;
+
+               removed = TRUE;
        }
 
-       return (processor != PROCESSOR_NULL);
+       simple_unlock(&rt_lock);
+
+       return (removed);
 }
 
 #if defined(CONFIG_SCHED_TRADITIONAL)
@@ -4360,14 +4524,18 @@ thread_get_urgency(thread_t thread, uint64_t *arg1, uint64_t *arg2)
                   ((thread->sched_pri <= MAXPRI_THROTTLE) && (thread->priority <= MAXPRI_THROTTLE)))  {
                /*
                 * Background urgency applied when thread priority is MAXPRI_THROTTLE or lower and thread is not promoted
+                * TODO: Use TH_SFLAG_THROTTLED instead?
                 */
                *arg1 = thread->sched_pri;
                *arg2 = thread->priority;
 
                return (THREAD_URGENCY_BACKGROUND);
        } else {
-               *arg1 = thread->sched_pri;
-               *arg2 = thread->priority;
+               /* For otherwise unclassified threads, report throughput QoS
+                * parameters
+                */
+               *arg1 = thread->effective_policy.t_through_qos;
+               *arg2 = thread->task->effective_policy.t_through_qos;
                
                return (THREAD_URGENCY_NORMAL);
        }
@@ -4409,13 +4577,10 @@ processor_idle(
        PROCESSOR_DATA(processor, current_state) = &PROCESSOR_DATA(processor, idle_state);
 
        while (1) {
-
                if (processor->state != PROCESSOR_IDLE) /* unsafe, but worst case we loop around once */
                        break;
                if (pset->pending_AST_cpu_mask & (1U << processor->cpu_id))
                        break;
-               if (!SCHED(processor_queue_empty)(processor))
-                       break;
                if (rt_runq.count)
                        break;
 #if CONFIG_SCHED_IDLE_IN_PLACE
@@ -4440,8 +4605,13 @@ processor_idle(
                IDLE_KERNEL_DEBUG_CONSTANT(
                        MACHDBG_CODE(DBG_MACH_SCHED,MACH_IDLE) | DBG_FUNC_NONE, (uintptr_t)thread_tid(thread), rt_runq.count, SCHED(processor_runq_count)(processor), -2, 0);
 
-               if (processor->state == PROCESSOR_INACTIVE && !machine_processor_is_inactive(processor))
-                       break;
+               if (!SCHED(processor_queue_empty)(processor)) {
+                       /* Secondary SMT processors respond to directed wakeups
+                        * exclusively. Some platforms induce 'spurious' SMT wakeups.
+                        */
+                       if (processor->processor_primary == processor)
+                                       break;
+               }
        }
 
        timer_switch(&PROCESSOR_DATA(processor, idle_state),
@@ -4464,8 +4634,10 @@ processor_idle(
 
                if ((new_thread != THREAD_NULL) && (SCHED(processor_queue_has_priority)(processor, new_thread->sched_pri, FALSE)                                        ||
                                                                                        (rt_runq.count > 0 && BASEPRI_RTQUEUES >= new_thread->sched_pri))       ) {
+                       /* Something higher priority has popped up on the runqueue - redispatch this thread elsewhere */
                        processor->current_pri = IDLEPRI;
                        processor->current_thmode = TH_MODE_FIXED;
+                       processor->current_sfi_class = SFI_CLASS_KERNEL;
                        processor->deadline = UINT64_MAX;
 
                        pset_unlock(pset);
@@ -4497,15 +4669,11 @@ processor_idle(
                processor->state = PROCESSOR_RUNNING;
                processor->current_pri = IDLEPRI;
                processor->current_thmode = TH_MODE_FIXED;
+               processor->current_sfi_class = SFI_CLASS_KERNEL;
                processor->deadline = UINT64_MAX;
                enqueue_tail(&pset->active_queue, (queue_entry_t)processor);
        }
        else
-       if (state == PROCESSOR_INACTIVE) {
-               processor->state = PROCESSOR_RUNNING;
-               enqueue_tail(&pset->active_queue, (queue_entry_t)processor);
-       }
-       else
        if (state == PROCESSOR_SHUTDOWN) {
                /*
                 *      Going off-line.  Force a
@@ -4515,6 +4683,7 @@ processor_idle(
                        processor->next_thread = THREAD_NULL;
                        processor->current_pri = IDLEPRI;
                        processor->current_thmode = TH_MODE_FIXED;
+                       processor->current_sfi_class = SFI_CLASS_KERNEL;
                        processor->deadline = UINT64_MAX;
 
                        pset_unlock(pset);
@@ -4619,9 +4788,12 @@ sched_startup(void)
        thread_block(THREAD_CONTINUE_NULL);
 }
 
-#if defined(CONFIG_SCHED_TRADITIONAL)
+#if defined(CONFIG_SCHED_TIMESHARE_CORE)
 
 static volatile uint64_t               sched_maintenance_deadline;
+#if defined(CONFIG_TELEMETRY)
+static volatile uint64_t               sched_telemetry_deadline = 0;
+#endif
 static uint64_t                                sched_tick_last_abstime;
 static uint64_t                                sched_tick_delta;
 uint64_t                               sched_tick_max_delta;
@@ -4631,17 +4803,20 @@ uint64_t                                sched_tick_max_delta;
  *     Perform periodic bookkeeping functions about ten
  *     times per second.
  */
-static void
+void
 sched_traditional_maintenance_continue(void)
 {
-       uint64_t        sched_tick_ctime;
-       sched_tick_ctime = mach_absolute_time();
+       uint64_t        sched_tick_ctime, late_time;
+
+       sched_tick_ctime = mach_absolute_time();        
 
        if (__improbable(sched_tick_last_abstime == 0)) {
                sched_tick_last_abstime = sched_tick_ctime;
+               late_time = 0;
                sched_tick_delta = 1;
        } else {
-               sched_tick_delta = ((sched_tick_ctime) - sched_tick_last_abstime) / sched_tick_interval;
+               late_time = sched_tick_ctime - sched_tick_last_abstime;
+               sched_tick_delta = late_time / sched_tick_interval;
                /* Ensure a delta of 1, since the interval could be slightly
                 * smaller than the sched_tick_interval due to dispatch
                 * latencies.
@@ -4660,6 +4835,13 @@ sched_traditional_maintenance_continue(void)
                sched_tick_max_delta = MAX(sched_tick_delta, sched_tick_max_delta);
        }
 
+       KERNEL_DEBUG_CONSTANT(MACHDBG_CODE(DBG_MACH_SCHED, MACH_SCHED_MAINTENANCE)|DBG_FUNC_START,
+                                                 sched_tick_delta,
+                                                 late_time,
+                                                 0,
+                                                 0,
+                                                 0);
+
        /* Add a number of pseudo-ticks corresponding to the elapsed interval
         * This could be greater than 1 if substantial intervals where
         * all processors are idle occur, which rarely occurs in practice.
@@ -4676,7 +4858,14 @@ sched_traditional_maintenance_continue(void)
         *  Scan the run queues for threads which
         *  may need to be updated.
         */
-       thread_update_scan();
+       SCHED(thread_update_scan)();
+
+       KERNEL_DEBUG_CONSTANT(MACHDBG_CODE(DBG_MACH_SCHED, MACH_SCHED_MAINTENANCE)|DBG_FUNC_END,
+                                                 sched_pri_shift,
+                                                 sched_background_pri_shift,
+                                                 0,
+                                                 0,
+                                                 0);
 
        assert_wait((event_t)sched_traditional_maintenance_continue, THREAD_UNINT);
        thread_block((thread_continue_t)sched_traditional_maintenance_continue);
@@ -4711,9 +4900,29 @@ sched_traditional_consider_maintenance(uint64_t ctime) {
                        sched_maintenance_wakeups++;
                }
        }
+
+#if defined(CONFIG_TELEMETRY)
+       /*
+        * Windowed telemetry is driven by the scheduler.  It should be safe
+        * to call compute_telemetry_windowed() even when windowed telemetry
+        * is disabled, but we should try to avoid doing extra work for no
+        * reason.
+        */
+       if (telemetry_window_enabled) {
+               deadline = sched_telemetry_deadline;
+
+               if (__improbable(ctime >= deadline)) {
+                       ndeadline = ctime + sched_telemetry_interval;
+
+                       if (__probable(__sync_bool_compare_and_swap(&sched_telemetry_deadline, deadline, ndeadline))) {
+                               compute_telemetry_windowed();
+                       }
+               }
+       }
+#endif /* CONFIG_TELEMETRY */
 }
 
-#endif /* CONFIG_SCHED_TRADITIONAL */
+#endif /* CONFIG_SCHED_TIMESHARE_CORE */
 
 void
 sched_init_thread(void (*continuation)(void))
@@ -4726,7 +4935,7 @@ sched_init_thread(void (*continuation)(void))
        /*NOTREACHED*/
 }
 
-#if defined(CONFIG_SCHED_TRADITIONAL)
+#if defined(CONFIG_SCHED_TIMESHARE_CORE)
 
 /*
  *     thread_update_scan / runq_scan:
@@ -4749,12 +4958,44 @@ sched_init_thread(void (*continuation)(void))
 static thread_t                thread_update_array[THREAD_UPDATE_SIZE];
 static int                     thread_update_count = 0;
 
+/* Returns TRUE if thread was added, FALSE if thread_update_array is full */
+boolean_t
+thread_update_add_thread(thread_t thread)
+{
+       if (thread_update_count == THREAD_UPDATE_SIZE)
+               return (FALSE);
+
+       thread_update_array[thread_update_count++] = thread;
+       thread_reference_internal(thread);
+       return (TRUE);
+}
+
+void
+thread_update_process_threads(void)
+{
+       while (thread_update_count > 0) {
+               spl_t   s;
+               thread_t thread = thread_update_array[--thread_update_count];
+               thread_update_array[thread_update_count] = THREAD_NULL;
+
+               s = splsched();
+               thread_lock(thread);
+               if (!(thread->state & (TH_WAIT)) && (SCHED(can_update_priority)(thread))) {
+                       SCHED(update_priority)(thread);
+               }
+               thread_unlock(thread);
+               splx(s);
+
+               thread_deallocate(thread);
+       }
+}
+
 /*
  *     Scan a runq for candidate threads.
  *
  *     Returns TRUE if retry is needed.
  */
-static boolean_t
+boolean_t
 runq_scan(
        run_queue_t                             runq)
 {
@@ -4768,11 +5009,8 @@ runq_scan(
                        queue_iterate(q, thread, thread_t, links) {
                                if (            thread->sched_stamp != sched_tick               &&
                                                (thread->sched_mode == TH_MODE_TIMESHARE)       ) {
-                                       if (thread_update_count == THREAD_UPDATE_SIZE)
+                                       if (thread_update_add_thread(thread) == FALSE)
                                                return (TRUE);
-
-                                       thread_update_array[thread_update_count++] = thread;
-                                       thread_reference_internal(thread);
                                }
 
                                count--;
@@ -4785,6 +5023,10 @@ runq_scan(
        return (FALSE);
 }
 
+#endif /* CONFIG_SCHED_TIMESHARE_CORE */
+
+#if defined(CONFIG_SCHED_TRADITIONAL)
+
 static void
 thread_update_scan(void)
 {
@@ -4796,6 +5038,10 @@ thread_update_scan(void)
 
        do {
                do {
+                       /*
+                        * TODO: in sched_traditional_use_pset_runqueue case,
+                        *  avoid scanning the same runq multiple times
+                        */
                        pset = processor->processor_set;
 
                        s = splsched();
@@ -4811,34 +5057,15 @@ thread_update_scan(void)
 
                        thread = processor->idle_thread;
                        if (thread != THREAD_NULL && thread->sched_stamp != sched_tick) {
-                               if (thread_update_count == THREAD_UPDATE_SIZE) {
+                               if (thread_update_add_thread(thread) == FALSE) {
                                        restart_needed = TRUE;
                                        break;
                                }
-
-                               thread_update_array[thread_update_count++] = thread;
-                               thread_reference_internal(thread);
                        }
                } while ((processor = processor->processor_list) != NULL);
 
-           /*
-            *  Ok, we now have a collection of candidates -- fix them.
-            */
-           while (thread_update_count > 0) {
-                       thread = thread_update_array[--thread_update_count];
-                       thread_update_array[thread_update_count] = THREAD_NULL;
-
-                       s = splsched();
-                       thread_lock(thread);
-                       if (    !(thread->state & (TH_WAIT))    ) {
-                               if (SCHED(can_update_priority)(thread))
-                                       SCHED(update_priority)(thread);
-                       }
-                       thread_unlock(thread);
-                       splx(s);
-
-                       thread_deallocate(thread);
-           }
+               /* Ok, we now have a collection of candidates -- fix them. */
+               thread_update_process_threads();
        } while (restart_needed);
 }
 
@@ -4864,9 +5091,9 @@ thread_set_eager_preempt(thread_t thread)
        thread->sched_flags |= TH_SFLAG_EAGERPREEMPT;
 
        if (thread == current_thread()) {
-               thread_unlock(thread);
 
-               ast = csw_check(p);
+               ast = csw_check(p, AST_NONE);
+               thread_unlock(thread);
                if (ast != AST_NONE) {
                        (void) thread_block_reason(THREAD_CONTINUE_NULL, NULL, ast);
                }
@@ -4958,14 +5185,12 @@ preemption_enabled(void)
        return (get_preemption_level() == 0 && ml_get_interrupts_enabled());
 }
 
-#if    DEBUG
-static boolean_t
+__assert_only static boolean_t
 thread_runnable(
        thread_t        thread)
 {
        return ((thread->state & (TH_RUN|TH_WAIT)) == TH_RUN);
 }
-#endif /* DEBUG */
 
 static void
 sched_timer_deadline_tracking_init(void) {