darling-xnu/osfmk/kern/sched.h
2023-05-16 21:41:14 -07:00

437 lines
15 KiB
C

/*
* Copyright (c) 2000-2019 Apple Inc. All rights reserved.
*
* @APPLE_OSREFERENCE_LICENSE_HEADER_START@
*
* This file contains Original Code and/or Modifications of Original Code
* as defined in and that are subject to the Apple Public Source License
* Version 2.0 (the 'License'). You may not use this file except in
* compliance with the License. The rights granted to you under the License
* may not be used to create, or enable the creation or redistribution of,
* unlawful or unlicensed copies of an Apple operating system, or to
* circumvent, violate, or enable the circumvention or violation of, any
* terms of an Apple operating system software license agreement.
*
* Please obtain a copy of the License at
* http://www.opensource.apple.com/apsl/ and read it before using this file.
*
* The Original Code and all software distributed under the License are
* distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
* EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
* INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
* Please see the License for the specific language governing rights and
* limitations under the License.
*
* @APPLE_OSREFERENCE_LICENSE_HEADER_END@
*/
/*
* @OSF_COPYRIGHT@
*/
/*
* Mach Operating System
* Copyright (c) 1991,1990,1989,1988,1987 Carnegie Mellon University
* All Rights Reserved.
*
* Permission to use, copy, modify and distribute this software and its
* documentation is hereby granted, provided that both the copyright
* notice and this permission notice appear in all copies of the
* software, derivative works or modified versions, and any portions
* thereof, and that both notices appear in supporting documentation.
*
* CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS"
* CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND FOR
* ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE.
*
* Carnegie Mellon requests users of this software to return to
*
* Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU
* School of Computer Science
* Carnegie Mellon University
* Pittsburgh PA 15213-3890
*
* any improvements or extensions that they make and grant Carnegie Mellon
* the rights to redistribute these changes.
*/
/*
*/
/*
* File: sched.h
* Author: Avadis Tevanian, Jr.
* Date: 1985
*
* Header file for scheduler.
*
*/
#ifndef _KERN_SCHED_H_
#define _KERN_SCHED_H_
#include <mach/policy.h>
#include <kern/kern_types.h>
#include <kern/smp.h>
#include <kern/circle_queue.h>
#include <kern/macro_help.h>
#include <kern/timer_call.h>
#include <kern/ast.h>
#include <kern/bits.h>
#define NRQS_MAX (128) /* maximum number of priority levels */
#define MAXPRI (NRQS_MAX-1)
#define MINPRI 0 /* lowest legal priority schedulable */
#define IDLEPRI MINPRI /* idle thread priority */
#define NOPRI -1
/*
* High-level priority assignments
*
*************************************************************************
* 127 Reserved (real-time)
* A
* +
* (32 levels)
* +
* V
* 96 Reserved (real-time)
* 95 Kernel mode only
* A
* +
* (16 levels)
* +
* V
* 80 Kernel mode only
* 79 System high priority
* A
* +
* (16 levels)
* +
* V
* 64 System high priority
* 63 Elevated priorities
* A
* +
* (12 levels)
* +
* V
* 52 Elevated priorities
* 51 Elevated priorities (incl. BSD +nice)
* A
* +
* (20 levels)
* +
* V
* 32 Elevated priorities (incl. BSD +nice)
* 31 Default (default base for threads)
* 30 Lowered priorities (incl. BSD -nice)
* A
* +
* (20 levels)
* +
* V
* 11 Lowered priorities (incl. BSD -nice)
* 10 Lowered priorities (aged pri's)
* A
* +
* (11 levels)
* +
* V
* 0 Lowered priorities (aged pri's / idle)
*************************************************************************
*/
#define BASEPRI_RTQUEUES (BASEPRI_REALTIME + 1) /* 97 */
#define BASEPRI_REALTIME (MAXPRI - (NRQS_MAX / 4) + 1) /* 96 */
#define MAXPRI_KERNEL (BASEPRI_REALTIME - 1) /* 95 */
#define BASEPRI_PREEMPT_HIGH (BASEPRI_PREEMPT + 1) /* 93 */
#define BASEPRI_PREEMPT (MAXPRI_KERNEL - 3) /* 92 */
#define BASEPRI_VM (BASEPRI_PREEMPT - 1) /* 91 */
#define BASEPRI_KERNEL (MINPRI_KERNEL + 1) /* 81 */
#define MINPRI_KERNEL (MAXPRI_KERNEL - (NRQS_MAX / 8) + 1) /* 80 */
#define MAXPRI_RESERVED (MINPRI_KERNEL - 1) /* 79 */
#define BASEPRI_GRAPHICS (MAXPRI_RESERVED - 3) /* 76 */
#define MINPRI_RESERVED (MAXPRI_RESERVED - (NRQS_MAX / 8) + 1) /* 64 */
#define MAXPRI_USER (MINPRI_RESERVED - 1) /* 63 */
#define BASEPRI_CONTROL (BASEPRI_DEFAULT + 17) /* 48 */
#define BASEPRI_FOREGROUND (BASEPRI_DEFAULT + 16) /* 47 */
#define BASEPRI_BACKGROUND (BASEPRI_DEFAULT + 15) /* 46 */
#define BASEPRI_USER_INITIATED (BASEPRI_DEFAULT + 6) /* 37 */
#define BASEPRI_DEFAULT (MAXPRI_USER - (NRQS_MAX / 4)) /* 31 */
#define MAXPRI_SUPPRESSED (BASEPRI_DEFAULT - 3) /* 28 */
#define BASEPRI_UTILITY (BASEPRI_DEFAULT - 11) /* 20 */
#define MAXPRI_THROTTLE (MINPRI + 4) /* 4 */
#define MINPRI_USER MINPRI /* 0 */
#define DEPRESSPRI (MINPRI) /* depress priority */
#define MAXPRI_PROMOTE (MAXPRI_KERNEL) /* ceiling for mutex promotion */
#define MINPRI_RWLOCK (BASEPRI_BACKGROUND) /* floor when holding rwlock count */
#define MINPRI_EXEC (BASEPRI_DEFAULT) /* floor when in exec state */
#define MINPRI_WAITQ (BASEPRI_DEFAULT) /* floor when in waitq handover state */
#define NRQS (BASEPRI_REALTIME) /* Non-realtime levels for runqs */
/* Ensure that NRQS is large enough to represent all non-realtime threads; even promoted ones */
_Static_assert((NRQS == (MAXPRI_PROMOTE + 1)), "Runqueues are too small to hold all non-realtime threads");
/* Type used for thread->sched_mode and saved_mode */
typedef enum {
TH_MODE_NONE = 0, /* unassigned, usually for saved_mode only */
TH_MODE_REALTIME, /* time constraints supplied */
TH_MODE_FIXED, /* use fixed priorities, no decay */
TH_MODE_TIMESHARE, /* use timesharing algorithm */
} sched_mode_t;
/*
* Since the clutch scheduler organizes threads based on the thread group
* and the scheduling bucket, its important to not mix threads from multiple
* priority bands into the same bucket. To achieve that, in the clutch bucket
* world, there is a scheduling bucket per QoS effectively.
*/
/* Buckets used for load calculation */
typedef enum {
TH_BUCKET_FIXPRI = 0, /* Fixed-priority */
TH_BUCKET_SHARE_FG, /* Timeshare thread above BASEPRI_DEFAULT */
#if CONFIG_SCHED_CLUTCH
TH_BUCKET_SHARE_IN, /* Timeshare thread between BASEPRI_USER_INITIATED and BASEPRI_DEFAULT */
#endif /* CONFIG_SCHED_CLUTCH */
TH_BUCKET_SHARE_DF, /* Timeshare thread between BASEPRI_DEFAULT and BASEPRI_UTILITY */
TH_BUCKET_SHARE_UT, /* Timeshare thread between BASEPRI_UTILITY and MAXPRI_THROTTLE */
TH_BUCKET_SHARE_BG, /* Timeshare thread between MAXPRI_THROTTLE and MINPRI */
TH_BUCKET_RUN, /* All runnable threads */
TH_BUCKET_SCHED_MAX = TH_BUCKET_RUN, /* Maximum schedulable buckets */
TH_BUCKET_MAX,
} sched_bucket_t;
/*
* Macro to check for invalid priorities.
*/
#define invalid_pri(pri) ((pri) < MINPRI || (pri) > MAXPRI)
struct runq_stats {
uint64_t count_sum;
uint64_t last_change_timestamp;
};
#if defined(CONFIG_SCHED_TIMESHARE_CORE) || defined(CONFIG_SCHED_PROTO)
struct run_queue {
int highq; /* highest runnable queue */
bitmap_t bitmap[BITMAP_LEN(NRQS)]; /* run queue bitmap array */
int count; /* # of threads total */
int urgency; /* level of preemption urgency */
circle_queue_head_t queues[NRQS]; /* one for each priority */
struct runq_stats runq_stats;
};
inline static void
rq_bitmap_set(bitmap_t *map, u_int n)
{
assert(n < NRQS);
bitmap_set(map, n);
}
inline static void
rq_bitmap_clear(bitmap_t *map, u_int n)
{
assert(n < NRQS);
bitmap_clear(map, n);
}
#endif /* defined(CONFIG_SCHED_TIMESHARE_CORE) || defined(CONFIG_SCHED_PROTO) */
struct rt_queue {
_Atomic int count; /* # of threads total */
queue_head_t queue; /* all runnable RT threads */
struct runq_stats runq_stats;
};
typedef struct rt_queue *rt_queue_t;
#if defined(CONFIG_SCHED_GRRR_CORE)
/*
* We map standard Mach priorities to an abstract scale that more properly
* indicates how we want processor time allocated under contention.
*/
typedef uint8_t grrr_proportional_priority_t;
typedef uint8_t grrr_group_index_t;
#define NUM_GRRR_PROPORTIONAL_PRIORITIES 256
#define MAX_GRRR_PROPORTIONAL_PRIORITY ((grrr_proportional_priority_t)255)
#if 0
#define NUM_GRRR_GROUPS 8 /* log(256) */
#endif
#define NUM_GRRR_GROUPS 64 /* 256/4 */
struct grrr_group {
queue_chain_t priority_order; /* next greatest weight group */
grrr_proportional_priority_t minpriority;
grrr_group_index_t index;
queue_head_t clients;
int count;
uint32_t weight;
#if 0
uint32_t deferred_removal_weight;
#endif
uint32_t work;
thread_t current_client;
};
struct grrr_run_queue {
int count;
uint32_t last_rescale_tick;
struct grrr_group groups[NUM_GRRR_GROUPS];
queue_head_t sorted_group_list;
uint32_t weight;
grrr_group_t current_group;
struct runq_stats runq_stats;
};
#endif /* defined(CONFIG_SCHED_GRRR_CORE) */
extern int rt_runq_count(processor_set_t);
extern void rt_runq_count_incr(processor_set_t);
extern void rt_runq_count_decr(processor_set_t);
#if defined(CONFIG_SCHED_MULTIQ)
sched_group_t sched_group_create(void);
void sched_group_destroy(sched_group_t sched_group);
#endif /* defined(CONFIG_SCHED_MULTIQ) */
/*
* Scheduler routines.
*/
/* Handle quantum expiration for an executing thread */
extern void thread_quantum_expire(
timer_call_param_t processor,
timer_call_param_t thread);
/* Context switch check for current processor */
extern ast_t csw_check(
thread_t thread,
processor_t processor,
ast_t check_reason);
/* Check for pending ASTs */
extern void ast_check(processor_t processor);
extern void sched_update_generation_count(void);
#if defined(CONFIG_SCHED_TIMESHARE_CORE)
extern uint32_t std_quantum, min_std_quantum;
extern uint32_t std_quantum_us;
#endif /* CONFIG_SCHED_TIMESHARE_CORE */
extern uint32_t thread_depress_time;
extern uint32_t default_timeshare_computation;
extern uint32_t default_timeshare_constraint;
extern uint32_t max_rt_quantum, min_rt_quantum;
extern int default_preemption_rate;
#if defined(CONFIG_SCHED_TIMESHARE_CORE)
/*
* Age usage at approximately (1 << SCHED_TICK_SHIFT) times per second
* Aging may be deferred during periods where all processors are idle
* and cumulatively applied during periods of activity.
*/
#define SCHED_TICK_SHIFT 3
#define SCHED_TICK_MAX_DELTA (8)
extern unsigned sched_tick;
extern uint32_t sched_tick_interval;
#endif /* CONFIG_SCHED_TIMESHARE_CORE */
extern uint64_t sched_one_second_interval;
/* Periodic computation of various averages */
extern void compute_sched_load(void);
extern void compute_averages(uint64_t);
extern void compute_averunnable(
void *nrun);
extern void compute_stack_target(
void *arg);
extern void compute_pageout_gc_throttle(
void *arg);
extern void compute_pmap_gc_throttle(
void *arg);
/*
* Conversion factor from usage
* to priority.
*/
#if defined(CONFIG_SCHED_TIMESHARE_CORE)
#define MAX_LOAD (NRQS - 1)
#define SCHED_PRI_SHIFT_MAX ((8 * sizeof(uint32_t)) - 1)
extern uint32_t sched_pri_shifts[TH_BUCKET_MAX];
extern uint32_t sched_fixed_shift;
extern int8_t sched_load_shifts[NRQS];
extern uint32_t sched_decay_usage_age_factor;
void sched_timeshare_consider_maintenance(uint64_t ctime);
#endif /* CONFIG_SCHED_TIMESHARE_CORE */
void sched_consider_recommended_cores(uint64_t ctime, thread_t thread);
extern int32_t sched_poll_yield_shift;
extern uint64_t sched_safe_duration;
extern uint32_t sched_load_average, sched_mach_factor;
extern uint32_t avenrun[3], mach_factor[3];
extern uint64_t max_unsafe_computation;
extern uint64_t max_poll_computation;
extern uint32_t sched_run_buckets[TH_BUCKET_MAX];
extern uint32_t sched_run_incr(thread_t thread);
extern uint32_t sched_run_decr(thread_t thread);
extern void sched_update_thread_bucket(thread_t thread);
extern uint32_t sched_smt_run_incr(thread_t thread);
extern uint32_t sched_smt_run_decr(thread_t thread);
extern void sched_smt_update_thread_bucket(thread_t thread);
#define SCHED_DECAY_TICKS 32
struct shift_data {
int shift1;
int shift2;
};
/*
* thread_timer_delta macro takes care of both thread timers.
*/
#define thread_timer_delta(thread, delta) \
MACRO_BEGIN \
(delta) = (typeof(delta))timer_delta(&(thread)->system_timer, \
&(thread)->system_timer_save); \
(delta) += (typeof(delta))timer_delta(&(thread)->user_timer, \
&(thread)->user_timer_save); \
MACRO_END
extern bool system_is_SMT;
#endif /* _KERN_SCHED_H_ */