blob: a6e658ba41c12048772023747d75f682fdca3d7f [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001#ifndef _LINUX_SCHED_H
2#define _LINUX_SCHED_H
3
David Howells607ca462012-10-13 10:46:48 +01004#include <uapi/linux/sched.h>
David Woodhouseb7b3c762006-04-27 00:12:56 +01005
Dongsheng Yang5c228072014-01-27 17:15:37 -05006#include <linux/sched/prio.h>
7
David Woodhouseb7b3c762006-04-27 00:12:56 +01008
9struct sched_param {
10 int sched_priority;
11};
12
Linus Torvalds1da177e2005-04-16 15:20:36 -070013#include <asm/param.h> /* for HZ */
14
Linus Torvalds1da177e2005-04-16 15:20:36 -070015#include <linux/capability.h>
16#include <linux/threads.h>
17#include <linux/kernel.h>
18#include <linux/types.h>
19#include <linux/timex.h>
20#include <linux/jiffies.h>
Peter Zijlstrafb00aca2013-11-07 14:43:43 +010021#include <linux/plist.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070022#include <linux/rbtree.h>
23#include <linux/thread_info.h>
24#include <linux/cpumask.h>
25#include <linux/errno.h>
26#include <linux/nodemask.h>
Martin Schwidefskyc92ff1b2007-10-16 01:24:43 -070027#include <linux/mm_types.h>
Frederic Weisbecker92cf2112015-05-12 16:41:46 +020028#include <linux/preempt.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070029
Linus Torvalds1da177e2005-04-16 15:20:36 -070030#include <asm/page.h>
31#include <asm/ptrace.h>
Frederic Weisbeckerbfc3f022014-03-05 16:33:42 +010032#include <linux/cputime.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070033
34#include <linux/smp.h>
35#include <linux/sem.h>
Jack Millerab602f72014-08-08 14:23:19 -070036#include <linux/shm.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070037#include <linux/signal.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070038#include <linux/compiler.h>
39#include <linux/completion.h>
40#include <linux/pid.h>
41#include <linux/percpu.h>
42#include <linux/topology.h>
43#include <linux/seccomp.h>
Ingo Molnare56d0902006-01-08 01:01:37 -080044#include <linux/rcupdate.h>
Jiri Pirko05725f72009-04-14 20:17:16 +020045#include <linux/rculist.h>
Ingo Molnar23f78d4a2006-06-27 02:54:53 -070046#include <linux/rtmutex.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070047
David Woodhousea3b67142006-04-25 14:54:40 +010048#include <linux/time.h>
49#include <linux/param.h>
50#include <linux/resource.h>
51#include <linux/timer.h>
52#include <linux/hrtimer.h>
Dmitry Vyukov5c9a8752016-03-22 14:27:30 -070053#include <linux/kcov.h>
Andrew Morton7c3ab732006-12-10 02:19:19 -080054#include <linux/task_io_accounting.h>
Arjan van de Ven97455122008-01-25 21:08:34 +010055#include <linux/latencytop.h>
David Howells9e2b2dc2008-08-13 16:20:04 +010056#include <linux/cred.h>
Peter Zijlstrafa14ff42011-09-12 13:06:17 +020057#include <linux/llist.h>
Eric W. Biederman7b44ab92011-11-16 23:20:58 -080058#include <linux/uidgid.h>
Ming Lei21caf2f2013-02-22 16:34:08 -080059#include <linux/gfp.h>
Aaron Tomlind4311ff2014-09-12 14:16:17 +010060#include <linux/magic.h>
Tejun Heo7d7efec2015-05-13 16:35:16 -040061#include <linux/cgroup-defs.h>
David Woodhousea3b67142006-04-25 14:54:40 +010062
63#include <asm/processor.h>
H. J. Lu36d57ac2005-09-06 15:16:49 -070064
Dario Faggiolid50dde52013-11-07 14:43:36 +010065#define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
66
67/*
68 * Extended scheduling parameters data structure.
69 *
70 * This is needed because the original struct sched_param can not be
71 * altered without introducing ABI issues with legacy applications
72 * (e.g., in sched_getparam()).
73 *
74 * However, the possibility of specifying more than just a priority for
75 * the tasks may be useful for a wide variety of application fields, e.g.,
76 * multimedia, streaming, automation and control, and many others.
77 *
78 * This variant (sched_attr) is meant at describing a so-called
79 * sporadic time-constrained task. In such model a task is specified by:
80 * - the activation period or minimum instance inter-arrival time;
81 * - the maximum (or average, depending on the actual scheduling
82 * discipline) computation time of all instances, a.k.a. runtime;
83 * - the deadline (relative to the actual activation time) of each
84 * instance.
85 * Very briefly, a periodic (sporadic) task asks for the execution of
86 * some specific computation --which is typically called an instance--
87 * (at most) every period. Moreover, each instance typically lasts no more
88 * than the runtime and must be completed by time instant t equal to
89 * the instance activation time + the deadline.
90 *
91 * This is reflected by the actual fields of the sched_attr structure:
92 *
93 * @size size of the structure, for fwd/bwd compat.
94 *
95 * @sched_policy task's scheduling policy
96 * @sched_flags for customizing the scheduler behaviour
97 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
98 * @sched_priority task's static priority (SCHED_FIFO/RR)
99 * @sched_deadline representative of the task's deadline
100 * @sched_runtime representative of the task's runtime
101 * @sched_period representative of the task's period
102 *
103 * Given this task model, there are a multiplicity of scheduling algorithms
104 * and policies, that can be used to ensure all the tasks will make their
105 * timing constraints.
Dario Faggioliaab03e02013-11-28 11:14:43 +0100106 *
107 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
108 * only user of this new interface. More information about the algorithm
109 * available in the scheduling class file or in Documentation/.
Dario Faggiolid50dde52013-11-07 14:43:36 +0100110 */
111struct sched_attr {
112 u32 size;
113
114 u32 sched_policy;
115 u64 sched_flags;
116
117 /* SCHED_NORMAL, SCHED_BATCH */
118 s32 sched_nice;
119
120 /* SCHED_FIFO, SCHED_RR */
121 u32 sched_priority;
122
123 /* SCHED_DEADLINE */
124 u64 sched_runtime;
125 u64 sched_deadline;
126 u64 sched_period;
127};
128
Ingo Molnarc87e2832006-06-27 02:54:58 -0700129struct futex_pi_state;
Alexey Dobriyan286100a2008-01-25 21:08:34 +0100130struct robust_list_head;
Akinobu Mitabddd87c2010-02-23 08:55:42 +0100131struct bio_list;
Al Viro5ad4e532009-03-29 19:50:06 -0400132struct fs_struct;
Ingo Molnarcdd6c482009-09-21 12:02:48 +0200133struct perf_event_context;
Jens Axboe73c10102011-03-08 13:19:51 +0100134struct blk_plug;
Linus Torvaldsc4ad8f92014-02-05 12:54:53 -0800135struct filename;
Al Viro89076bc2015-05-12 08:29:38 -0400136struct nameidata;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700137
Davidlohr Bueso615d6e82014-04-07 15:37:25 -0700138#define VMACACHE_BITS 2
139#define VMACACHE_SIZE (1U << VMACACHE_BITS)
140#define VMACACHE_MASK (VMACACHE_SIZE - 1)
141
Linus Torvalds1da177e2005-04-16 15:20:36 -0700142/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700143 * These are the constant used to fake the fixed-point load-average
144 * counting. Some notes:
145 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
146 * a load-average precision of 10 bits integer + 11 bits fractional
147 * - if you want to count load-averages more often, you need more
148 * precision, or rounding will get you. With 2-second counting freq,
149 * the EXP_n values would be 1981, 2034 and 2043 if still using only
150 * 11 bit fractions.
151 */
152extern unsigned long avenrun[]; /* Load averages */
Thomas Gleixner2d024942009-05-02 20:08:52 +0200153extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700154
155#define FSHIFT 11 /* nr of bits of precision */
156#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
Linus Torvalds0c2043a2007-10-07 16:17:38 -0700157#define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700158#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
159#define EXP_5 2014 /* 1/exp(5sec/5min) */
160#define EXP_15 2037 /* 1/exp(5sec/15min) */
161
162#define CALC_LOAD(load,exp,n) \
163 load *= exp; \
164 load += n*(FIXED_1-exp); \
165 load >>= FSHIFT;
166
167extern unsigned long total_forks;
168extern int nr_threads;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700169DECLARE_PER_CPU(unsigned long, process_counts);
170extern int nr_processes(void);
171extern unsigned long nr_running(void);
Tim Chen2ee507c2014-07-31 10:29:48 -0700172extern bool single_task_running(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700173extern unsigned long nr_iowait(void);
Peter Zijlstra8c215bd2010-07-01 09:07:17 +0200174extern unsigned long nr_iowait_cpu(int cpu);
Mel Gorman372ba8c2014-08-06 14:19:21 +0100175extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
Arjan van de Ven69d25872009-09-21 17:04:08 -0700176
Peter Zijlstra0f004f52010-11-30 19:48:45 +0100177extern void calc_global_load(unsigned long ticks);
Peter Zijlstra3289bdb2015-04-14 13:19:42 +0200178
179#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
Frederic Weisbecker1f419062016-04-13 15:56:51 +0200180extern void cpu_load_update_nohz_start(void);
181extern void cpu_load_update_nohz_stop(void);
Peter Zijlstra3289bdb2015-04-14 13:19:42 +0200182#else
Frederic Weisbecker1f419062016-04-13 15:56:51 +0200183static inline void cpu_load_update_nohz_start(void) { }
184static inline void cpu_load_update_nohz_stop(void) { }
Peter Zijlstra3289bdb2015-04-14 13:19:42 +0200185#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700186
Paul E. McKenneyb637a322012-09-19 16:58:38 -0700187extern void dump_cpu_task(int cpu);
188
Ingo Molnar43ae34c2007-07-09 18:52:00 +0200189struct seq_file;
190struct cfs_rq;
Ingo Molnar4cf86d72007-10-15 17:00:14 +0200191struct task_group;
Ingo Molnar43ae34c2007-07-09 18:52:00 +0200192#ifdef CONFIG_SCHED_DEBUG
193extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
194extern void proc_sched_set_task(struct task_struct *p);
Ingo Molnar43ae34c2007-07-09 18:52:00 +0200195#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700196
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700197/*
198 * Task state bitmask. NOTE! These bits are also
199 * encoded in fs/proc/array.c: get_task_state().
200 *
201 * We have two separate sets of flags: task->state
202 * is about runnability, while task->exit_state are
203 * about the task exiting. Confusing, but this way
204 * modifying one set can't modify the other one by
205 * mistake.
206 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700207#define TASK_RUNNING 0
208#define TASK_INTERRUPTIBLE 1
209#define TASK_UNINTERRUPTIBLE 2
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500210#define __TASK_STOPPED 4
211#define __TASK_TRACED 8
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700212/* in tsk->exit_state */
Oleg Nesterovad866222014-04-07 15:38:46 -0700213#define EXIT_DEAD 16
214#define EXIT_ZOMBIE 32
Oleg Nesterovabd50b32014-04-07 15:38:42 -0700215#define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700216/* in tsk->state again */
Mike Galbraithaf927232007-10-15 17:00:13 +0200217#define TASK_DEAD 64
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500218#define TASK_WAKEKILL 128
Peter Zijlstrae9c84312009-09-15 14:43:03 +0200219#define TASK_WAKING 256
Thomas Gleixnerf2530dc2013-04-09 09:33:34 +0200220#define TASK_PARKED 512
Peter Zijlstra80ed87c2015-05-08 14:23:45 +0200221#define TASK_NOLOAD 1024
Peter Zijlstra7dc603c2016-06-16 13:29:28 +0200222#define TASK_NEW 2048
223#define TASK_STATE_MAX 4096
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500224
Peter Zijlstra7dc603c2016-06-16 13:29:28 +0200225#define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPNn"
Peter Zijlstra73342152009-12-17 13:16:27 +0100226
Peter Zijlstrae1781532009-12-17 13:16:30 +0100227extern char ___assert_task_state[1 - 2*!!(
228 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500229
230/* Convenience macros for the sake of set_task_state */
231#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
232#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
233#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700234
Peter Zijlstra80ed87c2015-05-08 14:23:45 +0200235#define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
236
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500237/* Convenience macros for the sake of wake_up */
238#define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500239#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500240
241/* get_task_state() */
242#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500243 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
Oleg Nesterov74e37202014-01-23 15:55:35 -0800244 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500245
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500246#define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
247#define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500248#define task_is_stopped_or_traced(task) \
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500249 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500250#define task_contributes_to_load(task) \
Nathan Lynche3c8ca82009-04-08 19:45:12 -0500251 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
Peter Zijlstra80ed87c2015-05-08 14:23:45 +0200252 (task->flags & PF_FROZEN) == 0 && \
253 (task->state & TASK_NOLOAD) == 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700254
Peter Zijlstra8eb23b92014-09-24 10:18:55 +0200255#ifdef CONFIG_DEBUG_ATOMIC_SLEEP
256
257#define __set_task_state(tsk, state_value) \
258 do { \
259 (tsk)->task_state_change = _THIS_IP_; \
260 (tsk)->state = (state_value); \
261 } while (0)
262#define set_task_state(tsk, state_value) \
263 do { \
264 (tsk)->task_state_change = _THIS_IP_; \
Peter Zijlstrab92b8b32015-05-12 10:51:55 +0200265 smp_store_mb((tsk)->state, (state_value)); \
Peter Zijlstra8eb23b92014-09-24 10:18:55 +0200266 } while (0)
267
268/*
269 * set_current_state() includes a barrier so that the write of current->state
270 * is correctly serialised wrt the caller's subsequent test of whether to
271 * actually sleep:
272 *
273 * set_current_state(TASK_UNINTERRUPTIBLE);
274 * if (do_i_need_to_sleep())
275 * schedule();
276 *
277 * If the caller does not need such serialisation then use __set_current_state()
278 */
279#define __set_current_state(state_value) \
280 do { \
281 current->task_state_change = _THIS_IP_; \
282 current->state = (state_value); \
283 } while (0)
284#define set_current_state(state_value) \
285 do { \
286 current->task_state_change = _THIS_IP_; \
Peter Zijlstrab92b8b32015-05-12 10:51:55 +0200287 smp_store_mb(current->state, (state_value)); \
Peter Zijlstra8eb23b92014-09-24 10:18:55 +0200288 } while (0)
289
290#else
291
Linus Torvalds1da177e2005-04-16 15:20:36 -0700292#define __set_task_state(tsk, state_value) \
293 do { (tsk)->state = (state_value); } while (0)
294#define set_task_state(tsk, state_value) \
Peter Zijlstrab92b8b32015-05-12 10:51:55 +0200295 smp_store_mb((tsk)->state, (state_value))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700296
Andrew Morton498d0c52005-09-13 01:25:14 -0700297/*
298 * set_current_state() includes a barrier so that the write of current->state
299 * is correctly serialised wrt the caller's subsequent test of whether to
300 * actually sleep:
301 *
302 * set_current_state(TASK_UNINTERRUPTIBLE);
303 * if (do_i_need_to_sleep())
304 * schedule();
305 *
306 * If the caller does not need such serialisation then use __set_current_state()
307 */
Peter Zijlstra8eb23b92014-09-24 10:18:55 +0200308#define __set_current_state(state_value) \
Linus Torvalds1da177e2005-04-16 15:20:36 -0700309 do { current->state = (state_value); } while (0)
Peter Zijlstra8eb23b92014-09-24 10:18:55 +0200310#define set_current_state(state_value) \
Peter Zijlstrab92b8b32015-05-12 10:51:55 +0200311 smp_store_mb(current->state, (state_value))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700312
Peter Zijlstra8eb23b92014-09-24 10:18:55 +0200313#endif
314
Linus Torvalds1da177e2005-04-16 15:20:36 -0700315/* Task command name length */
316#define TASK_COMM_LEN 16
317
Linus Torvalds1da177e2005-04-16 15:20:36 -0700318#include <linux/spinlock.h>
319
320/*
321 * This serializes "schedule()" and also protects
322 * the run-queue from deletions/modifications (but
323 * _adding_ to the beginning of the run-queue has
324 * a separate lock).
325 */
326extern rwlock_t tasklist_lock;
327extern spinlock_t mmlist_lock;
328
Ingo Molnar36c8b582006-07-03 00:25:41 -0700329struct task_struct;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700330
Paul E. McKenneydb1466b2010-03-03 07:46:56 -0800331#ifdef CONFIG_PROVE_RCU
332extern int lockdep_tasklist_lock_is_held(void);
333#endif /* #ifdef CONFIG_PROVE_RCU */
334
Linus Torvalds1da177e2005-04-16 15:20:36 -0700335extern void sched_init(void);
336extern void sched_init_smp(void);
Harvey Harrison2d07b252008-02-15 09:56:34 -0800337extern asmlinkage void schedule_tail(struct task_struct *prev);
Ingo Molnar36c8b582006-07-03 00:25:41 -0700338extern void init_idle(struct task_struct *idle, int cpu);
Ingo Molnar1df21052007-07-09 18:51:58 +0200339extern void init_idle_bootup_task(struct task_struct *idle);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700340
Rik van Riel3fa08182015-03-09 12:12:07 -0400341extern cpumask_var_t cpu_isolated_map;
342
Andrew Morton89f19f02009-09-19 11:55:44 -0700343extern int runqueue_is_locked(int cpu);
Ingo Molnar017730c2008-05-12 21:20:52 +0200344
Frederic Weisbecker3451d022011-08-10 23:21:01 +0200345#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
Alex Shic1cc0172012-09-10 15:10:58 +0800346extern void nohz_balance_enter_idle(int cpu);
Suresh Siddha69e1e812011-12-01 17:07:33 -0800347extern void set_cpu_sd_state_idle(void);
Thomas Gleixnerbc7a34b2015-05-26 22:50:33 +0000348extern int get_nohz_timer_target(void);
Siddha, Suresh B46cb4b72007-05-08 00:32:51 -0700349#else
Alex Shic1cc0172012-09-10 15:10:58 +0800350static inline void nohz_balance_enter_idle(int cpu) { }
Peter Zijlstrafdaabd82011-12-06 12:47:55 +0100351static inline void set_cpu_sd_state_idle(void) { }
Siddha, Suresh B46cb4b72007-05-08 00:32:51 -0700352#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700353
Ingo Molnare59e2ae2006-12-06 20:35:59 -0800354/*
Ingo Molnar39bc89f2007-04-25 20:50:03 -0700355 * Only dump TASK_* tasks. (0 for all tasks)
Ingo Molnare59e2ae2006-12-06 20:35:59 -0800356 */
357extern void show_state_filter(unsigned long state_filter);
358
359static inline void show_state(void)
360{
Ingo Molnar39bc89f2007-04-25 20:50:03 -0700361 show_state_filter(0);
Ingo Molnare59e2ae2006-12-06 20:35:59 -0800362}
363
Linus Torvalds1da177e2005-04-16 15:20:36 -0700364extern void show_regs(struct pt_regs *);
365
366/*
367 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
368 * task), SP is the stack pointer of the first frame that should be shown in the back
369 * trace (or NULL if the entire call-chain of the task should be shown).
370 */
371extern void show_stack(struct task_struct *task, unsigned long *sp);
372
Linus Torvalds1da177e2005-04-16 15:20:36 -0700373extern void cpu_init (void);
374extern void trap_init(void);
375extern void update_process_times(int user);
376extern void scheduler_tick(void);
Thomas Gleixner9cf72432016-03-10 12:54:09 +0100377extern int sched_cpu_starting(unsigned int cpu);
Thomas Gleixner40190a72016-03-10 12:54:13 +0100378extern int sched_cpu_activate(unsigned int cpu);
379extern int sched_cpu_deactivate(unsigned int cpu);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700380
Thomas Gleixnerf2785dd2016-03-10 12:54:18 +0100381#ifdef CONFIG_HOTPLUG_CPU
382extern int sched_cpu_dying(unsigned int cpu);
383#else
384# define sched_cpu_dying NULL
385#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700386
Ingo Molnar82a1fcb2008-01-25 21:08:02 +0100387extern void sched_show_task(struct task_struct *p);
388
Frederic Weisbecker19cc36c2010-05-13 02:30:49 +0200389#ifdef CONFIG_LOCKUP_DETECTOR
Tejun Heo03e0d462015-12-08 11:28:04 -0500390extern void touch_softlockup_watchdog_sched(void);
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700391extern void touch_softlockup_watchdog(void);
Jason Wesseld6ad3e22010-01-27 16:25:22 -0600392extern void touch_softlockup_watchdog_sync(void);
Jeremy Fitzhardinge04c91672007-05-08 00:28:05 -0700393extern void touch_all_softlockup_watchdogs(void);
Don Zickus332fbdb2010-05-07 17:11:45 -0400394extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
395 void __user *buffer,
396 size_t *lenp, loff_t *ppos);
Ingo Molnar9c44bc02008-05-12 21:21:04 +0200397extern unsigned int softlockup_panic;
Don Zickusac1f5912015-11-05 18:44:44 -0800398extern unsigned int hardlockup_panic;
Peter Zijlstra004417a2010-11-25 18:38:29 +0100399void lockup_detector_init(void);
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700400#else
Tejun Heo03e0d462015-12-08 11:28:04 -0500401static inline void touch_softlockup_watchdog_sched(void)
402{
403}
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700404static inline void touch_softlockup_watchdog(void)
405{
406}
Jason Wesseld6ad3e22010-01-27 16:25:22 -0600407static inline void touch_softlockup_watchdog_sync(void)
408{
409}
Jeremy Fitzhardinge04c91672007-05-08 00:28:05 -0700410static inline void touch_all_softlockup_watchdogs(void)
411{
412}
Peter Zijlstra004417a2010-11-25 18:38:29 +0100413static inline void lockup_detector_init(void)
414{
415}
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700416#endif
417
Marcelo Tosatti8b414522013-10-11 21:39:26 -0300418#ifdef CONFIG_DETECT_HUNG_TASK
419void reset_hung_task_detector(void);
420#else
421static inline void reset_hung_task_detector(void)
422{
423}
424#endif
425
Linus Torvalds1da177e2005-04-16 15:20:36 -0700426/* Attach to any functions which should be ignored in wchan output. */
427#define __sched __attribute__((__section__(".sched.text")))
Ingo Molnardeaf2222007-11-28 15:52:56 +0100428
429/* Linker adds these: start and end of __sched functions */
430extern char __sched_text_start[], __sched_text_end[];
431
Linus Torvalds1da177e2005-04-16 15:20:36 -0700432/* Is this address in the __sched functions? */
433extern int in_sched_functions(unsigned long addr);
434
435#define MAX_SCHEDULE_TIMEOUT LONG_MAX
Harvey Harrisonb3c97522008-02-13 15:03:15 -0800436extern signed long schedule_timeout(signed long timeout);
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700437extern signed long schedule_timeout_interruptible(signed long timeout);
Matthew Wilcox294d5cc2007-12-06 11:59:46 -0500438extern signed long schedule_timeout_killable(signed long timeout);
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700439extern signed long schedule_timeout_uninterruptible(signed long timeout);
Andrew Morton69b27ba2016-03-25 14:20:21 -0700440extern signed long schedule_timeout_idle(signed long timeout);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700441asmlinkage void schedule(void);
Thomas Gleixnerc5491ea2011-03-21 12:09:35 +0100442extern void schedule_preempt_disabled(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700443
NeilBrown9cff8ad2015-02-13 15:49:17 +1100444extern long io_schedule_timeout(long timeout);
445
446static inline void io_schedule(void)
447{
448 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
449}
450
Peter Zijlstra9af65282016-09-13 18:37:29 +0200451void __noreturn do_task_dead(void);
452
Serge E. Hallynab516012006-10-02 02:18:06 -0700453struct nsproxy;
Cedric Le Goateracce2922007-07-15 23:40:59 -0700454struct user_namespace;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700455
David Howellsefc1a3b2010-01-15 17:01:35 -0800456#ifdef CONFIG_MMU
457extern void arch_pick_mmap_layout(struct mm_struct *mm);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700458extern unsigned long
459arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
460 unsigned long, unsigned long);
461extern unsigned long
462arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
463 unsigned long len, unsigned long pgoff,
464 unsigned long flags);
David Howellsefc1a3b2010-01-15 17:01:35 -0800465#else
466static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
467#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700468
Kees Cookd049f742013-11-12 15:11:17 -0800469#define SUID_DUMP_DISABLE 0 /* No setuid dumping */
470#define SUID_DUMP_USER 1 /* Dump as user of process */
471#define SUID_DUMP_ROOT 2 /* Dump as root */
472
Kawai, Hidehiro6c5d5232007-07-19 01:48:27 -0700473/* mm flags */
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700474
Oleg Nesterov7288e112014-01-23 15:55:32 -0800475/* for SUID_DUMP_* above */
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700476#define MMF_DUMPABLE_BITS 2
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700477#define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700478
Oleg Nesterov942be382014-01-23 15:55:34 -0800479extern void set_dumpable(struct mm_struct *mm, int value);
480/*
481 * This returns the actual value of the suid_dumpable flag. For things
482 * that are using this for checking for privilege transitions, it must
483 * test against SUID_DUMP_USER rather than treating it as a boolean
484 * value.
485 */
486static inline int __get_dumpable(unsigned long mm_flags)
487{
488 return mm_flags & MMF_DUMPABLE_MASK;
489}
490
491static inline int get_dumpable(struct mm_struct *mm)
492{
493 return __get_dumpable(mm->flags);
494}
495
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700496/* coredump filter bits */
497#define MMF_DUMP_ANON_PRIVATE 2
498#define MMF_DUMP_ANON_SHARED 3
499#define MMF_DUMP_MAPPED_PRIVATE 4
500#define MMF_DUMP_MAPPED_SHARED 5
Roland McGrath82df3972007-10-16 23:27:02 -0700501#define MMF_DUMP_ELF_HEADERS 6
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700502#define MMF_DUMP_HUGETLB_PRIVATE 7
503#define MMF_DUMP_HUGETLB_SHARED 8
Ross Zwisler50378352015-10-05 16:33:36 -0600504#define MMF_DUMP_DAX_PRIVATE 9
505#define MMF_DUMP_DAX_SHARED 10
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700506
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700507#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
Ross Zwisler50378352015-10-05 16:33:36 -0600508#define MMF_DUMP_FILTER_BITS 9
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700509#define MMF_DUMP_FILTER_MASK \
510 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
511#define MMF_DUMP_FILTER_DEFAULT \
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700512 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
Roland McGrath656eb2c2008-10-18 20:28:23 -0700513 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
514
515#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
516# define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
517#else
518# define MMF_DUMP_MASK_DEFAULT_ELF 0
519#endif
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700520 /* leave room for more dump flags */
521#define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
Andrea Arcangeliba761492011-01-13 15:46:58 -0800522#define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
Konstantin Khlebnikovbafb2822012-06-07 14:21:11 -0700523#define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700524
Oleg Nesterov9f68f672012-08-19 16:15:09 +0200525#define MMF_HAS_UPROBES 19 /* has uprobes */
526#define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
Michal Hocko862e3072016-10-07 16:58:57 -0700527#define MMF_OOM_SKIP 21 /* mm is of no interest for the OOM killer */
Michal Hocko3f70dc32016-10-07 16:59:06 -0700528#define MMF_UNSTABLE 22 /* mm is unstable for copy_from_user */
Aaron Lu6fcb52a2016-10-07 17:00:08 -0700529#define MMF_HUGE_ZERO_PAGE 23 /* mm has ever used the global huge zero page */
Tetsuo Handa7f382992019-02-01 14:20:31 -0800530#define MMF_OOM_REAP_QUEUED 26 /* mm was queued for oom_reaper */
Oleg Nesterovf8ac4ec2012-08-08 17:11:42 +0200531
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700532#define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
Kawai, Hidehiro6c5d5232007-07-19 01:48:27 -0700533
Linus Torvalds1da177e2005-04-16 15:20:36 -0700534struct sighand_struct {
535 atomic_t count;
536 struct k_sigaction action[_NSIG];
537 spinlock_t siglock;
Davide Libenzib8fceee2007-09-20 12:40:16 -0700538 wait_queue_head_t signalfd_wqh;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700539};
540
KaiGai Kohei0e464812006-06-25 05:49:24 -0700541struct pacct_struct {
KaiGai Koheif6ec29a2006-06-25 05:49:25 -0700542 int ac_flag;
543 long ac_exitcode;
KaiGai Kohei0e464812006-06-25 05:49:24 -0700544 unsigned long ac_mem;
KaiGai Kohei77787bf2006-06-25 05:49:26 -0700545 cputime_t ac_utime, ac_stime;
546 unsigned long ac_minflt, ac_majflt;
KaiGai Kohei0e464812006-06-25 05:49:24 -0700547};
548
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200549struct cpu_itimer {
550 cputime_t expires;
551 cputime_t incr;
Stanislaw Gruszka8356b5f2009-07-29 12:15:27 +0200552 u32 error;
553 u32 incr_error;
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200554};
555
Frank Mayharf06febc2008-09-12 09:54:39 -0700556/**
Peter Zijlstra9d7fb042015-06-30 11:30:54 +0200557 * struct prev_cputime - snaphsot of system and user cputime
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +0100558 * @utime: time spent in user mode
559 * @stime: time spent in system mode
Peter Zijlstra9d7fb042015-06-30 11:30:54 +0200560 * @lock: protects the above two fields
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +0100561 *
Peter Zijlstra9d7fb042015-06-30 11:30:54 +0200562 * Stores previous user/system time values such that we can guarantee
563 * monotonicity.
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +0100564 */
Peter Zijlstra9d7fb042015-06-30 11:30:54 +0200565struct prev_cputime {
566#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +0100567 cputime_t utime;
568 cputime_t stime;
Peter Zijlstra9d7fb042015-06-30 11:30:54 +0200569 raw_spinlock_t lock;
570#endif
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +0100571};
572
Peter Zijlstra9d7fb042015-06-30 11:30:54 +0200573static inline void prev_cputime_init(struct prev_cputime *prev)
574{
575#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
576 prev->utime = prev->stime = 0;
577 raw_spin_lock_init(&prev->lock);
578#endif
579}
580
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +0100581/**
Frank Mayharf06febc2008-09-12 09:54:39 -0700582 * struct task_cputime - collected CPU time counts
583 * @utime: time spent in user mode, in &cputime_t units
584 * @stime: time spent in kernel mode, in &cputime_t units
585 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
Ingo Molnar5ce73a42008-09-14 17:11:46 +0200586 *
Peter Zijlstra9d7fb042015-06-30 11:30:54 +0200587 * This structure groups together three kinds of CPU time that are tracked for
588 * threads and thread groups. Most things considering CPU time want to group
589 * these counts together and treat all three of them in parallel.
Frank Mayharf06febc2008-09-12 09:54:39 -0700590 */
591struct task_cputime {
592 cputime_t utime;
593 cputime_t stime;
594 unsigned long long sum_exec_runtime;
595};
Peter Zijlstra9d7fb042015-06-30 11:30:54 +0200596
Frank Mayharf06febc2008-09-12 09:54:39 -0700597/* Alternate field names when used to cache expirations. */
Frank Mayharf06febc2008-09-12 09:54:39 -0700598#define virt_exp utime
Peter Zijlstra9d7fb042015-06-30 11:30:54 +0200599#define prof_exp stime
Frank Mayharf06febc2008-09-12 09:54:39 -0700600#define sched_exp sum_exec_runtime
601
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100602#define INIT_CPUTIME \
603 (struct task_cputime) { \
Martin Schwidefsky64861632011-12-15 14:56:09 +0100604 .utime = 0, \
605 .stime = 0, \
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100606 .sum_exec_runtime = 0, \
607 }
608
Jason Low971e8a982015-04-28 13:00:23 -0700609/*
610 * This is the atomic variant of task_cputime, which can be used for
611 * storing and updating task_cputime statistics without locking.
612 */
613struct task_cputime_atomic {
614 atomic64_t utime;
615 atomic64_t stime;
616 atomic64_t sum_exec_runtime;
617};
618
619#define INIT_CPUTIME_ATOMIC \
620 (struct task_cputime_atomic) { \
621 .utime = ATOMIC64_INIT(0), \
622 .stime = ATOMIC64_INIT(0), \
623 .sum_exec_runtime = ATOMIC64_INIT(0), \
624 }
625
Peter Zijlstra609ca062015-09-28 17:52:18 +0200626#define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
Peter Zijlstraa233f112013-09-23 19:04:26 +0200627
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200628/*
Peter Zijlstra87dcbc02015-09-28 17:45:40 +0200629 * Disable preemption until the scheduler is running -- use an unconditional
630 * value so that it also works on !PREEMPT_COUNT kernels.
Peter Zijlstrad86ee482009-07-10 14:57:57 +0200631 *
Peter Zijlstra87dcbc02015-09-28 17:45:40 +0200632 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200633 */
Peter Zijlstra87dcbc02015-09-28 17:45:40 +0200634#define INIT_PREEMPT_COUNT PREEMPT_OFFSET
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200635
Peter Zijlstra609ca062015-09-28 17:52:18 +0200636/*
637 * Initial preempt_count value; reflects the preempt_count schedule invariant
638 * which states that during context switches:
639 *
640 * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
641 *
642 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
643 * Note: See finish_task_switch().
644 */
645#define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100646
Frank Mayharf06febc2008-09-12 09:54:39 -0700647/**
648 * struct thread_group_cputimer - thread group interval timer counts
Jason Low920ce392015-05-08 14:31:50 -0700649 * @cputime_atomic: atomic thread group interval timers.
Jason Lowd5c373e2015-10-14 12:07:55 -0700650 * @running: true when there are timers running and
651 * @cputime_atomic receives updates.
Jason Lowc8d75aa2015-10-14 12:07:56 -0700652 * @checking_timer: true when a thread in the group is in the
653 * process of checking for thread group timers.
Frank Mayharf06febc2008-09-12 09:54:39 -0700654 *
655 * This structure contains the version of task_cputime, above, that is
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100656 * used for thread group CPU timer calculations.
Frank Mayharf06febc2008-09-12 09:54:39 -0700657 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100658struct thread_group_cputimer {
Jason Low71107442015-04-28 13:00:24 -0700659 struct task_cputime_atomic cputime_atomic;
Jason Lowd5c373e2015-10-14 12:07:55 -0700660 bool running;
Jason Lowc8d75aa2015-10-14 12:07:56 -0700661 bool checking_timer;
Frank Mayharf06febc2008-09-12 09:54:39 -0700662};
Frank Mayharf06febc2008-09-12 09:54:39 -0700663
Ben Blum4714d1d2011-05-26 16:25:18 -0700664#include <linux/rwsem.h>
Mike Galbraith5091faa2010-11-30 14:18:03 +0100665struct autogroup;
666
Linus Torvalds1da177e2005-04-16 15:20:36 -0700667/*
Jonathan Neuschäfere815f0a2011-03-21 20:24:47 +0100668 * NOTE! "signal_struct" does not have its own
Linus Torvalds1da177e2005-04-16 15:20:36 -0700669 * locking, because a shared signal_struct always
670 * implies a shared sighand_struct, so locking
671 * sighand_struct is always a proper superset of
672 * the locking of signal_struct.
673 */
674struct signal_struct {
Oleg Nesterovea6d2902010-05-26 14:43:16 -0700675 atomic_t sigcnt;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700676 atomic_t live;
Oleg Nesterovb3ac0222010-05-26 14:43:24 -0700677 int nr_threads;
Oleg Nesterov0c740d02014-01-21 15:49:56 -0800678 struct list_head thread_head;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700679
680 wait_queue_head_t wait_chldexit; /* for wait4() */
681
682 /* current thread group signal load-balancing target: */
Ingo Molnar36c8b582006-07-03 00:25:41 -0700683 struct task_struct *curr_target;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700684
685 /* shared signal handling: */
686 struct sigpending shared_pending;
687
688 /* thread group exit support */
689 int group_exit_code;
690 /* overloaded:
691 * - notify group_exit_task when ->count is equal to notify_count
692 * - everyone except group_exit_task is stopped during signal delivery
693 * of fatal signals, group_exit_task processes the signal.
694 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700695 int notify_count;
Richard Kennedy07dd20e2008-08-01 13:18:04 +0100696 struct task_struct *group_exit_task;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700697
698 /* thread group stop support, overloads group_exit_code too */
699 int group_stop_count;
700 unsigned int flags; /* see SIGNAL_* flags below */
701
Lennart Poetteringebec18a2012-03-23 15:01:54 -0700702 /*
703 * PR_SET_CHILD_SUBREAPER marks a process, like a service
704 * manager, to re-parent orphan (double-forking) child processes
705 * to this process instead of 'init'. The service manager is
706 * able to receive SIGCHLD signals and is able to investigate
707 * the process until it calls wait(). All children of this
708 * process will inherit a flag if they should look for a
709 * child_subreaper process at exit.
710 */
711 unsigned int is_child_subreaper:1;
712 unsigned int has_child_subreaper:1;
713
Linus Torvalds1da177e2005-04-16 15:20:36 -0700714 /* POSIX.1b Interval Timers */
Pavel Emelyanov5ed67f02013-03-11 13:12:21 +0400715 int posix_timer_id;
716 struct list_head posix_timers;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700717
718 /* ITIMER_REAL timer for the process */
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800719 struct hrtimer real_timer;
Oleg Nesterovfea9d172008-02-08 04:19:19 -0800720 struct pid *leader_pid;
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800721 ktime_t it_real_incr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700722
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200723 /*
724 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
725 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
726 * values are defined to 0 and 1 respectively
727 */
728 struct cpu_itimer it[2];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700729
Frank Mayharf06febc2008-09-12 09:54:39 -0700730 /*
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100731 * Thread group totals for process CPU timers.
732 * See thread_group_cputimer(), et al, for details.
Frank Mayharf06febc2008-09-12 09:54:39 -0700733 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100734 struct thread_group_cputimer cputimer;
Frank Mayharf06febc2008-09-12 09:54:39 -0700735
736 /* Earliest-expiration cache. */
737 struct task_cputime cputime_expires;
738
Frederic Weisbeckerd027d452015-06-07 15:54:30 +0200739#ifdef CONFIG_NO_HZ_FULL
Frederic Weisbeckerf009a7a2016-03-24 15:38:00 +0100740 atomic_t tick_dep_mask;
Frederic Weisbeckerd027d452015-06-07 15:54:30 +0200741#endif
742
Frank Mayharf06febc2008-09-12 09:54:39 -0700743 struct list_head cpu_timers[3];
744
Eric W. Biedermanab521dc2007-02-12 00:53:00 -0800745 struct pid *tty_old_pgrp;
Cedric Le Goater1ec320a2006-12-08 02:37:55 -0800746
Linus Torvalds1da177e2005-04-16 15:20:36 -0700747 /* boolean value for session group leader */
748 int leader;
749
750 struct tty_struct *tty; /* NULL if no tty */
751
Mike Galbraith5091faa2010-11-30 14:18:03 +0100752#ifdef CONFIG_SCHED_AUTOGROUP
753 struct autogroup *autogroup;
754#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700755 /*
756 * Cumulative resource counters for dead threads in the group,
757 * and for reaped dead child processes forked by this group.
758 * Live threads maintain their own counters and add to these
759 * in __exit_signal, except for the group leader.
760 */
Rik van Riele78c3492014-08-16 13:40:10 -0400761 seqlock_t stats_lock;
Peter Zijlstra32bd6712009-02-05 12:24:15 +0100762 cputime_t utime, stime, cutime, cstime;
Laurent Vivier9ac52312007-10-15 17:00:19 +0200763 cputime_t gtime;
764 cputime_t cgtime;
Peter Zijlstra9d7fb042015-06-30 11:30:54 +0200765 struct prev_cputime prev_cputime;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700766 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
767 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
Eric Dumazet6eaeeab2007-05-10 22:22:37 -0700768 unsigned long inblock, oublock, cinblock, coublock;
Jiri Pirko1f102062009-09-22 16:44:10 -0700769 unsigned long maxrss, cmaxrss;
Andrea Righi940389b2008-07-28 00:48:12 +0200770 struct task_io_accounting ioac;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700771
772 /*
Peter Zijlstra32bd6712009-02-05 12:24:15 +0100773 * Cumulative ns of schedule CPU time fo dead threads in the
774 * group, not including a zombie group leader, (This only differs
775 * from jiffies_to_ns(utime + stime) if sched_clock uses something
776 * other than jiffies.)
777 */
778 unsigned long long sum_sched_runtime;
779
780 /*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700781 * We don't bother to synchronize most readers of this at all,
782 * because there is no reader checking a limit that actually needs
783 * to get both rlim_cur and rlim_max atomically, and either one
784 * alone is a single word that can safely be read normally.
785 * getrlimit/setrlimit use task_lock(current->group_leader) to
786 * protect this instead of the siglock, because they really
787 * have no need to disable irqs.
788 */
789 struct rlimit rlim[RLIM_NLIMITS];
790
KaiGai Kohei0e464812006-06-25 05:49:24 -0700791#ifdef CONFIG_BSD_PROCESS_ACCT
792 struct pacct_struct pacct; /* per-process accounting information */
793#endif
Shailabh Nagarad4ecbc2006-07-14 00:24:44 -0700794#ifdef CONFIG_TASKSTATS
Shailabh Nagarad4ecbc2006-07-14 00:24:44 -0700795 struct taskstats *stats;
796#endif
Miloslav Trmac522ed772007-07-15 23:40:56 -0700797#ifdef CONFIG_AUDIT
798 unsigned audit_tty;
799 struct tty_audit_buf *tty_audit_buf;
800#endif
KOSAKI Motohiro28b83c52009-09-21 17:03:13 -0700801
Tetsuo Handac96fc2d2016-05-23 16:23:57 -0700802 /*
803 * Thread is the potential origin of an oom condition; kill first on
804 * oom
805 */
806 bool oom_flag_origin;
David Rientjesa9c58b902012-12-11 16:02:54 -0800807 short oom_score_adj; /* OOM kill score adjustment */
808 short oom_score_adj_min; /* OOM kill score adjustment min value.
809 * Only settable by CAP_SYS_RESOURCE. */
Michal Hocko26db62f2016-10-07 16:58:51 -0700810 struct mm_struct *oom_mm; /* recorded mm when the thread group got
811 * killed by the oom killer */
KOSAKI Motohiro9b1bf122010-10-27 15:34:08 -0700812
813 struct mutex cred_guard_mutex; /* guard against foreign influences on
814 * credential calculations
815 * (notably. ptrace) */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700816};
817
818/*
819 * Bits in flags field of signal_struct.
820 */
821#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
Oleg Nesterovee77f072011-04-01 20:12:38 +0200822#define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
823#define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
Oleg Nesterov403bad72013-04-30 15:28:10 -0700824#define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
Oleg Nesterove4420552008-04-30 00:52:44 -0700825/*
826 * Pending notifications to parent.
827 */
828#define SIGNAL_CLD_STOPPED 0x00000010
829#define SIGNAL_CLD_CONTINUED 0x00000020
830#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700831
Oleg Nesterovfae5fa42008-04-30 00:53:03 -0700832#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
833
Jamie Iles916a05b2017-01-10 16:57:54 -0800834#define SIGNAL_STOP_MASK (SIGNAL_CLD_MASK | SIGNAL_STOP_STOPPED | \
835 SIGNAL_STOP_CONTINUED)
836
837static inline void signal_set_stop_flags(struct signal_struct *sig,
838 unsigned int flags)
839{
840 WARN_ON(sig->flags & (SIGNAL_GROUP_EXIT|SIGNAL_GROUP_COREDUMP));
841 sig->flags = (sig->flags & ~SIGNAL_STOP_MASK) | flags;
842}
843
Oleg Nesteroved5d2ca2008-02-04 22:27:24 -0800844/* If true, all threads except ->group_exit_task have pending SIGKILL */
845static inline int signal_group_exit(const struct signal_struct *sig)
846{
847 return (sig->flags & SIGNAL_GROUP_EXIT) ||
848 (sig->group_exit_task != NULL);
849}
850
Linus Torvalds1da177e2005-04-16 15:20:36 -0700851/*
852 * Some day this will be a full-fledged user tracking system..
853 */
854struct user_struct {
855 atomic_t __count; /* reference count */
856 atomic_t processes; /* How many processes does this user have? */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700857 atomic_t sigpending; /* How many pending signals does this user have? */
Amy Griffis2d9048e2006-06-01 13:10:59 -0700858#ifdef CONFIG_INOTIFY_USER
Robert Love0eeca282005-07-12 17:06:03 -0400859 atomic_t inotify_watches; /* How many inotify watches does this user have? */
860 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
861#endif
Eric Paris4afeff82010-10-28 17:21:58 -0400862#ifdef CONFIG_FANOTIFY
863 atomic_t fanotify_listeners;
864#endif
Davide Libenzi7ef99642008-12-01 13:13:55 -0800865#ifdef CONFIG_EPOLL
Robin Holt52bd19f72011-01-12 17:00:01 -0800866 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
Davide Libenzi7ef99642008-12-01 13:13:55 -0800867#endif
Alexey Dobriyan970a8642007-10-16 23:30:09 -0700868#ifdef CONFIG_POSIX_MQUEUE
Linus Torvalds1da177e2005-04-16 15:20:36 -0700869 /* protected by mq_lock */
870 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
Alexey Dobriyan970a8642007-10-16 23:30:09 -0700871#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700872 unsigned long locked_shm; /* How many pages of mlocked shm ? */
willy tarreau712f4aa2016-01-10 07:54:56 +0100873 unsigned long unix_inflight; /* How many files in flight in unix sockets */
Willy Tarreau759c0112016-01-18 16:36:09 +0100874 atomic_long_t pipe_bufs; /* how many pages are allocated in pipe buffers */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700875
876#ifdef CONFIG_KEYS
877 struct key *uid_keyring; /* UID specific keyring */
878 struct key *session_keyring; /* UID's default session keyring */
879#endif
880
881 /* Hash table maintenance information */
Pavel Emelyanov735de222007-09-18 22:46:44 -0700882 struct hlist_node uidhash_node;
Eric W. Biederman7b44ab92011-11-16 23:20:58 -0800883 kuid_t uid;
Srivatsa Vaddagiri24e377a2007-10-15 17:00:09 +0200884
Alexei Starovoitovaaac3ba2015-10-07 22:23:22 -0700885#if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
Peter Zijlstra789f90f2009-05-15 15:19:27 +0200886 atomic_long_t locked_vm;
887#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700888};
889
Kay Sieverseb41d942007-11-02 13:47:53 +0100890extern int uids_sysfs_init(void);
Dhaval Giani5cb350b2007-10-15 17:00:14 +0200891
Eric W. Biederman7b44ab92011-11-16 23:20:58 -0800892extern struct user_struct *find_user(kuid_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700893
894extern struct user_struct root_user;
895#define INIT_USER (&root_user)
896
David Howellsb6dff3e2008-11-14 10:39:16 +1100897
Linus Torvalds1da177e2005-04-16 15:20:36 -0700898struct backing_dev_info;
899struct reclaim_state;
900
Naveen N. Raof6db8342015-06-25 23:53:37 +0530901#ifdef CONFIG_SCHED_INFO
Linus Torvalds1da177e2005-04-16 15:20:36 -0700902struct sched_info {
903 /* cumulative counters */
Ingo Molnar2d723762007-10-15 17:00:12 +0200904 unsigned long pcount; /* # of times run on this cpu */
Ken Chen9c2c4802008-12-16 23:41:22 -0800905 unsigned long long run_delay; /* time spent waiting on a runqueue */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700906
907 /* timestamps */
Balbir Singh172ba842007-07-09 18:52:00 +0200908 unsigned long long last_arrival,/* when we last ran on a cpu */
909 last_queued; /* when we were last queued to run */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700910};
Naveen N. Raof6db8342015-06-25 23:53:37 +0530911#endif /* CONFIG_SCHED_INFO */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700912
Shailabh Nagarca74e922006-07-14 00:24:36 -0700913#ifdef CONFIG_TASK_DELAY_ACCT
914struct task_delay_info {
915 spinlock_t lock;
916 unsigned int flags; /* Private per-task flags */
917
918 /* For each stat XXX, add following, aligned appropriately
919 *
920 * struct timespec XXX_start, XXX_end;
921 * u64 XXX_delay;
922 * u32 XXX_count;
923 *
924 * Atomicity of updates to XXX_delay, XXX_count protected by
925 * single lock above (split into XXX_lock if contention is an issue).
926 */
Shailabh Nagar0ff92242006-07-14 00:24:37 -0700927
928 /*
929 * XXX_count is incremented on every XXX operation, the delay
930 * associated with the operation is added to XXX_delay.
931 * XXX_delay contains the accumulated delay time in nanoseconds.
932 */
Thomas Gleixner9667a232014-07-16 21:04:35 +0000933 u64 blkio_start; /* Shared by blkio, swapin */
Shailabh Nagar0ff92242006-07-14 00:24:37 -0700934 u64 blkio_delay; /* wait for sync block io completion */
935 u64 swapin_delay; /* wait for swapin block io completion */
936 u32 blkio_count; /* total count of the number of sync block */
937 /* io operations performed */
938 u32 swapin_count; /* total count of the number of swapin block */
939 /* io operations performed */
Keika Kobayashi873b4772008-07-25 01:48:52 -0700940
Thomas Gleixner9667a232014-07-16 21:04:35 +0000941 u64 freepages_start;
Keika Kobayashi873b4772008-07-25 01:48:52 -0700942 u64 freepages_delay; /* wait for memory reclaim */
943 u32 freepages_count; /* total count of memory reclaim */
Shailabh Nagarca74e922006-07-14 00:24:36 -0700944};
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700945#endif /* CONFIG_TASK_DELAY_ACCT */
946
947static inline int sched_info_on(void)
948{
949#ifdef CONFIG_SCHEDSTATS
950 return 1;
951#elif defined(CONFIG_TASK_DELAY_ACCT)
952 extern int delayacct_on;
953 return delayacct_on;
954#else
955 return 0;
Shailabh Nagarca74e922006-07-14 00:24:36 -0700956#endif
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700957}
Shailabh Nagarca74e922006-07-14 00:24:36 -0700958
Mel Gormancb251762016-02-05 09:08:36 +0000959#ifdef CONFIG_SCHEDSTATS
960void force_schedstat_enabled(void);
961#endif
962
Ingo Molnard15bcfd2007-07-09 18:51:57 +0200963enum cpu_idle_type {
964 CPU_IDLE,
965 CPU_NOT_IDLE,
966 CPU_NEWLY_IDLE,
967 CPU_MAX_IDLE_TYPES
Linus Torvalds1da177e2005-04-16 15:20:36 -0700968};
969
970/*
Yuyang Du6ecdd742016-04-05 12:12:26 +0800971 * Integer metrics need fixed point arithmetic, e.g., sched/fair
972 * has a few: load, load_avg, util_avg, freq, and capacity.
973 *
974 * We define a basic fixed point arithmetic range, and then formalize
975 * all these metrics based on that basic range.
976 */
977# define SCHED_FIXEDPOINT_SHIFT 10
978# define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
979
980/*
Nicolas Pitreca8ce3d2014-05-26 18:19:39 -0400981 * Increase resolution of cpu_capacity calculations
Nikhil Rao1399fa72011-05-18 10:09:39 -0700982 */
Yuyang Du6ecdd742016-04-05 12:12:26 +0800983#define SCHED_CAPACITY_SHIFT SCHED_FIXEDPOINT_SHIFT
Nicolas Pitreca8ce3d2014-05-26 18:19:39 -0400984#define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700985
Nikhil Rao1399fa72011-05-18 10:09:39 -0700986/*
Peter Zijlstra76751042015-05-01 08:27:50 -0700987 * Wake-queues are lists of tasks with a pending wakeup, whose
988 * callers have already marked the task as woken internally,
989 * and can thus carry on. A common use case is being able to
990 * do the wakeups once the corresponding user lock as been
991 * released.
992 *
993 * We hold reference to each task in the list across the wakeup,
994 * thus guaranteeing that the memory is still valid by the time
995 * the actual wakeups are performed in wake_up_q().
996 *
997 * One per task suffices, because there's never a need for a task to be
998 * in two wake queues simultaneously; it is forbidden to abandon a task
999 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
1000 * already in a wake queue, the wakeup will happen soon and the second
1001 * waker can just skip it.
1002 *
1003 * The WAKE_Q macro declares and initializes the list head.
1004 * wake_up_q() does NOT reinitialize the list; it's expected to be
1005 * called near the end of a function, where the fact that the queue is
1006 * not used again will be easy to see by inspection.
1007 *
1008 * Note that this can cause spurious wakeups. schedule() callers
1009 * must ensure the call is done inside a loop, confirming that the
1010 * wakeup condition has in fact occurred.
1011 */
1012struct wake_q_node {
1013 struct wake_q_node *next;
1014};
1015
1016struct wake_q_head {
1017 struct wake_q_node *first;
1018 struct wake_q_node **lastp;
1019};
1020
1021#define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
1022
1023#define WAKE_Q(name) \
1024 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
1025
1026extern void wake_q_add(struct wake_q_head *head,
1027 struct task_struct *task);
1028extern void wake_up_q(struct wake_q_head *head);
1029
1030/*
Nikhil Rao1399fa72011-05-18 10:09:39 -07001031 * sched-domains (multiprocessor balancing) declarations:
1032 */
Peter Williams2dd73a42006-06-27 02:54:34 -07001033#ifdef CONFIG_SMP
Peter Zijlstrab5d978e2009-09-01 10:34:33 +02001034#define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
1035#define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
1036#define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
1037#define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
Peter Zijlstrac88d5912009-09-10 13:50:02 +02001038#define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +02001039#define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
Morten Rasmussen1f6e6c72016-07-25 14:34:22 +01001040#define SD_ASYM_CPUCAPACITY 0x0040 /* Groups have different max cpu capacities */
Morten Rasmussenbd425d42016-06-22 18:03:12 +01001041#define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu capacity */
Vincent Guittotd77b3ed2014-04-11 11:44:40 +02001042#define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +02001043#define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
1044#define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
Michael Neuling532cb4c2010-06-08 14:57:02 +10001045#define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +02001046#define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
Peter Zijlstrae3589f62011-07-15 10:35:52 +02001047#define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
Mel Gorman3a7053b2013-10-07 11:29:00 +01001048#define SD_NUMA 0x4000 /* cross-node balancing */
Siddha, Suresh B5c45bf22006-06-27 02:54:42 -07001049
Vincent Guittot143e1e22014-04-11 11:44:37 +02001050#ifdef CONFIG_SCHED_SMT
Guenter Roeckb6220ad2014-06-24 18:05:29 -07001051static inline int cpu_smt_flags(void)
Vincent Guittot143e1e22014-04-11 11:44:37 +02001052{
Nicolas Pitre5d4dfdd2014-05-27 13:50:41 -04001053 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
Vincent Guittot143e1e22014-04-11 11:44:37 +02001054}
1055#endif
1056
1057#ifdef CONFIG_SCHED_MC
Guenter Roeckb6220ad2014-06-24 18:05:29 -07001058static inline int cpu_core_flags(void)
Vincent Guittot143e1e22014-04-11 11:44:37 +02001059{
1060 return SD_SHARE_PKG_RESOURCES;
1061}
1062#endif
1063
1064#ifdef CONFIG_NUMA
Guenter Roeckb6220ad2014-06-24 18:05:29 -07001065static inline int cpu_numa_flags(void)
Vincent Guittot143e1e22014-04-11 11:44:37 +02001066{
1067 return SD_NUMA;
1068}
1069#endif
Michael Neuling532cb4c2010-06-08 14:57:02 +10001070
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +09001071struct sched_domain_attr {
1072 int relax_domain_level;
1073};
1074
1075#define SD_ATTR_INIT (struct sched_domain_attr) { \
1076 .relax_domain_level = -1, \
1077}
1078
Peter Zijlstra60495e72011-04-07 14:10:04 +02001079extern int sched_domain_level_max;
1080
Li Zefan5e6521e2013-03-05 16:06:23 +08001081struct sched_group;
1082
Peter Zijlstra24fc7ed2016-05-09 10:37:59 +02001083struct sched_domain_shared {
1084 atomic_t ref;
Peter Zijlstra0e369d72016-05-09 10:38:01 +02001085 atomic_t nr_busy_cpus;
Peter Zijlstra10e2f1a2016-05-09 10:38:05 +02001086 int has_idle_cores;
Peter Zijlstra24fc7ed2016-05-09 10:37:59 +02001087};
1088
Linus Torvalds1da177e2005-04-16 15:20:36 -07001089struct sched_domain {
1090 /* These fields must be setup */
1091 struct sched_domain *parent; /* top domain must be null terminated */
Siddha, Suresh B1a848872006-10-03 01:14:08 -07001092 struct sched_domain *child; /* bottom domain must be null terminated */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001093 struct sched_group *groups; /* the balancing groups of the domain */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001094 unsigned long min_interval; /* Minimum balance interval ms */
1095 unsigned long max_interval; /* Maximum balance interval ms */
1096 unsigned int busy_factor; /* less balancing by factor if busy */
1097 unsigned int imbalance_pct; /* No balance until over watermark */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001098 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
Nick Piggin78979862005-06-25 14:57:13 -07001099 unsigned int busy_idx;
1100 unsigned int idle_idx;
1101 unsigned int newidle_idx;
1102 unsigned int wake_idx;
Nick Piggin147cbb42005-06-25 14:57:19 -07001103 unsigned int forkexec_idx;
Peter Zijlstraa52bfd72009-09-01 10:34:35 +02001104 unsigned int smt_gain;
Vincent Guittot25f55d92013-04-23 16:59:02 +02001105
1106 int nohz_idle; /* NOHZ IDLE status */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001107 int flags; /* See SD_* */
Peter Zijlstra60495e72011-04-07 14:10:04 +02001108 int level;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001109
1110 /* Runtime fields. */
1111 unsigned long last_balance; /* init to jiffies. units in jiffies */
1112 unsigned int balance_interval; /* initialise to 1. units in ms. */
1113 unsigned int nr_balance_failed; /* initialise to 0 */
1114
Jason Lowf48627e2013-09-13 11:26:53 -07001115 /* idle_balance() stats */
Jason Low9bd721c2013-09-13 11:26:52 -07001116 u64 max_newidle_lb_cost;
Jason Lowf48627e2013-09-13 11:26:53 -07001117 unsigned long next_decay_max_lb_cost;
Peter Zijlstra2398f2c2008-06-27 13:41:35 +02001118
Peter Zijlstra10e2f1a2016-05-09 10:38:05 +02001119 u64 avg_scan_cost; /* select_idle_sibling */
1120
Linus Torvalds1da177e2005-04-16 15:20:36 -07001121#ifdef CONFIG_SCHEDSTATS
1122 /* load_balance() stats */
Ken Chen480b9432007-10-18 21:32:56 +02001123 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1124 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1125 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1126 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1127 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1128 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1129 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1130 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
Linus Torvalds1da177e2005-04-16 15:20:36 -07001131
1132 /* Active load balancing */
Ken Chen480b9432007-10-18 21:32:56 +02001133 unsigned int alb_count;
1134 unsigned int alb_failed;
1135 unsigned int alb_pushed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001136
Nick Piggin68767a02005-06-25 14:57:20 -07001137 /* SD_BALANCE_EXEC stats */
Ken Chen480b9432007-10-18 21:32:56 +02001138 unsigned int sbe_count;
1139 unsigned int sbe_balanced;
1140 unsigned int sbe_pushed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001141
Nick Piggin68767a02005-06-25 14:57:20 -07001142 /* SD_BALANCE_FORK stats */
Ken Chen480b9432007-10-18 21:32:56 +02001143 unsigned int sbf_count;
1144 unsigned int sbf_balanced;
1145 unsigned int sbf_pushed;
Nick Piggin68767a02005-06-25 14:57:20 -07001146
Linus Torvalds1da177e2005-04-16 15:20:36 -07001147 /* try_to_wake_up() stats */
Ken Chen480b9432007-10-18 21:32:56 +02001148 unsigned int ttwu_wake_remote;
1149 unsigned int ttwu_move_affine;
1150 unsigned int ttwu_move_balance;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001151#endif
Ingo Molnara5d8c342008-10-09 11:35:51 +02001152#ifdef CONFIG_SCHED_DEBUG
1153 char *name;
1154#endif
Peter Zijlstradce840a2011-04-07 14:09:50 +02001155 union {
1156 void *private; /* used during construction */
1157 struct rcu_head rcu; /* used during destruction */
1158 };
Peter Zijlstra24fc7ed2016-05-09 10:37:59 +02001159 struct sched_domain_shared *shared;
Rusty Russell6c99e9a2008-11-25 02:35:04 +10301160
Peter Zijlstra669c55e2010-04-16 14:59:29 +02001161 unsigned int span_weight;
Ingo Molnar4200efd2009-05-19 09:22:19 +02001162 /*
1163 * Span of all CPUs in this domain.
1164 *
1165 * NOTE: this field is variable length. (Allocated dynamically
1166 * by attaching extra space to the end of the structure,
1167 * depending on how many CPUs the kernel has booted up with)
Ingo Molnar4200efd2009-05-19 09:22:19 +02001168 */
1169 unsigned long span[0];
Linus Torvalds1da177e2005-04-16 15:20:36 -07001170};
1171
Rusty Russell758b2cd2008-11-25 02:35:04 +10301172static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1173{
Rusty Russell6c99e9a2008-11-25 02:35:04 +10301174 return to_cpumask(sd->span);
Rusty Russell758b2cd2008-11-25 02:35:04 +10301175}
1176
Rusty Russellacc3f5d2009-11-03 14:53:40 +10301177extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +09001178 struct sched_domain_attr *dattr_new);
Paul Jackson029190c2007-10-18 23:40:20 -07001179
Rusty Russellacc3f5d2009-11-03 14:53:40 +10301180/* Allocate an array of sched domains, for partition_sched_domains(). */
1181cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1182void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1183
Peter Zijlstra39be3502012-01-26 12:44:34 +01001184bool cpus_share_cache(int this_cpu, int that_cpu);
1185
Vincent Guittot143e1e22014-04-11 11:44:37 +02001186typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
Guenter Roeckb6220ad2014-06-24 18:05:29 -07001187typedef int (*sched_domain_flags_f)(void);
Vincent Guittot143e1e22014-04-11 11:44:37 +02001188
1189#define SDTL_OVERLAP 0x01
1190
1191struct sd_data {
1192 struct sched_domain **__percpu sd;
Peter Zijlstra24fc7ed2016-05-09 10:37:59 +02001193 struct sched_domain_shared **__percpu sds;
Vincent Guittot143e1e22014-04-11 11:44:37 +02001194 struct sched_group **__percpu sg;
Nicolas Pitre63b2ca32014-05-26 18:19:37 -04001195 struct sched_group_capacity **__percpu sgc;
Vincent Guittot143e1e22014-04-11 11:44:37 +02001196};
1197
1198struct sched_domain_topology_level {
1199 sched_domain_mask_f mask;
1200 sched_domain_flags_f sd_flags;
1201 int flags;
1202 int numa_level;
1203 struct sd_data data;
1204#ifdef CONFIG_SCHED_DEBUG
1205 char *name;
1206#endif
1207};
1208
Vincent Guittot143e1e22014-04-11 11:44:37 +02001209extern void set_sched_topology(struct sched_domain_topology_level *tl);
Chuansheng Liuf6be8af2014-09-04 15:17:53 +08001210extern void wake_up_if_idle(int cpu);
Vincent Guittot143e1e22014-04-11 11:44:37 +02001211
1212#ifdef CONFIG_SCHED_DEBUG
1213# define SD_INIT_NAME(type) .name = #type
1214#else
1215# define SD_INIT_NAME(type)
1216#endif
1217
Ingo Molnar1b427c12008-07-18 14:01:39 +02001218#else /* CONFIG_SMP */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001219
Ingo Molnar1b427c12008-07-18 14:01:39 +02001220struct sched_domain_attr;
1221
1222static inline void
Rusty Russellacc3f5d2009-11-03 14:53:40 +10301223partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
Ingo Molnar1b427c12008-07-18 14:01:39 +02001224 struct sched_domain_attr *dattr_new)
Con Kolivasd02c7a82007-07-26 13:40:43 +02001225{
Con Kolivasd02c7a82007-07-26 13:40:43 +02001226}
Peter Zijlstra39be3502012-01-26 12:44:34 +01001227
1228static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1229{
1230 return true;
1231}
1232
Ingo Molnar1b427c12008-07-18 14:01:39 +02001233#endif /* !CONFIG_SMP */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001234
Peter Zijlstra47fe38f2009-09-02 13:49:18 +02001235
Linus Torvalds1da177e2005-04-16 15:20:36 -07001236struct io_context; /* See blkdev.h */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001237
Linus Torvalds1da177e2005-04-16 15:20:36 -07001238
Chen, Kenneth W383f2832005-09-09 13:02:02 -07001239#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
Ingo Molnar36c8b582006-07-03 00:25:41 -07001240extern void prefetch_stack(struct task_struct *t);
Chen, Kenneth W383f2832005-09-09 13:02:02 -07001241#else
1242static inline void prefetch_stack(struct task_struct *t) { }
1243#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001244
1245struct audit_context; /* See audit.c */
1246struct mempolicy;
Jens Axboeb92ce552006-04-11 13:52:07 +02001247struct pipe_inode_info;
Serge E. Hallyn4865ecf2006-10-02 02:18:14 -07001248struct uts_namespace;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001249
Ingo Molnar20b8a592007-07-09 18:51:58 +02001250struct load_weight {
Peter Zijlstra9dbdb152013-11-18 18:27:06 +01001251 unsigned long weight;
1252 u32 inv_weight;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001253};
1254
Yuyang Du9d89c252015-07-15 08:04:37 +08001255/*
Yuyang Du7b595332016-04-05 12:12:28 +08001256 * The load_avg/util_avg accumulates an infinite geometric series
1257 * (see __update_load_avg() in kernel/sched/fair.c).
1258 *
1259 * [load_avg definition]
1260 *
1261 * load_avg = runnable% * scale_load_down(load)
1262 *
1263 * where runnable% is the time ratio that a sched_entity is runnable.
1264 * For cfs_rq, it is the aggregated load_avg of all runnable and
Yuyang Du9d89c252015-07-15 08:04:37 +08001265 * blocked sched_entities.
Yuyang Du7b595332016-04-05 12:12:28 +08001266 *
1267 * load_avg may also take frequency scaling into account:
1268 *
1269 * load_avg = runnable% * scale_load_down(load) * freq%
1270 *
1271 * where freq% is the CPU frequency normalized to the highest frequency.
1272 *
1273 * [util_avg definition]
1274 *
1275 * util_avg = running% * SCHED_CAPACITY_SCALE
1276 *
1277 * where running% is the time ratio that a sched_entity is running on
1278 * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
1279 * and blocked sched_entities.
1280 *
1281 * util_avg may also factor frequency scaling and CPU capacity scaling:
1282 *
1283 * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity%
1284 *
1285 * where freq% is the same as above, and capacity% is the CPU capacity
1286 * normalized to the greatest capacity (due to uarch differences, etc).
1287 *
1288 * N.B., the above ratios (runnable%, running%, freq%, and capacity%)
1289 * themselves are in the range of [0, 1]. To do fixed point arithmetics,
1290 * we therefore scale them to as large a range as necessary. This is for
1291 * example reflected by util_avg's SCHED_CAPACITY_SCALE.
1292 *
1293 * [Overflow issue]
1294 *
1295 * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
1296 * with the highest load (=88761), always runnable on a single cfs_rq,
1297 * and should not overflow as the number already hits PID_MAX_LIMIT.
1298 *
1299 * For all other cases (including 32-bit kernels), struct load_weight's
1300 * weight will overflow first before we do, because:
1301 *
1302 * Max(load_avg) <= Max(load.weight)
1303 *
1304 * Then it is the load_weight's responsibility to consider overflow
1305 * issues.
Yuyang Du9d89c252015-07-15 08:04:37 +08001306 */
Paul Turner9d85f212012-10-04 13:18:29 +02001307struct sched_avg {
Yuyang Du9d89c252015-07-15 08:04:37 +08001308 u64 last_update_time, load_sum;
1309 u32 util_sum, period_contrib;
1310 unsigned long load_avg, util_avg;
Paul Turner9d85f212012-10-04 13:18:29 +02001311};
1312
Ingo Molnar94c18222007-08-02 17:41:40 +02001313#ifdef CONFIG_SCHEDSTATS
Lucas De Marchi41acab82010-03-10 23:37:45 -03001314struct sched_statistics {
Ingo Molnar94c18222007-08-02 17:41:40 +02001315 u64 wait_start;
1316 u64 wait_max;
Arjan van de Ven6d082592008-01-25 21:08:35 +01001317 u64 wait_count;
1318 u64 wait_sum;
Arjan van de Ven8f0dfc32009-07-20 11:26:58 -07001319 u64 iowait_count;
1320 u64 iowait_sum;
Ingo Molnar94c18222007-08-02 17:41:40 +02001321
1322 u64 sleep_start;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001323 u64 sleep_max;
Ingo Molnar94c18222007-08-02 17:41:40 +02001324 s64 sum_sleep_runtime;
1325
1326 u64 block_start;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001327 u64 block_max;
1328 u64 exec_max;
Ingo Molnareba1ed42007-10-15 17:00:02 +02001329 u64 slice_max;
Ingo Molnarcc367732007-10-15 17:00:18 +02001330
Ingo Molnarcc367732007-10-15 17:00:18 +02001331 u64 nr_migrations_cold;
1332 u64 nr_failed_migrations_affine;
1333 u64 nr_failed_migrations_running;
1334 u64 nr_failed_migrations_hot;
1335 u64 nr_forced_migrations;
Ingo Molnarcc367732007-10-15 17:00:18 +02001336
1337 u64 nr_wakeups;
1338 u64 nr_wakeups_sync;
1339 u64 nr_wakeups_migrate;
1340 u64 nr_wakeups_local;
1341 u64 nr_wakeups_remote;
1342 u64 nr_wakeups_affine;
1343 u64 nr_wakeups_affine_attempts;
1344 u64 nr_wakeups_passive;
1345 u64 nr_wakeups_idle;
Lucas De Marchi41acab82010-03-10 23:37:45 -03001346};
1347#endif
1348
1349struct sched_entity {
1350 struct load_weight load; /* for load-balancing */
1351 struct rb_node run_node;
1352 struct list_head group_node;
1353 unsigned int on_rq;
1354
1355 u64 exec_start;
1356 u64 sum_exec_runtime;
1357 u64 vruntime;
1358 u64 prev_sum_exec_runtime;
1359
Lucas De Marchi41acab82010-03-10 23:37:45 -03001360 u64 nr_migrations;
1361
Lucas De Marchi41acab82010-03-10 23:37:45 -03001362#ifdef CONFIG_SCHEDSTATS
1363 struct sched_statistics statistics;
Ingo Molnar94c18222007-08-02 17:41:40 +02001364#endif
1365
Ingo Molnar20b8a592007-07-09 18:51:58 +02001366#ifdef CONFIG_FAIR_GROUP_SCHED
Peter Zijlstrafed14d42012-02-11 06:05:00 +01001367 int depth;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001368 struct sched_entity *parent;
1369 /* rq on which this entity is (to be) queued: */
1370 struct cfs_rq *cfs_rq;
1371 /* rq "owned" by this entity/group: */
1372 struct cfs_rq *my_q;
1373#endif
Clark Williams8bd75c72013-02-07 09:47:07 -06001374
Alex Shi141965c2013-06-26 13:05:39 +08001375#ifdef CONFIG_SMP
Jiri Olsa5a107802015-12-08 21:23:59 +01001376 /*
1377 * Per entity load average tracking.
1378 *
1379 * Put into separate cache line so it does not
1380 * collide with read-mostly values above.
1381 */
1382 struct sched_avg avg ____cacheline_aligned_in_smp;
Paul Turner9d85f212012-10-04 13:18:29 +02001383#endif
Ingo Molnar20b8a592007-07-09 18:51:58 +02001384};
Ingo Molnar70b97a72006-07-03 00:25:42 -07001385
Peter Zijlstrafa717062008-01-25 21:08:27 +01001386struct sched_rt_entity {
1387 struct list_head run_list;
Peter Zijlstra78f2c7d2008-01-25 21:08:27 +01001388 unsigned long timeout;
Ying Xue57d2aa02012-07-17 15:03:43 +08001389 unsigned long watchdog_stamp;
Richard Kennedybee367e2008-08-01 13:24:08 +01001390 unsigned int time_slice;
Peter Zijlstraff77e462016-01-18 15:27:07 +01001391 unsigned short on_rq;
1392 unsigned short on_list;
Peter Zijlstra6f505b12008-01-25 21:08:30 +01001393
Peter Zijlstra58d6c2d2008-04-19 19:45:00 +02001394 struct sched_rt_entity *back;
Peter Zijlstra052f1dc2008-02-13 15:45:40 +01001395#ifdef CONFIG_RT_GROUP_SCHED
Peter Zijlstra6f505b12008-01-25 21:08:30 +01001396 struct sched_rt_entity *parent;
1397 /* rq on which this entity is (to be) queued: */
1398 struct rt_rq *rt_rq;
1399 /* rq "owned" by this entity/group: */
1400 struct rt_rq *my_q;
1401#endif
Peter Zijlstrafa717062008-01-25 21:08:27 +01001402};
1403
Dario Faggioliaab03e02013-11-28 11:14:43 +01001404struct sched_dl_entity {
1405 struct rb_node rb_node;
1406
1407 /*
1408 * Original scheduling parameters. Copied here from sched_attr
xiaofeng.yan4027d082014-05-09 03:21:27 +00001409 * during sched_setattr(), they will remain the same until
1410 * the next sched_setattr().
Dario Faggioliaab03e02013-11-28 11:14:43 +01001411 */
1412 u64 dl_runtime; /* maximum runtime for each instance */
1413 u64 dl_deadline; /* relative deadline of each instance */
Harald Gustafsson755378a2013-11-07 14:43:40 +01001414 u64 dl_period; /* separation of two instances (period) */
Dario Faggioli332ac172013-11-07 14:43:45 +01001415 u64 dl_bw; /* dl_runtime / dl_deadline */
Daniel Bristot de Oliveira0559ea32017-05-29 16:24:03 +02001416 u64 dl_density; /* dl_runtime / dl_deadline */
Dario Faggioliaab03e02013-11-28 11:14:43 +01001417
1418 /*
1419 * Actual scheduling parameters. Initialized with the values above,
1420 * they are continously updated during task execution. Note that
1421 * the remaining runtime could be < 0 in case we are in overrun.
1422 */
1423 s64 runtime; /* remaining runtime for this instance */
1424 u64 deadline; /* absolute deadline for this instance */
1425 unsigned int flags; /* specifying the scheduler behaviour */
1426
1427 /*
1428 * Some bool flags:
1429 *
1430 * @dl_throttled tells if we exhausted the runtime. If so, the
1431 * task has to wait for a replenishment to be performed at the
1432 * next firing of dl_timer.
1433 *
Dario Faggioli2d3d8912013-11-07 14:43:44 +01001434 * @dl_boosted tells if we are boosted due to DI. If so we are
1435 * outside bandwidth enforcement mechanism (but only until we
Juri Lelli5bfd1262014-04-15 13:49:04 +02001436 * exit the critical section);
1437 *
1438 * @dl_yielded tells if task gave up the cpu before consuming
1439 * all its available runtime during the last job.
Dario Faggioliaab03e02013-11-28 11:14:43 +01001440 */
Luca Abeni72f9f3f2016-03-07 12:27:04 +01001441 int dl_throttled, dl_boosted, dl_yielded;
Dario Faggioliaab03e02013-11-28 11:14:43 +01001442
1443 /*
1444 * Bandwidth enforcement timer. Each -deadline task has its
1445 * own bandwidth to be enforced, thus we need one timer per task.
1446 */
1447 struct hrtimer dl_timer;
1448};
Clark Williams8bd75c72013-02-07 09:47:07 -06001449
Paul E. McKenney1d082fd2014-08-14 16:01:53 -07001450union rcu_special {
1451 struct {
Paul E. McKenney8203d6d2015-08-02 13:53:17 -07001452 u8 blocked;
1453 u8 need_qs;
1454 u8 exp_need_qs;
1455 u8 pad; /* Otherwise the compiler can store garbage here. */
1456 } b; /* Bits. */
1457 u32 s; /* Set of bits. */
Paul E. McKenney1d082fd2014-08-14 16:01:53 -07001458};
Paul E. McKenney86848962009-08-27 15:00:12 -07001459struct rcu_node;
1460
Peter Zijlstra8dc85d52010-09-02 16:50:03 +02001461enum perf_event_task_context {
1462 perf_invalid_context = -1,
1463 perf_hw_context = 0,
Peter Zijlstra89a1e182010-09-07 17:34:50 +02001464 perf_sw_context,
Peter Zijlstra8dc85d52010-09-02 16:50:03 +02001465 perf_nr_task_contexts,
1466};
1467
Mel Gorman72b252a2015-09-04 15:47:32 -07001468/* Track pages that require TLB flushes */
1469struct tlbflush_unmap_batch {
1470 /*
1471 * Each bit set is a CPU that potentially has a TLB entry for one of
1472 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1473 */
1474 struct cpumask cpumask;
1475
1476 /* True if any bit in cpumask is set */
1477 bool flush_required;
Mel Gormand950c942015-09-04 15:47:35 -07001478
1479 /*
1480 * If true then the PTE was dirty when unmapped. The entry must be
1481 * flushed before IO is initiated or a stale TLB entry potentially
1482 * allows an update without redirtying the page.
1483 */
1484 bool writable;
Mel Gorman72b252a2015-09-04 15:47:32 -07001485};
1486
Linus Torvalds1da177e2005-04-16 15:20:36 -07001487struct task_struct {
Andy Lutomirskic65eacb2016-09-13 14:29:24 -07001488#ifdef CONFIG_THREAD_INFO_IN_TASK
1489 /*
1490 * For reasons of header soup (see current_thread_info()), this
1491 * must be the first element of task_struct.
1492 */
1493 struct thread_info thread_info;
1494#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001495 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
Roman Zippelf7e42172007-05-09 02:35:17 -07001496 void *stack;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001497 atomic_t usage;
William Cohen97dc32c2007-05-08 00:23:41 -07001498 unsigned int flags; /* per process flags, defined below */
1499 unsigned int ptrace;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001500
Peter Williams2dd73a42006-06-27 02:54:34 -07001501#ifdef CONFIG_SMP
Peter Zijlstrafa14ff42011-09-12 13:06:17 +02001502 struct llist_node wake_entry;
Peter Zijlstra3ca7a442011-04-05 17:23:40 +02001503 int on_cpu;
Andy Lutomirskic65eacb2016-09-13 14:29:24 -07001504#ifdef CONFIG_THREAD_INFO_IN_TASK
1505 unsigned int cpu; /* current CPU */
1506#endif
Mike Galbraith63b0e9e2015-07-14 17:39:50 +02001507 unsigned int wakee_flips;
Michael Wang62470412013-07-04 12:55:51 +08001508 unsigned long wakee_flip_decay_ts;
Mike Galbraith63b0e9e2015-07-14 17:39:50 +02001509 struct task_struct *last_wakee;
Peter Zijlstraac66f542013-10-07 11:29:16 +01001510
1511 int wake_cpu;
Nick Piggin4866cde2005-06-25 14:57:23 -07001512#endif
Peter Zijlstrafd2f4412011-04-05 17:23:44 +02001513 int on_rq;
Ingo Molnar50e645a2007-07-09 18:52:00 +02001514
Ingo Molnarb29739f2006-06-27 02:54:51 -07001515 int prio, static_prio, normal_prio;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001516 unsigned int rt_priority;
Ingo Molnar5522d5d2007-10-15 17:00:12 +02001517 const struct sched_class *sched_class;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001518 struct sched_entity se;
Peter Zijlstrafa717062008-01-25 21:08:27 +01001519 struct sched_rt_entity rt;
Peter Zijlstra8323f262012-06-22 13:36:05 +02001520#ifdef CONFIG_CGROUP_SCHED
1521 struct task_group *sched_task_group;
1522#endif
Dario Faggioliaab03e02013-11-28 11:14:43 +01001523 struct sched_dl_entity dl;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001524
Avi Kivitye107be32007-07-26 13:40:43 +02001525#ifdef CONFIG_PREEMPT_NOTIFIERS
1526 /* list of struct preempt_notifier: */
1527 struct hlist_head preempt_notifiers;
1528#endif
1529
Alexey Dobriyan6c5c9342006-09-29 01:59:40 -07001530#ifdef CONFIG_BLK_DEV_IO_TRACE
Jens Axboe2056a782006-03-23 20:00:26 +01001531 unsigned int btrace_seq;
Alexey Dobriyan6c5c9342006-09-29 01:59:40 -07001532#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001533
William Cohen97dc32c2007-05-08 00:23:41 -07001534 unsigned int policy;
Peter Zijlstra29baa742012-04-23 12:11:21 +02001535 int nr_cpus_allowed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001536 cpumask_t cpus_allowed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001537
Paul E. McKenneya57eb942010-06-29 16:49:16 -07001538#ifdef CONFIG_PREEMPT_RCU
Paul E. McKenneye260be62008-01-25 21:08:24 +01001539 int rcu_read_lock_nesting;
Paul E. McKenney1d082fd2014-08-14 16:01:53 -07001540 union rcu_special rcu_read_unlock_special;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001541 struct list_head rcu_node_entry;
Paul E. McKenneya57eb942010-06-29 16:49:16 -07001542 struct rcu_node *rcu_blocked_node;
Pranith Kumar28f65692014-09-22 14:00:48 -04001543#endif /* #ifdef CONFIG_PREEMPT_RCU */
Paul E. McKenney8315f422014-06-27 13:42:20 -07001544#ifdef CONFIG_TASKS_RCU
1545 unsigned long rcu_tasks_nvcsw;
1546 bool rcu_tasks_holdout;
1547 struct list_head rcu_tasks_holdout_list;
Paul E. McKenney176f8f72014-08-04 17:43:50 -07001548 int rcu_tasks_idle_cpu;
Paul E. McKenney8315f422014-06-27 13:42:20 -07001549#endif /* #ifdef CONFIG_TASKS_RCU */
Paul E. McKenneye260be62008-01-25 21:08:24 +01001550
Naveen N. Raof6db8342015-06-25 23:53:37 +05301551#ifdef CONFIG_SCHED_INFO
Linus Torvalds1da177e2005-04-16 15:20:36 -07001552 struct sched_info sched_info;
1553#endif
1554
1555 struct list_head tasks;
Dario Faggioli806c09a2010-11-30 19:51:33 +01001556#ifdef CONFIG_SMP
Gregory Haskins917b6272008-12-29 09:39:53 -05001557 struct plist_node pushable_tasks;
Juri Lelli1baca4c2013-11-07 14:43:38 +01001558 struct rb_node pushable_dl_tasks;
Dario Faggioli806c09a2010-11-30 19:51:33 +01001559#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001560
1561 struct mm_struct *mm, *active_mm;
Davidlohr Bueso615d6e82014-04-07 15:37:25 -07001562 /* per-thread vma caching */
Linus Torvalds84580562018-09-12 23:57:48 -10001563 u64 vmacache_seqnum;
Davidlohr Bueso615d6e82014-04-07 15:37:25 -07001564 struct vm_area_struct *vmacache[VMACACHE_SIZE];
KAMEZAWA Hiroyuki34e55232010-03-05 13:41:40 -08001565#if defined(SPLIT_RSS_COUNTING)
1566 struct task_rss_stat rss_stat;
1567#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001568/* task state */
William Cohen97dc32c2007-05-08 00:23:41 -07001569 int exit_state;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001570 int exit_code, exit_signal;
1571 int pdeath_signal; /* The signal sent when the parent dies */
Palmer Dabbelte7cc4172015-04-30 21:19:55 -07001572 unsigned long jobctl; /* JOBCTL_*, siglock protected */
Andrei Epure9b89f6b2013-04-11 20:30:29 +03001573
1574 /* Used for emulating ABI behavior of previous Linux versions */
William Cohen97dc32c2007-05-08 00:23:41 -07001575 unsigned int personality;
Andrei Epure9b89f6b2013-04-11 20:30:29 +03001576
Peter Zijlstrabe958bd2015-11-25 16:02:07 +01001577 /* scheduler bits, serialized by scheduler locks */
Lennart Poetteringca94c442009-06-15 17:17:47 +02001578 unsigned sched_reset_on_fork:1;
Peter Zijlstraa8e4f2e2011-04-05 17:23:49 +02001579 unsigned sched_contributes_to_load:1;
Peter Zijlstraff303e62015-04-17 20:05:30 +02001580 unsigned sched_migrated:1;
Peter Zijlstrab7e7ade2016-05-23 11:19:07 +02001581 unsigned sched_remote_wakeup:1;
Peter Zijlstrabe958bd2015-11-25 16:02:07 +01001582 unsigned :0; /* force alignment to the next boundary */
1583
1584 /* unserialized, strictly 'current' */
1585 unsigned in_execve:1; /* bit to tell LSMs we're in execve */
1586 unsigned in_iowait:1;
Andy Lutomirski7e781412016-08-02 14:05:36 -07001587#if !defined(TIF_RESTORE_SIGMASK)
1588 unsigned restore_sigmask:1;
1589#endif
Tejun Heo626ebc42015-11-05 18:46:09 -08001590#ifdef CONFIG_MEMCG
1591 unsigned memcg_may_oom:1;
Johannes Weiner127424c2016-01-20 15:02:32 -08001592#ifndef CONFIG_SLOB
Vladimir Davydov6f185c22014-12-12 16:55:15 -08001593 unsigned memcg_kmem_skip_account:1;
1594#endif
Johannes Weiner127424c2016-01-20 15:02:32 -08001595#endif
Peter Zijlstraff303e62015-04-17 20:05:30 +02001596#ifdef CONFIG_COMPAT_BRK
1597 unsigned brk_randomized:1;
1598#endif
Tejun Heof44236a2017-03-16 16:54:24 -04001599#ifdef CONFIG_CGROUPS
1600 /* disallow userland-initiated cgroup migration */
1601 unsigned no_cgroup_migration:1;
1602#endif
Vladimir Davydov6f185c22014-12-12 16:55:15 -08001603
Kees Cook1d4457f2014-05-21 15:23:46 -07001604 unsigned long atomic_flags; /* Flags needing atomic access. */
1605
Andy Lutomirskif56141e2015-02-12 15:01:14 -08001606 struct restart_block restart_block;
1607
Linus Torvalds1da177e2005-04-16 15:20:36 -07001608 pid_t pid;
1609 pid_t tgid;
Arjan van de Ven0a425402006-09-26 10:52:38 +02001610
Hiroshi Shimamoto13145622009-08-18 15:06:02 +09001611#ifdef CONFIG_CC_STACKPROTECTOR
Arjan van de Ven0a425402006-09-26 10:52:38 +02001612 /* Canary value for the -fstack-protector gcc feature */
1613 unsigned long stack_canary;
Hiroshi Shimamoto13145622009-08-18 15:06:02 +09001614#endif
Oleg Nesterov4d1d61a2012-05-11 10:59:08 +10001615 /*
Linus Torvalds1da177e2005-04-16 15:20:36 -07001616 * pointers to (original) parent process, youngest child, younger sibling,
Oleg Nesterov4d1d61a2012-05-11 10:59:08 +10001617 * older sibling, respectively. (p->father can be replaced with
Roland McGrathf4700212008-03-24 18:36:23 -07001618 * p->real_parent->pid)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001619 */
Kees Cookabd63bc2011-12-14 14:39:26 -08001620 struct task_struct __rcu *real_parent; /* real parent process */
1621 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001622 /*
Roland McGrathf4700212008-03-24 18:36:23 -07001623 * children/sibling forms the list of my natural children
Linus Torvalds1da177e2005-04-16 15:20:36 -07001624 */
1625 struct list_head children; /* list of my children */
1626 struct list_head sibling; /* linkage in my parent's children list */
1627 struct task_struct *group_leader; /* threadgroup leader */
1628
Roland McGrathf4700212008-03-24 18:36:23 -07001629 /*
1630 * ptraced is the list of tasks this task is using ptrace on.
1631 * This includes both natural children and PTRACE_ATTACH targets.
1632 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1633 */
1634 struct list_head ptraced;
1635 struct list_head ptrace_entry;
1636
Linus Torvalds1da177e2005-04-16 15:20:36 -07001637 /* PID/PID hash table linkage. */
Eric W. Biederman92476d72006-03-31 02:31:42 -08001638 struct pid_link pids[PIDTYPE_MAX];
Oleg Nesterov47e65322006-03-28 16:11:25 -08001639 struct list_head thread_group;
Oleg Nesterov0c740d02014-01-21 15:49:56 -08001640 struct list_head thread_node;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001641
1642 struct completion *vfork_done; /* for vfork() */
1643 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1644 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1645
Michael Neulingc66f08b2007-10-18 03:06:34 -07001646 cputime_t utime, stime, utimescaled, stimescaled;
Laurent Vivier9ac52312007-10-15 17:00:19 +02001647 cputime_t gtime;
Peter Zijlstra9d7fb042015-06-30 11:30:54 +02001648 struct prev_cputime prev_cputime;
Frederic Weisbecker6a616712012-12-16 20:00:34 +01001649#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
Frederic Weisbeckerb7ce2272015-11-19 16:47:34 +01001650 seqcount_t vtime_seqcount;
Frederic Weisbecker6a616712012-12-16 20:00:34 +01001651 unsigned long long vtime_snap;
1652 enum {
Frederic Weisbecker7098c1e2015-11-19 16:47:30 +01001653 /* Task is sleeping or running in a CPU with VTIME inactive */
1654 VTIME_INACTIVE = 0,
1655 /* Task runs in userspace in a CPU with VTIME active */
Frederic Weisbecker6a616712012-12-16 20:00:34 +01001656 VTIME_USER,
Frederic Weisbecker7098c1e2015-11-19 16:47:30 +01001657 /* Task runs in kernelspace in a CPU with VTIME active */
Frederic Weisbecker6a616712012-12-16 20:00:34 +01001658 VTIME_SYS,
1659 } vtime_snap_whence;
1660#endif
Frederic Weisbeckerd027d452015-06-07 15:54:30 +02001661
1662#ifdef CONFIG_NO_HZ_FULL
Frederic Weisbeckerf009a7a2016-03-24 15:38:00 +01001663 atomic_t tick_dep_mask;
Frederic Weisbeckerd027d452015-06-07 15:54:30 +02001664#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001665 unsigned long nvcsw, nivcsw; /* context switch counts */
Thomas Gleixnerccbf62d2014-07-16 21:04:34 +00001666 u64 start_time; /* monotonic time in nsec */
Thomas Gleixner57e0be02014-07-16 21:04:32 +00001667 u64 real_start_time; /* boot based time in nsec */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001668/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1669 unsigned long min_flt, maj_flt;
1670
Frank Mayharf06febc2008-09-12 09:54:39 -07001671 struct task_cputime cputime_expires;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001672 struct list_head cpu_timers[3];
1673
1674/* process credentials */
Eric W. Biedermane747b4a2016-11-14 18:48:07 -06001675 const struct cred __rcu *ptracer_cred; /* Tracer's credentials at attach */
Arnd Bergmann1b0ba1c2010-02-24 19:45:09 +01001676 const struct cred __rcu *real_cred; /* objective and real subjective task
David Howells3b11a1d2008-11-14 10:39:26 +11001677 * credentials (COW) */
Arnd Bergmann1b0ba1c2010-02-24 19:45:09 +01001678 const struct cred __rcu *cred; /* effective (overridable) subjective task
David Howells3b11a1d2008-11-14 10:39:26 +11001679 * credentials (COW) */
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -07001680 char comm[TASK_COMM_LEN]; /* executable name excluding path
1681 - access with [gs]et_task_comm (which lock
1682 it with task_lock())
Linus Torvalds221af7f2010-01-28 22:14:42 -08001683 - initialized normally by setup_new_exec */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001684/* file system info */
NeilBrown756daf22015-03-23 13:37:38 +11001685 struct nameidata *nameidata;
Alexey Dobriyan3d5b6fc2006-09-29 01:59:40 -07001686#ifdef CONFIG_SYSVIPC
Linus Torvalds1da177e2005-04-16 15:20:36 -07001687/* ipc stuff */
1688 struct sysv_sem sysvsem;
Jack Millerab602f72014-08-08 14:23:19 -07001689 struct sysv_shm sysvshm;
Alexey Dobriyan3d5b6fc2006-09-29 01:59:40 -07001690#endif
Mandeep Singh Bainese162b392009-01-15 11:08:40 -08001691#ifdef CONFIG_DETECT_HUNG_TASK
Ingo Molnar82a1fcb2008-01-25 21:08:02 +01001692/* hung task detection */
Ingo Molnar82a1fcb2008-01-25 21:08:02 +01001693 unsigned long last_switch_count;
1694#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001695/* filesystem information */
1696 struct fs_struct *fs;
1697/* open file information */
1698 struct files_struct *files;
Serge E. Hallyn1651e142006-10-02 02:18:08 -07001699/* namespaces */
Serge E. Hallynab516012006-10-02 02:18:06 -07001700 struct nsproxy *nsproxy;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001701/* signal handlers */
1702 struct signal_struct *signal;
1703 struct sighand_struct *sighand;
1704
1705 sigset_t blocked, real_blocked;
Roland McGrathf3de2722008-04-30 00:53:09 -07001706 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001707 struct sigpending pending;
1708
1709 unsigned long sas_ss_sp;
1710 size_t sas_ss_size;
Stas Sergeev2a742132016-04-14 23:20:04 +03001711 unsigned sas_ss_flags;
Oleg Nesterov2e01fab2015-11-06 16:32:19 -08001712
Al Viro67d12142012-06-27 11:07:19 +04001713 struct callback_head *task_works;
Oleg Nesterove73f8952012-05-11 10:59:07 +10001714
Linus Torvalds1da177e2005-04-16 15:20:36 -07001715 struct audit_context *audit_context;
Al Virobfef93a2008-01-10 04:53:18 -05001716#ifdef CONFIG_AUDITSYSCALL
Eric W. Biedermane1760bd2012-09-10 22:39:43 -07001717 kuid_t loginuid;
Eric Paris4746ec52008-01-08 10:06:53 -05001718 unsigned int sessionid;
Al Virobfef93a2008-01-10 04:53:18 -05001719#endif
Will Drewry932eceb2012-04-12 16:47:54 -05001720 struct seccomp seccomp;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001721
1722/* Thread group tracking */
Eric W. Biederman110012a2020-03-30 19:01:04 -05001723 u64 parent_exec_id;
1724 u64 self_exec_id;
Miao Xie58568d22009-06-16 15:31:49 -07001725/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1726 * mempolicy */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001727 spinlock_t alloc_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001728
Ingo Molnarb29739f2006-06-27 02:54:51 -07001729 /* Protection of the PI data structures: */
Thomas Gleixner1d615482009-11-17 14:54:03 +01001730 raw_spinlock_t pi_lock;
Ingo Molnarb29739f2006-06-27 02:54:51 -07001731
Peter Zijlstra76751042015-05-01 08:27:50 -07001732 struct wake_q_node wake_q;
1733
Ingo Molnar23f78d4a2006-06-27 02:54:53 -07001734#ifdef CONFIG_RT_MUTEXES
1735 /* PI waiters blocked on a rt_mutex held by this task */
Peter Zijlstrafb00aca2013-11-07 14:43:43 +01001736 struct rb_root pi_waiters;
1737 struct rb_node *pi_waiters_leftmost;
Ingo Molnar23f78d4a2006-06-27 02:54:53 -07001738 /* Deadlock detection and priority inheritance handling */
1739 struct rt_mutex_waiter *pi_blocked_on;
Ingo Molnar23f78d4a2006-06-27 02:54:53 -07001740#endif
1741
Ingo Molnar408894e2006-01-09 15:59:20 -08001742#ifdef CONFIG_DEBUG_MUTEXES
1743 /* mutex deadlock detection */
1744 struct mutex_waiter *blocked_on;
1745#endif
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001746#ifdef CONFIG_TRACE_IRQFLAGS
1747 unsigned int irq_events;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001748 unsigned long hardirq_enable_ip;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001749 unsigned long hardirq_disable_ip;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001750 unsigned int hardirq_enable_event;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001751 unsigned int hardirq_disable_event;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001752 int hardirqs_enabled;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001753 int hardirq_context;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001754 unsigned long softirq_disable_ip;
1755 unsigned long softirq_enable_ip;
1756 unsigned int softirq_disable_event;
1757 unsigned int softirq_enable_event;
1758 int softirqs_enabled;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001759 int softirq_context;
1760#endif
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001761#ifdef CONFIG_LOCKDEP
Peter Zijlstrabdb94412008-02-25 23:02:48 +01001762# define MAX_LOCK_DEPTH 48UL
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001763 u64 curr_chain_key;
1764 int lockdep_depth;
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001765 unsigned int lockdep_recursion;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001766 struct held_lock held_locks[MAX_LOCK_DEPTH];
Nick Piggincf40bd12009-01-21 08:12:39 +01001767 gfp_t lockdep_reclaim_gfp;
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001768#endif
Andrey Ryabininc6d30852016-01-20 15:00:55 -08001769#ifdef CONFIG_UBSAN
1770 unsigned int in_ubsan;
1771#endif
Ingo Molnar408894e2006-01-09 15:59:20 -08001772
Linus Torvalds1da177e2005-04-16 15:20:36 -07001773/* journalling filesystem info */
1774 void *journal_info;
1775
Neil Brownd89d8792007-05-01 09:53:42 +02001776/* stacked block device info */
Akinobu Mitabddd87c2010-02-23 08:55:42 +01001777 struct bio_list *bio_list;
Neil Brownd89d8792007-05-01 09:53:42 +02001778
Jens Axboe73c10102011-03-08 13:19:51 +01001779#ifdef CONFIG_BLOCK
1780/* stack plugging */
1781 struct blk_plug *plug;
1782#endif
1783
Linus Torvalds1da177e2005-04-16 15:20:36 -07001784/* VM state */
1785 struct reclaim_state *reclaim_state;
1786
Linus Torvalds1da177e2005-04-16 15:20:36 -07001787 struct backing_dev_info *backing_dev_info;
1788
1789 struct io_context *io_context;
1790
1791 unsigned long ptrace_message;
1792 siginfo_t *last_siginfo; /* For ptrace use. */
Andrew Morton7c3ab732006-12-10 02:19:19 -08001793 struct task_io_accounting ioac;
Jay Lan8f0ab512006-09-30 23:28:59 -07001794#if defined(CONFIG_TASK_XACCT)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001795 u64 acct_rss_mem1; /* accumulated rss usage */
1796 u64 acct_vm_mem1; /* accumulated virtual memory usage */
Jonathan Lim49b5cf32008-07-25 01:48:40 -07001797 cputime_t acct_timexpd; /* stime + utime since last update */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001798#endif
1799#ifdef CONFIG_CPUSETS
Miao Xie58568d22009-06-16 15:31:49 -07001800 nodemask_t mems_allowed; /* Protected by alloc_lock */
Mel Gormancc9a6c82012-03-21 16:34:11 -07001801 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
Paul Jackson825a46a2006-03-24 03:16:03 -08001802 int cpuset_mem_spread_rotor;
Jack Steiner6adef3e2010-05-26 14:42:49 -07001803 int cpuset_slab_spread_rotor;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001804#endif
Paul Menageddbcc7e2007-10-18 23:39:30 -07001805#ifdef CONFIG_CGROUPS
Paul Menage817929e2007-10-18 23:39:36 -07001806 /* Control Group info protected by css_set_lock */
Arnd Bergmann2c392b82010-02-24 19:41:39 +01001807 struct css_set __rcu *cgroups;
Paul Menage817929e2007-10-18 23:39:36 -07001808 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1809 struct list_head cg_list;
Paul Menageddbcc7e2007-10-18 23:39:30 -07001810#endif
Alexey Dobriyan42b2dd02007-10-16 23:27:30 -07001811#ifdef CONFIG_FUTEX
Ingo Molnar0771dfe2006-03-27 01:16:22 -08001812 struct robust_list_head __user *robust_list;
Ingo Molnar34f192c2006-03-27 01:16:24 -08001813#ifdef CONFIG_COMPAT
1814 struct compat_robust_list_head __user *compat_robust_list;
1815#endif
Ingo Molnarc87e2832006-06-27 02:54:58 -07001816 struct list_head pi_state_list;
1817 struct futex_pi_state *pi_state_cache;
Thomas Gleixnerad3466a2021-02-01 10:01:41 +00001818 struct mutex futex_exit_mutex;
Thomas Gleixner2c116892021-02-01 10:01:34 +00001819 unsigned int futex_state;
Alexey Dobriyan42b2dd02007-10-16 23:27:30 -07001820#endif
Ingo Molnarcdd6c482009-09-21 12:02:48 +02001821#ifdef CONFIG_PERF_EVENTS
Peter Zijlstra8dc85d52010-09-02 16:50:03 +02001822 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
Ingo Molnarcdd6c482009-09-21 12:02:48 +02001823 struct mutex perf_event_mutex;
1824 struct list_head perf_event_list;
Paul Mackerrasa63eaf32009-05-22 14:17:31 +10001825#endif
Thomas Gleixner8f47b182014-02-07 20:58:39 +01001826#ifdef CONFIG_DEBUG_PREEMPT
1827 unsigned long preempt_disable_ip;
1828#endif
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001829#ifdef CONFIG_NUMA
Miao Xie58568d22009-06-16 15:31:49 -07001830 struct mempolicy *mempolicy; /* Protected by alloc_lock */
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001831 short il_next;
Eric Dumazet207205a2011-03-22 16:30:44 -07001832 short pref_node_fork;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001833#endif
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001834#ifdef CONFIG_NUMA_BALANCING
1835 int numa_scan_seq;
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001836 unsigned int numa_scan_period;
Mel Gorman598f0ec2013-10-07 11:28:55 +01001837 unsigned int numa_scan_period_max;
Rik van Rielde1c9ce2013-10-07 11:29:39 +01001838 int numa_preferred_nid;
Mel Gorman6b9a7462013-10-07 11:29:11 +01001839 unsigned long numa_migrate_retry;
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001840 u64 node_stamp; /* migration stamp */
Rik van Riel7e2703e2014-01-27 17:03:45 -05001841 u64 last_task_numa_placement;
1842 u64 last_sum_exec_runtime;
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001843 struct callback_head numa_work;
Mel Gormanf809ca92013-10-07 11:28:57 +01001844
Peter Zijlstra8c8a7432013-10-07 11:29:21 +01001845 struct list_head numa_entry;
1846 struct numa_group *numa_group;
1847
Mel Gorman745d6142013-10-07 11:28:59 +01001848 /*
Iulia Manda44dba3d2014-10-31 02:13:31 +02001849 * numa_faults is an array split into four regions:
1850 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1851 * in this precise order.
1852 *
1853 * faults_memory: Exponential decaying average of faults on a per-node
1854 * basis. Scheduling placement decisions are made based on these
1855 * counts. The values remain static for the duration of a PTE scan.
1856 * faults_cpu: Track the nodes the process was running on when a NUMA
1857 * hinting fault was incurred.
1858 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1859 * during the current scan window. When the scan completes, the counts
1860 * in faults_memory and faults_cpu decay and these values are copied.
Mel Gorman745d6142013-10-07 11:28:59 +01001861 */
Iulia Manda44dba3d2014-10-31 02:13:31 +02001862 unsigned long *numa_faults;
Mel Gorman83e1d2c2013-10-07 11:29:27 +01001863 unsigned long total_numa_faults;
Mel Gorman745d6142013-10-07 11:28:59 +01001864
1865 /*
Rik van Riel04bb2f92013-10-07 11:29:36 +01001866 * numa_faults_locality tracks if faults recorded during the last
Mel Gorman074c2382015-03-25 15:55:42 -07001867 * scan window were remote/local or failed to migrate. The task scan
1868 * period is adapted based on the locality of the faults with different
1869 * weights depending on whether they were shared or private faults
Rik van Riel04bb2f92013-10-07 11:29:36 +01001870 */
Mel Gorman074c2382015-03-25 15:55:42 -07001871 unsigned long numa_faults_locality[3];
Rik van Riel04bb2f92013-10-07 11:29:36 +01001872
Ingo Molnarb32e86b2013-10-07 11:29:30 +01001873 unsigned long numa_pages_migrated;
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001874#endif /* CONFIG_NUMA_BALANCING */
1875
Mel Gorman72b252a2015-09-04 15:47:32 -07001876#ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1877 struct tlbflush_unmap_batch tlb_ubc;
1878#endif
1879
Ingo Molnare56d0902006-01-08 01:01:37 -08001880 struct rcu_head rcu;
Jens Axboeb92ce552006-04-11 13:52:07 +02001881
1882 /*
1883 * cache last used pipe for splice
1884 */
1885 struct pipe_inode_info *splice_pipe;
Eric Dumazet5640f762012-09-23 23:04:42 +00001886
1887 struct page_frag task_frag;
1888
Shailabh Nagarca74e922006-07-14 00:24:36 -07001889#ifdef CONFIG_TASK_DELAY_ACCT
1890 struct task_delay_info *delays;
1891#endif
Akinobu Mitaf4f154f2006-12-08 02:39:47 -08001892#ifdef CONFIG_FAULT_INJECTION
1893 int make_it_fail;
1894#endif
Wu Fengguang9d823e82011-06-11 18:10:12 -06001895 /*
1896 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1897 * balance_dirty_pages() for some dirty throttling pause
1898 */
1899 int nr_dirtied;
1900 int nr_dirtied_pause;
Wu Fengguang83712352011-06-11 19:25:42 -06001901 unsigned long dirty_paused_when; /* start of a write-and-pause period */
Wu Fengguang9d823e82011-06-11 18:10:12 -06001902
Arjan van de Ven97455122008-01-25 21:08:34 +01001903#ifdef CONFIG_LATENCYTOP
1904 int latency_record_count;
1905 struct latency_record latency_record[LT_SAVECOUNT];
1906#endif
Arjan van de Ven69766752008-09-01 15:52:40 -07001907 /*
1908 * time slack values; these are used to round up poll() and
1909 * select() etc timeout values. These are in nanoseconds.
1910 */
John Stultzda8b44d2016-03-17 14:20:51 -07001911 u64 timer_slack_ns;
1912 u64 default_timer_slack_ns;
David Millerf8d570a2008-11-06 00:37:40 -08001913
Andrey Ryabinin0b24bec2015-02-13 14:39:17 -08001914#ifdef CONFIG_KASAN
1915 unsigned int kasan_depth;
1916#endif
Frederic Weisbeckerfb526072008-11-25 21:07:04 +01001917#ifdef CONFIG_FUNCTION_GRAPH_TRACER
Daniel Mack3ad2f3f2010-02-03 08:01:28 +08001918 /* Index of current stored address in ret_stack */
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001919 int curr_ret_stack;
1920 /* Stack of return addresses for return function tracing */
1921 struct ftrace_ret_stack *ret_stack;
Steven Rostedt8aef2d22009-03-24 01:10:15 -04001922 /* time stamp for last schedule */
1923 unsigned long long ftrace_timestamp;
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001924 /*
1925 * Number of functions that haven't been traced
1926 * because of depth overrun.
1927 */
1928 atomic_t trace_overrun;
Frederic Weisbecker380c4b12008-12-06 03:43:41 +01001929 /* Pause for the tracing */
1930 atomic_t tracing_graph_pause;
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001931#endif
Steven Rostedtea4e2bc2008-12-03 15:36:57 -05001932#ifdef CONFIG_TRACING
1933 /* state flags for use by tracers */
1934 unsigned long trace;
Steven Rostedtb1cff0a2011-05-25 14:27:43 -04001935 /* bitmask and counter of trace recursion */
Steven Rostedt261842b2009-04-16 21:41:52 -04001936 unsigned long trace_recursion;
1937#endif /* CONFIG_TRACING */
Dmitry Vyukov5c9a8752016-03-22 14:27:30 -07001938#ifdef CONFIG_KCOV
1939 /* Coverage collection mode enabled for this task (0 if disabled). */
1940 enum kcov_mode kcov_mode;
1941 /* Size of the kcov_area. */
1942 unsigned kcov_size;
1943 /* Buffer for coverage collection. */
1944 void *kcov_area;
1945 /* kcov desciptor wired with this task or NULL. */
1946 struct kcov *kcov;
1947#endif
Vladimir Davydov6f185c22014-12-12 16:55:15 -08001948#ifdef CONFIG_MEMCG
Tejun Heo626ebc42015-11-05 18:46:09 -08001949 struct mem_cgroup *memcg_in_oom;
1950 gfp_t memcg_oom_gfp_mask;
1951 int memcg_oom_order;
Tejun Heob23afb92015-11-05 18:46:11 -08001952
1953 /* number of pages to reclaim on returning to userland */
1954 unsigned int memcg_nr_pages_over_high;
KAMEZAWA Hiroyuki569b8462009-12-15 16:47:03 -08001955#endif
Srikar Dronamraju0326f5a2012-03-13 23:30:11 +05301956#ifdef CONFIG_UPROBES
1957 struct uprobe_task *utask;
Srikar Dronamraju0326f5a2012-03-13 23:30:11 +05301958#endif
Kent Overstreetcafe5632013-03-23 16:11:31 -07001959#if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1960 unsigned int sequential_io;
1961 unsigned int sequential_io_avg;
1962#endif
Peter Zijlstra8eb23b92014-09-24 10:18:55 +02001963#ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1964 unsigned long task_state_change;
1965#endif
David Hildenbrand8bcbde52015-05-11 17:52:06 +02001966 int pagefault_disabled;
Michal Hocko03049262016-03-25 14:20:33 -07001967#ifdef CONFIG_MMU
Vladimir Davydov29c696e2016-03-25 14:20:39 -07001968 struct task_struct *oom_reaper_list;
Michal Hocko03049262016-03-25 14:20:33 -07001969#endif
Andy Lutomirskiba14a192016-08-11 02:35:21 -07001970#ifdef CONFIG_VMAP_STACK
1971 struct vm_struct *stack_vm_area;
1972#endif
Andy Lutomirski68f24b082016-09-15 22:45:48 -07001973#ifdef CONFIG_THREAD_INFO_IN_TASK
1974 /* A live task holds one reference. */
1975 atomic_t stack_refcount;
1976#endif
Dave Hansen0c8c0f02015-07-17 12:28:11 +02001977/* CPU-specific state of this task */
1978 struct thread_struct thread;
1979/*
1980 * WARNING: on x86, 'thread_struct' contains a variable-sized
1981 * structure. It *MUST* be at the end of 'task_struct'.
1982 *
1983 * Do not put anything below here!
1984 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001985};
1986
Ingo Molnar5aaeb5c2015-07-17 12:28:12 +02001987#ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1988extern int arch_task_struct_size __read_mostly;
1989#else
1990# define arch_task_struct_size (sizeof(struct task_struct))
1991#endif
Dave Hansen0c8c0f02015-07-17 12:28:11 +02001992
Andy Lutomirskiba14a192016-08-11 02:35:21 -07001993#ifdef CONFIG_VMAP_STACK
1994static inline struct vm_struct *task_stack_vm_area(const struct task_struct *t)
1995{
1996 return t->stack_vm_area;
1997}
1998#else
1999static inline struct vm_struct *task_stack_vm_area(const struct task_struct *t)
2000{
2001 return NULL;
2002}
2003#endif
2004
Rusty Russell76e6eee2009-03-12 14:35:43 -06002005/* Future-safe accessor for struct task_struct's cpus_allowed. */
Rusty Russella4636812009-12-17 11:43:29 -06002006#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
Rusty Russell76e6eee2009-03-12 14:35:43 -06002007
Thomas Gleixner50605ff2016-05-11 14:23:31 +02002008static inline int tsk_nr_cpus_allowed(struct task_struct *p)
2009{
2010 return p->nr_cpus_allowed;
2011}
2012
Peter Zijlstra6688cc02013-10-07 11:29:24 +01002013#define TNF_MIGRATED 0x01
2014#define TNF_NO_GROUP 0x02
Rik van Rieldabe1d92013-10-07 11:29:34 +01002015#define TNF_SHARED 0x04
Rik van Riel04bb2f92013-10-07 11:29:36 +01002016#define TNF_FAULT_LOCAL 0x08
Mel Gorman074c2382015-03-25 15:55:42 -07002017#define TNF_MIGRATE_FAIL 0x10
Peter Zijlstra6688cc02013-10-07 11:29:24 +01002018
Michal Hockob18dc5f2016-07-28 15:44:46 -07002019static inline bool in_vfork(struct task_struct *tsk)
2020{
2021 bool ret;
2022
2023 /*
2024 * need RCU to access ->real_parent if CLONE_VM was used along with
2025 * CLONE_PARENT.
2026 *
2027 * We check real_parent->mm == tsk->mm because CLONE_VFORK does not
2028 * imply CLONE_VM
2029 *
2030 * CLONE_VFORK can be used with CLONE_PARENT/CLONE_THREAD and thus
2031 * ->real_parent is not necessarily the task doing vfork(), so in
2032 * theory we can't rely on task_lock() if we want to dereference it.
2033 *
2034 * And in this case we can't trust the real_parent->mm == tsk->mm
2035 * check, it can be false negative. But we do not care, if init or
2036 * another oom-unkillable task does this it should blame itself.
2037 */
2038 rcu_read_lock();
2039 ret = tsk->vfork_done && tsk->real_parent->mm == tsk->mm;
2040 rcu_read_unlock();
2041
2042 return ret;
2043}
2044
Peter Zijlstracbee9f82012-10-25 14:16:43 +02002045#ifdef CONFIG_NUMA_BALANCING
Peter Zijlstra6688cc02013-10-07 11:29:24 +01002046extern void task_numa_fault(int last_node, int node, int pages, int flags);
Mel Gormane29cf082013-10-07 11:29:22 +01002047extern pid_t task_numa_group_id(struct task_struct *p);
Mel Gorman1a687c22012-11-22 11:16:36 +00002048extern void set_numabalancing_state(bool enabled);
Jann Horn837ffc92019-07-16 17:20:45 +02002049extern void task_numa_free(struct task_struct *p, bool final);
Rik van Riel10f39042014-01-27 17:03:44 -05002050extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
2051 int src_nid, int dst_cpu);
Peter Zijlstracbee9f82012-10-25 14:16:43 +02002052#else
Mel Gormanac8e8952013-10-07 11:29:03 +01002053static inline void task_numa_fault(int last_node, int node, int pages,
Peter Zijlstra6688cc02013-10-07 11:29:24 +01002054 int flags)
Peter Zijlstracbee9f82012-10-25 14:16:43 +02002055{
2056}
Mel Gormane29cf082013-10-07 11:29:22 +01002057static inline pid_t task_numa_group_id(struct task_struct *p)
2058{
2059 return 0;
2060}
Mel Gorman1a687c22012-11-22 11:16:36 +00002061static inline void set_numabalancing_state(bool enabled)
2062{
2063}
Jann Horn837ffc92019-07-16 17:20:45 +02002064static inline void task_numa_free(struct task_struct *p, bool final)
Rik van Riel82727012013-10-07 11:29:28 +01002065{
2066}
Rik van Riel10f39042014-01-27 17:03:44 -05002067static inline bool should_numa_migrate_memory(struct task_struct *p,
2068 struct page *page, int src_nid, int dst_cpu)
2069{
2070 return true;
2071}
Peter Zijlstracbee9f82012-10-25 14:16:43 +02002072#endif
2073
Alexey Dobriyane8681712007-10-26 12:17:22 +04002074static inline struct pid *task_pid(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07002075{
2076 return task->pids[PIDTYPE_PID].pid;
2077}
2078
Alexey Dobriyane8681712007-10-26 12:17:22 +04002079static inline struct pid *task_tgid(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07002080{
2081 return task->group_leader->pids[PIDTYPE_PID].pid;
2082}
2083
Oleg Nesterov6dda81f2009-04-02 16:58:35 -07002084/*
2085 * Without tasklist or rcu lock it is not safe to dereference
2086 * the result of task_pgrp/task_session even if task == current,
2087 * we can race with another thread doing sys_setsid/sys_setpgid.
2088 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04002089static inline struct pid *task_pgrp(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07002090{
2091 return task->group_leader->pids[PIDTYPE_PGID].pid;
2092}
2093
Alexey Dobriyane8681712007-10-26 12:17:22 +04002094static inline struct pid *task_session(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07002095{
2096 return task->group_leader->pids[PIDTYPE_SID].pid;
2097}
2098
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002099struct pid_namespace;
2100
2101/*
2102 * the helpers to get the task's different pids as they are seen
2103 * from various namespaces
2104 *
2105 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
Eric W. Biederman44c4e1b2008-02-08 04:19:15 -08002106 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
2107 * current.
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002108 * task_xid_nr_ns() : id seen from the ns specified;
2109 *
2110 * set_task_vxid() : assigns a virtual id to a task;
2111 *
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002112 * see also pid_nr() etc in include/linux/pid.h
2113 */
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07002114pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
2115 struct pid_namespace *ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002116
Alexey Dobriyane8681712007-10-26 12:17:22 +04002117static inline pid_t task_pid_nr(struct task_struct *tsk)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002118{
2119 return tsk->pid;
2120}
2121
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07002122static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
2123 struct pid_namespace *ns)
2124{
2125 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
2126}
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002127
2128static inline pid_t task_pid_vnr(struct task_struct *tsk)
2129{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07002130 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002131}
2132
2133
Alexey Dobriyane8681712007-10-26 12:17:22 +04002134static inline pid_t task_tgid_nr(struct task_struct *tsk)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002135{
2136 return tsk->tgid;
2137}
2138
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002139
Richard Guy Briggs80e0b6e2014-03-16 14:00:19 -04002140static inline int pid_alive(const struct task_struct *p);
Richard Guy Briggsad36d282013-08-15 18:05:12 -04002141
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07002142static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
2143 struct pid_namespace *ns)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002144{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07002145 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002146}
2147
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002148static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
2149{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07002150 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002151}
2152
2153
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07002154static inline pid_t task_session_nr_ns(struct task_struct *tsk,
2155 struct pid_namespace *ns)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002156{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07002157 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002158}
2159
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002160static inline pid_t task_session_vnr(struct task_struct *tsk)
2161{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07002162 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002163}
2164
Oleg Nesterov322cd322017-08-21 17:35:02 +02002165static inline pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
2166{
2167 return __task_pid_nr_ns(tsk, __PIDTYPE_TGID, ns);
2168}
2169
2170static inline pid_t task_tgid_vnr(struct task_struct *tsk)
2171{
2172 return __task_pid_nr_ns(tsk, __PIDTYPE_TGID, NULL);
2173}
2174
2175static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
2176{
2177 pid_t pid = 0;
2178
2179 rcu_read_lock();
2180 if (pid_alive(tsk))
2181 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
2182 rcu_read_unlock();
2183
2184 return pid;
2185}
2186
2187static inline pid_t task_ppid_nr(const struct task_struct *tsk)
2188{
2189 return task_ppid_nr_ns(tsk, &init_pid_ns);
2190}
2191
Oleg Nesterov1b0f7ff2009-04-02 16:58:39 -07002192/* obsolete, do not use */
2193static inline pid_t task_pgrp_nr(struct task_struct *tsk)
2194{
2195 return task_pgrp_nr_ns(tsk, &init_pid_ns);
2196}
Pavel Emelyanov7af57292007-10-18 23:40:06 -07002197
Linus Torvalds1da177e2005-04-16 15:20:36 -07002198/**
2199 * pid_alive - check that a task structure is not stale
2200 * @p: Task structure to be checked.
2201 *
2202 * Test if a process is not yet dead (at most zombie state)
2203 * If pid_alive fails, then pointers within the task structure
2204 * can be stale and must not be dereferenced.
Yacine Belkadie69f6182013-07-12 20:45:47 +02002205 *
2206 * Return: 1 if the process is alive. 0 otherwise.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002207 */
Richard Guy Briggsad36d282013-08-15 18:05:12 -04002208static inline int pid_alive(const struct task_struct *p)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002209{
Eric W. Biederman92476d72006-03-31 02:31:42 -08002210 return p->pids[PIDTYPE_PID].pid != NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002211}
2212
Sukadev Bhattiproluf400e192006-09-29 02:00:07 -07002213/**
Sergey Senozhatsky570f5242016-01-01 23:03:01 +09002214 * is_global_init - check if a task structure is init. Since init
2215 * is free to have sub-threads we need to check tgid.
Henne32602592006-10-06 00:44:01 -07002216 * @tsk: Task structure to be checked.
2217 *
2218 * Check if a task structure is the first user space task the kernel created.
Yacine Belkadie69f6182013-07-12 20:45:47 +02002219 *
2220 * Return: 1 if the task structure is init. 0 otherwise.
Sukadev Bhattiproluf400e192006-09-29 02:00:07 -07002221 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04002222static inline int is_global_init(struct task_struct *tsk)
Pavel Emelyanovb461cc02007-10-18 23:40:09 -07002223{
Sergey Senozhatsky570f5242016-01-01 23:03:01 +09002224 return task_tgid_nr(tsk) == 1;
Pavel Emelyanovb461cc02007-10-18 23:40:09 -07002225}
Serge E. Hallynb460cbc2007-10-18 23:39:52 -07002226
Cedric Le Goater9ec52092006-10-02 02:19:00 -07002227extern struct pid *cad_pid;
2228
Linus Torvalds1da177e2005-04-16 15:20:36 -07002229extern void free_task(struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002230#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
Ingo Molnare56d0902006-01-08 01:01:37 -08002231
Andrew Morton158d9eb2006-03-31 02:31:34 -08002232extern void __put_task_struct(struct task_struct *t);
Ingo Molnare56d0902006-01-08 01:01:37 -08002233
2234static inline void put_task_struct(struct task_struct *t)
2235{
2236 if (atomic_dec_and_test(&t->usage))
Eric W. Biederman8c7904a2006-03-31 02:31:37 -08002237 __put_task_struct(t);
Ingo Molnare56d0902006-01-08 01:01:37 -08002238}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002239
Oleg Nesterov150593b2016-05-18 19:02:18 +02002240struct task_struct *task_rcu_dereference(struct task_struct **ptask);
2241struct task_struct *try_get_task_struct(struct task_struct **ptask);
2242
Frederic Weisbecker6a616712012-12-16 20:00:34 +01002243#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2244extern void task_cputime(struct task_struct *t,
2245 cputime_t *utime, cputime_t *stime);
2246extern void task_cputime_scaled(struct task_struct *t,
2247 cputime_t *utimescaled, cputime_t *stimescaled);
2248extern cputime_t task_gtime(struct task_struct *t);
2249#else
Frederic Weisbecker6fac4822012-11-13 14:20:55 +01002250static inline void task_cputime(struct task_struct *t,
2251 cputime_t *utime, cputime_t *stime)
2252{
2253 if (utime)
2254 *utime = t->utime;
2255 if (stime)
2256 *stime = t->stime;
2257}
2258
2259static inline void task_cputime_scaled(struct task_struct *t,
2260 cputime_t *utimescaled,
2261 cputime_t *stimescaled)
2262{
2263 if (utimescaled)
2264 *utimescaled = t->utimescaled;
2265 if (stimescaled)
2266 *stimescaled = t->stimescaled;
2267}
Frederic Weisbecker6a616712012-12-16 20:00:34 +01002268
2269static inline cputime_t task_gtime(struct task_struct *t)
2270{
2271 return t->gtime;
2272}
2273#endif
Frederic Weisbeckere80d0a12012-11-21 16:26:44 +01002274extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2275extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
Balbir Singh49048622008-09-05 18:12:23 +02002276
Linus Torvalds1da177e2005-04-16 15:20:36 -07002277/*
2278 * Per process flags
2279 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002280#define PF_EXITING 0x00000004 /* getting shut down */
Laurent Vivier94886b82007-10-15 17:00:19 +02002281#define PF_VCPU 0x00000010 /* I'm a virtual CPU */
Tejun Heo21aa9af2010-06-08 21:40:37 +02002282#define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002283#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
Andi Kleen4db96cf2009-09-16 11:50:14 +02002284#define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002285#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2286#define PF_DUMPCORE 0x00000200 /* dumped core */
2287#define PF_SIGNALED 0x00000400 /* killed by a signal */
2288#define PF_MEMALLOC 0x00000800 /* Allocating memory */
Vasiliy Kulikov72fa5992011-08-08 19:02:04 +04002289#define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002290#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
Tejun Heo774a1222013-01-15 18:52:51 -08002291#define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002292#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2293#define PF_FROZEN 0x00010000 /* frozen for system suspend */
2294#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2295#define PF_KSWAPD 0x00040000 /* I am kswapd */
Ming Lei21caf2f2013-02-22 16:34:08 -08002296#define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002297#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
Oleg Nesterov246bb0b2008-07-25 01:47:38 -07002298#define PF_KTHREAD 0x00200000 /* I am a kernel thread */
Jens Axboeb31dc662006-06-13 08:26:10 +02002299#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2300#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
Tejun Heo14a40ff2013-03-19 13:45:20 -07002301#define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
Andi Kleen4db96cf2009-09-16 11:50:14 +02002302#define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
Thomas Gleixner61a87122006-06-27 02:54:56 -07002303#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
Tejun Heo58a69cb2011-02-16 09:25:31 +01002304#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
Colin Cross2b44c4d2013-07-24 17:41:33 -07002305#define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002306
2307/*
2308 * Only the _current_ task can read/write to tsk->flags, but other
2309 * tasks can access tsk->flags in readonly mode for example
2310 * with tsk_used_math (like during threaded core dumping).
2311 * There is however an exception to this rule during ptrace
2312 * or during fork: the ptracer task is allowed to write to the
2313 * child->flags of its traced child (same goes for fork, the parent
2314 * can write to the child->flags), because we're guaranteed the
2315 * child is not running and in turn not changing child->flags
2316 * at the same time the parent does it.
2317 */
2318#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2319#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2320#define clear_used_math() clear_stopped_child_used_math(current)
2321#define set_used_math() set_stopped_child_used_math(current)
2322#define conditional_stopped_child_used_math(condition, child) \
2323 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2324#define conditional_used_math(condition) \
2325 conditional_stopped_child_used_math(condition, current)
2326#define copy_to_stopped_child_used_math(child) \
2327 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2328/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2329#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2330#define used_math() tsk_used_math(current)
2331
Junxiao Bi934f30722014-10-09 15:28:23 -07002332/* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2333 * __GFP_FS is also cleared as it implies __GFP_IO.
2334 */
Ming Lei21caf2f2013-02-22 16:34:08 -08002335static inline gfp_t memalloc_noio_flags(gfp_t flags)
2336{
2337 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
Junxiao Bi934f30722014-10-09 15:28:23 -07002338 flags &= ~(__GFP_IO | __GFP_FS);
Ming Lei21caf2f2013-02-22 16:34:08 -08002339 return flags;
2340}
2341
2342static inline unsigned int memalloc_noio_save(void)
2343{
2344 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2345 current->flags |= PF_MEMALLOC_NOIO;
2346 return flags;
2347}
2348
2349static inline void memalloc_noio_restore(unsigned int flags)
2350{
2351 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2352}
2353
Kees Cook1d4457f2014-05-21 15:23:46 -07002354/* Per-process atomic flags. */
Zefan Lia2b86f72014-09-25 09:40:17 +08002355#define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
Zefan Li2ad654b2014-09-25 09:41:02 +08002356#define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2357#define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
Tetsuo Handa77ed2c52016-03-08 20:01:32 +09002358#define PFA_LMK_WAITING 3 /* Lowmemorykiller is waiting */
Thomas Gleixner036608d2018-05-03 22:09:15 +02002359#define PFA_SPEC_SSB_DISABLE 4 /* Speculative Store Bypass disabled */
2360#define PFA_SPEC_SSB_FORCE_DISABLE 5 /* Speculative Store Bypass force disabled*/
Thomas Gleixner2d99bc02018-11-25 19:33:53 +01002361#define PFA_SPEC_IB_DISABLE 6 /* Indirect branch speculation restricted */
2362#define PFA_SPEC_IB_FORCE_DISABLE 7 /* Indirect branch speculation permanently restricted */
Kees Cook1d4457f2014-05-21 15:23:46 -07002363
Kees Cook1d4457f2014-05-21 15:23:46 -07002364
Zefan Lie0e50702014-09-25 09:40:40 +08002365#define TASK_PFA_TEST(name, func) \
2366 static inline bool task_##func(struct task_struct *p) \
2367 { return test_bit(PFA_##name, &p->atomic_flags); }
2368#define TASK_PFA_SET(name, func) \
2369 static inline void task_set_##func(struct task_struct *p) \
2370 { set_bit(PFA_##name, &p->atomic_flags); }
2371#define TASK_PFA_CLEAR(name, func) \
2372 static inline void task_clear_##func(struct task_struct *p) \
2373 { clear_bit(PFA_##name, &p->atomic_flags); }
Kees Cook1d4457f2014-05-21 15:23:46 -07002374
Zefan Lie0e50702014-09-25 09:40:40 +08002375TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2376TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
Kees Cook1d4457f2014-05-21 15:23:46 -07002377
Zefan Li2ad654b2014-09-25 09:41:02 +08002378TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2379TASK_PFA_SET(SPREAD_PAGE, spread_page)
2380TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2381
2382TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2383TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2384TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
Tejun Heo544b2c92011-06-14 11:20:18 +02002385
Tetsuo Handa77ed2c52016-03-08 20:01:32 +09002386TASK_PFA_TEST(LMK_WAITING, lmk_waiting)
2387TASK_PFA_SET(LMK_WAITING, lmk_waiting)
2388
Thomas Gleixner036608d2018-05-03 22:09:15 +02002389TASK_PFA_TEST(SPEC_SSB_DISABLE, spec_ssb_disable)
2390TASK_PFA_SET(SPEC_SSB_DISABLE, spec_ssb_disable)
2391TASK_PFA_CLEAR(SPEC_SSB_DISABLE, spec_ssb_disable)
2392
2393TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
2394TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
2395
Thomas Gleixner2d99bc02018-11-25 19:33:53 +01002396TASK_PFA_TEST(SPEC_IB_DISABLE, spec_ib_disable)
2397TASK_PFA_SET(SPEC_IB_DISABLE, spec_ib_disable)
2398TASK_PFA_CLEAR(SPEC_IB_DISABLE, spec_ib_disable)
2399
2400TASK_PFA_TEST(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable)
2401TASK_PFA_SET(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable)
2402
Tejun Heo39efa3e2011-03-23 10:37:00 +01002403/*
Tejun Heoe5c19022011-03-23 10:37:00 +01002404 * task->jobctl flags
Tejun Heod79fdd62011-03-23 10:37:00 +01002405 */
Oleg Nesterovee77f072011-04-01 20:12:38 +02002406#define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
Tejun Heoe5c19022011-03-23 10:37:00 +01002407
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07002408#define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
Tejun Heoa8f072c2011-06-02 11:13:59 +02002409#define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2410#define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2411#define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
Tejun Heo73ddff22011-06-14 11:20:14 +02002412#define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
Tejun Heofb1d9102011-06-14 11:20:17 +02002413#define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
Tejun Heoa8f072c2011-06-02 11:13:59 +02002414#define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2415
Palmer Dabbeltb76808e2015-04-30 21:19:57 -07002416#define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2417#define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2418#define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2419#define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2420#define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2421#define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2422#define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
Tejun Heoa8f072c2011-06-02 11:13:59 +02002423
Tejun Heofb1d9102011-06-14 11:20:17 +02002424#define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
Tejun Heo73ddff22011-06-14 11:20:14 +02002425#define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
Tejun Heo3759a0d2011-06-02 11:14:00 +02002426
Tejun Heo7dd3db52011-06-02 11:14:00 +02002427extern bool task_set_jobctl_pending(struct task_struct *task,
Palmer Dabbeltb76808e2015-04-30 21:19:57 -07002428 unsigned long mask);
Tejun Heo73ddff22011-06-14 11:20:14 +02002429extern void task_clear_jobctl_trapping(struct task_struct *task);
Tejun Heo3759a0d2011-06-02 11:14:00 +02002430extern void task_clear_jobctl_pending(struct task_struct *task,
Palmer Dabbeltb76808e2015-04-30 21:19:57 -07002431 unsigned long mask);
Tejun Heo39efa3e2011-03-23 10:37:00 +01002432
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07002433static inline void rcu_copy_process(struct task_struct *p)
2434{
Linus Torvalds1da177e2005-04-16 15:20:36 -07002435#ifdef CONFIG_PREEMPT_RCU
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07002436 p->rcu_read_lock_nesting = 0;
Paul E. McKenney1d082fd2014-08-14 16:01:53 -07002437 p->rcu_read_unlock_special.s = 0;
Paul E. McKenneydd5d19b2009-08-27 14:58:16 -07002438 p->rcu_blocked_node = NULL;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07002439 INIT_LIST_HEAD(&p->rcu_node_entry);
Paul E. McKenney8315f422014-06-27 13:42:20 -07002440#endif /* #ifdef CONFIG_PREEMPT_RCU */
2441#ifdef CONFIG_TASKS_RCU
2442 p->rcu_tasks_holdout = false;
2443 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
Paul E. McKenney176f8f72014-08-04 17:43:50 -07002444 p->rcu_tasks_idle_cpu = -1;
Paul E. McKenney8315f422014-06-27 13:42:20 -07002445#endif /* #ifdef CONFIG_TASKS_RCU */
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07002446}
2447
Mel Gorman907aed42012-07-31 16:44:07 -07002448static inline void tsk_restore_flags(struct task_struct *task,
2449 unsigned long orig_flags, unsigned long flags)
2450{
2451 task->flags &= ~flags;
2452 task->flags |= orig_flags & flags;
2453}
2454
Juri Lellif82f8042014-10-07 09:52:11 +01002455extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2456 const struct cpumask *trial);
Juri Lelli7f514122014-09-19 10:22:40 +01002457extern int task_can_attach(struct task_struct *p,
2458 const struct cpumask *cs_cpus_allowed);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002459#ifdef CONFIG_SMP
KOSAKI Motohiro1e1b6c52011-05-19 15:08:58 +09002460extern void do_set_cpus_allowed(struct task_struct *p,
2461 const struct cpumask *new_mask);
2462
Mike Traviscd8ba7c2008-03-26 14:23:49 -07002463extern int set_cpus_allowed_ptr(struct task_struct *p,
Rusty Russell96f874e2008-11-25 02:35:14 +10302464 const struct cpumask *new_mask);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002465#else
KOSAKI Motohiro1e1b6c52011-05-19 15:08:58 +09002466static inline void do_set_cpus_allowed(struct task_struct *p,
2467 const struct cpumask *new_mask)
2468{
2469}
Mike Traviscd8ba7c2008-03-26 14:23:49 -07002470static inline int set_cpus_allowed_ptr(struct task_struct *p,
Rusty Russell96f874e2008-11-25 02:35:14 +10302471 const struct cpumask *new_mask)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002472{
Rusty Russell96f874e2008-11-25 02:35:14 +10302473 if (!cpumask_test_cpu(0, new_mask))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002474 return -EINVAL;
2475 return 0;
2476}
2477#endif
Rusty Russelle0ad9552009-09-24 09:34:38 -06002478
Frederic Weisbecker3451d022011-08-10 23:21:01 +02002479#ifdef CONFIG_NO_HZ_COMMON
Peter Zijlstra5167e8d2012-06-22 15:52:09 +02002480void calc_load_enter_idle(void);
2481void calc_load_exit_idle(void);
2482#else
2483static inline void calc_load_enter_idle(void) { }
2484static inline void calc_load_exit_idle(void) { }
Frederic Weisbecker3451d022011-08-10 23:21:01 +02002485#endif /* CONFIG_NO_HZ_COMMON */
Peter Zijlstra5167e8d2012-06-22 15:52:09 +02002486
Ingo Molnarb3425012009-02-26 20:20:29 +01002487/*
Peter Zijlstrac6763292010-05-25 10:48:51 +02002488 * Do not use outside of architecture code which knows its limitations.
2489 *
2490 * sched_clock() has no promise of monotonicity or bounded drift between
2491 * CPUs, use (which you should not) requires disabling IRQs.
2492 *
2493 * Please use one of the three interfaces below.
Ingo Molnarb3425012009-02-26 20:20:29 +01002494 */
Mike Frysinger1bbfa6f2009-12-09 20:07:03 -05002495extern unsigned long long notrace sched_clock(void);
Peter Zijlstrac6763292010-05-25 10:48:51 +02002496/*
Hiroshi Shimamoto489a71b2012-04-02 17:00:44 +09002497 * See the comment in kernel/sched/clock.c
Peter Zijlstrac6763292010-05-25 10:48:51 +02002498 */
Cyril Bur545a2bf2015-02-12 15:01:24 -08002499extern u64 running_clock(void);
Peter Zijlstrac6763292010-05-25 10:48:51 +02002500extern u64 sched_clock_cpu(int cpu);
2501
Ingo Molnare436d802007-07-19 21:28:35 +02002502
Peter Zijlstrac1955a32008-08-11 08:59:03 +02002503extern void sched_clock_init(void);
Peter Zijlstrac1955a32008-08-11 08:59:03 +02002504
Peter Zijlstra3e51f332008-05-03 18:29:28 +02002505#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
Peter Zijlstra3e51f332008-05-03 18:29:28 +02002506static inline void sched_clock_tick(void)
2507{
2508}
2509
2510static inline void sched_clock_idle_sleep_event(void)
2511{
2512}
2513
2514static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2515{
2516}
Daniel Lezcano2c923e92016-04-11 16:38:34 +02002517
2518static inline u64 cpu_clock(int cpu)
2519{
2520 return sched_clock();
2521}
2522
2523static inline u64 local_clock(void)
2524{
2525 return sched_clock();
2526}
Peter Zijlstra3e51f332008-05-03 18:29:28 +02002527#else
Peter Zijlstrac6763292010-05-25 10:48:51 +02002528/*
2529 * Architectures can set this to 1 if they have specified
2530 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2531 * but then during bootup it turns out that sched_clock()
2532 * is reliable after all:
2533 */
Peter Zijlstra35af99e2013-11-28 19:38:42 +01002534extern int sched_clock_stable(void);
2535extern void set_sched_clock_stable(void);
2536extern void clear_sched_clock_stable(void);
Peter Zijlstrac6763292010-05-25 10:48:51 +02002537
Peter Zijlstra3e51f332008-05-03 18:29:28 +02002538extern void sched_clock_tick(void);
2539extern void sched_clock_idle_sleep_event(void);
2540extern void sched_clock_idle_wakeup_event(u64 delta_ns);
Daniel Lezcano2c923e92016-04-11 16:38:34 +02002541
2542/*
2543 * As outlined in clock.c, provides a fast, high resolution, nanosecond
2544 * time source that is monotonic per cpu argument and has bounded drift
2545 * between cpus.
2546 *
2547 * ######################### BIG FAT WARNING ##########################
2548 * # when comparing cpu_clock(i) to cpu_clock(j) for i != j, time can #
2549 * # go backwards !! #
2550 * ####################################################################
2551 */
2552static inline u64 cpu_clock(int cpu)
2553{
2554 return sched_clock_cpu(cpu);
2555}
2556
2557static inline u64 local_clock(void)
2558{
2559 return sched_clock_cpu(raw_smp_processor_id());
2560}
Peter Zijlstra3e51f332008-05-03 18:29:28 +02002561#endif
2562
Venkatesh Pallipadib52bfee2010-10-04 17:03:19 -07002563#ifdef CONFIG_IRQ_TIME_ACCOUNTING
2564/*
2565 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2566 * The reason for this explicit opt-in is not to have perf penalty with
2567 * slow sched_clocks.
2568 */
2569extern void enable_sched_clock_irqtime(void);
2570extern void disable_sched_clock_irqtime(void);
2571#else
2572static inline void enable_sched_clock_irqtime(void) {}
2573static inline void disable_sched_clock_irqtime(void) {}
2574#endif
2575
Ingo Molnar36c8b582006-07-03 00:25:41 -07002576extern unsigned long long
Ingo Molnar41b86e92007-07-09 18:51:58 +02002577task_sched_runtime(struct task_struct *task);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002578
2579/* sched_exec is called by processes performing an exec */
2580#ifdef CONFIG_SMP
2581extern void sched_exec(void);
2582#else
2583#define sched_exec() {}
2584#endif
2585
Ingo Molnar2aa44d02007-08-23 15:18:02 +02002586extern void sched_clock_idle_sleep_event(void);
2587extern void sched_clock_idle_wakeup_event(u64 delta_ns);
Ingo Molnarbb29ab22007-07-09 18:51:59 +02002588
Linus Torvalds1da177e2005-04-16 15:20:36 -07002589#ifdef CONFIG_HOTPLUG_CPU
2590extern void idle_task_exit(void);
2591#else
2592static inline void idle_task_exit(void) {}
2593#endif
2594
Frederic Weisbecker3451d022011-08-10 23:21:01 +02002595#if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
Frederic Weisbecker1c200912011-08-10 23:21:01 +02002596extern void wake_up_nohz_cpu(int cpu);
Thomas Gleixner06d83082008-03-22 09:20:24 +01002597#else
Frederic Weisbecker1c200912011-08-10 23:21:01 +02002598static inline void wake_up_nohz_cpu(int cpu) { }
Thomas Gleixner06d83082008-03-22 09:20:24 +01002599#endif
2600
Frederic Weisbeckerce831b32013-04-20 15:15:35 +02002601#ifdef CONFIG_NO_HZ_FULL
Frederic Weisbecker265f22a2013-05-03 03:39:05 +02002602extern u64 scheduler_tick_max_deferment(void);
Ingo Molnarbf0f6f22007-07-09 18:51:58 +02002603#endif
Ingo Molnarbf0f6f22007-07-09 18:51:58 +02002604
Mike Galbraith5091faa2010-11-30 14:18:03 +01002605#ifdef CONFIG_SCHED_AUTOGROUP
Mike Galbraith5091faa2010-11-30 14:18:03 +01002606extern void sched_autogroup_create_attach(struct task_struct *p);
2607extern void sched_autogroup_detach(struct task_struct *p);
2608extern void sched_autogroup_fork(struct signal_struct *sig);
2609extern void sched_autogroup_exit(struct signal_struct *sig);
Oleg Nesterov8e5bfa82016-11-14 19:46:12 +01002610extern void sched_autogroup_exit_task(struct task_struct *p);
Mike Galbraith5091faa2010-11-30 14:18:03 +01002611#ifdef CONFIG_PROC_FS
2612extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
Hiroshi Shimamoto2e5b5b32012-02-23 17:41:27 +09002613extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
Mike Galbraith5091faa2010-11-30 14:18:03 +01002614#endif
2615#else
2616static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2617static inline void sched_autogroup_detach(struct task_struct *p) { }
2618static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2619static inline void sched_autogroup_exit(struct signal_struct *sig) { }
Oleg Nesterov8e5bfa82016-11-14 19:46:12 +01002620static inline void sched_autogroup_exit_task(struct task_struct *p) { }
Mike Galbraith5091faa2010-11-30 14:18:03 +01002621#endif
2622
Dan Carpenterfa933842014-05-23 13:20:42 +03002623extern int yield_to(struct task_struct *p, bool preempt);
Ingo Molnar36c8b582006-07-03 00:25:41 -07002624extern void set_user_nice(struct task_struct *p, long nice);
2625extern int task_prio(const struct task_struct *p);
Dongsheng Yangd0ea0262014-01-27 22:00:45 -05002626/**
2627 * task_nice - return the nice value of a given task.
2628 * @p: the task in question.
2629 *
2630 * Return: The nice value [ -20 ... 0 ... 19 ].
2631 */
2632static inline int task_nice(const struct task_struct *p)
2633{
2634 return PRIO_TO_NICE((p)->static_prio);
2635}
Ingo Molnar36c8b582006-07-03 00:25:41 -07002636extern int can_nice(const struct task_struct *p, const int nice);
2637extern int task_curr(const struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002638extern int idle_cpu(int cpu);
KOSAKI Motohirofe7de492010-10-20 16:01:12 -07002639extern int sched_setscheduler(struct task_struct *, int,
2640 const struct sched_param *);
Rusty Russell961ccdd2008-06-23 13:55:38 +10002641extern int sched_setscheduler_nocheck(struct task_struct *, int,
KOSAKI Motohirofe7de492010-10-20 16:01:12 -07002642 const struct sched_param *);
Dario Faggiolid50dde52013-11-07 14:43:36 +01002643extern int sched_setattr(struct task_struct *,
2644 const struct sched_attr *);
Ingo Molnar36c8b582006-07-03 00:25:41 -07002645extern struct task_struct *idle_task(int cpu);
Paul E. McKenneyc4f30602011-11-10 12:41:56 -08002646/**
2647 * is_idle_task - is the specified task an idle task?
Randy Dunlapfa757282012-01-21 11:03:13 -08002648 * @p: the task in question.
Yacine Belkadie69f6182013-07-12 20:45:47 +02002649 *
2650 * Return: 1 if @p is an idle task. 0 otherwise.
Paul E. McKenneyc4f30602011-11-10 12:41:56 -08002651 */
Paul E. McKenney7061ca32011-12-20 08:20:46 -08002652static inline bool is_idle_task(const struct task_struct *p)
Paul E. McKenneyc4f30602011-11-10 12:41:56 -08002653{
2654 return p->pid == 0;
2655}
Ingo Molnar36c8b582006-07-03 00:25:41 -07002656extern struct task_struct *curr_task(int cpu);
Peter Zijlstraa458ae22016-09-20 20:29:40 +02002657extern void ia64_set_curr_task(int cpu, struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002658
2659void yield(void);
2660
Linus Torvalds1da177e2005-04-16 15:20:36 -07002661union thread_union {
Andy Lutomirskic65eacb2016-09-13 14:29:24 -07002662#ifndef CONFIG_THREAD_INFO_IN_TASK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002663 struct thread_info thread_info;
Andy Lutomirskic65eacb2016-09-13 14:29:24 -07002664#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002665 unsigned long stack[THREAD_SIZE/sizeof(long)];
2666};
2667
2668#ifndef __HAVE_ARCH_KSTACK_END
2669static inline int kstack_end(void *addr)
2670{
2671 /* Reliable end of stack detection:
2672 * Some APM bios versions misalign the stack
2673 */
2674 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2675}
2676#endif
2677
2678extern union thread_union init_thread_union;
2679extern struct task_struct init_task;
2680
2681extern struct mm_struct init_mm;
2682
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002683extern struct pid_namespace init_pid_ns;
2684
2685/*
2686 * find a task by one of its numerical ids
2687 *
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002688 * find_task_by_pid_ns():
2689 * finds a task by its pid in the specified namespace
Pavel Emelyanov228ebcb2007-10-18 23:40:16 -07002690 * find_task_by_vpid():
2691 * finds a task by its virtual pid
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002692 *
Pavel Emelyanove49859e2008-07-25 01:48:36 -07002693 * see also find_vpid() etc in include/linux/pid.h
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002694 */
2695
Pavel Emelyanov228ebcb2007-10-18 23:40:16 -07002696extern struct task_struct *find_task_by_vpid(pid_t nr);
2697extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2698 struct pid_namespace *ns);
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002699
Linus Torvalds1da177e2005-04-16 15:20:36 -07002700/* per-UID process charging. */
Eric W. Biederman7b44ab92011-11-16 23:20:58 -08002701extern struct user_struct * alloc_uid(kuid_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002702static inline struct user_struct *get_uid(struct user_struct *u)
2703{
2704 atomic_inc(&u->__count);
2705 return u;
2706}
2707extern void free_uid(struct user_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002708
2709#include <asm/current.h>
2710
Torben Hohnf0af911a92011-01-27 15:59:10 +01002711extern void xtime_update(unsigned long ticks);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002712
Harvey Harrisonb3c97522008-02-13 15:03:15 -08002713extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2714extern int wake_up_process(struct task_struct *tsk);
Samir Bellabes3e51e3e2011-05-11 18:18:05 +02002715extern void wake_up_new_task(struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002716#ifdef CONFIG_SMP
2717 extern void kick_process(struct task_struct *tsk);
2718#else
2719 static inline void kick_process(struct task_struct *tsk) { }
2720#endif
Dario Faggioliaab03e02013-11-28 11:14:43 +01002721extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
Ingo Molnarad46c2c2007-07-09 18:52:00 +02002722extern void sched_dead(struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002723
Linus Torvalds1da177e2005-04-16 15:20:36 -07002724extern void proc_caches_init(void);
2725extern void flush_signals(struct task_struct *);
Oleg Nesterov10ab8252007-05-09 02:34:37 -07002726extern void ignore_signals(struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002727extern void flush_signal_handlers(struct task_struct *, int force_default);
2728extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2729
Oleg Nesterovbe0e6f22015-11-06 16:32:22 -08002730static inline int kernel_dequeue_signal(siginfo_t *info)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002731{
Oleg Nesterovbe0e6f22015-11-06 16:32:22 -08002732 struct task_struct *tsk = current;
2733 siginfo_t __info;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002734 int ret;
2735
Oleg Nesterovbe0e6f22015-11-06 16:32:22 -08002736 spin_lock_irq(&tsk->sighand->siglock);
2737 ret = dequeue_signal(tsk, &tsk->blocked, info ?: &__info);
2738 spin_unlock_irq(&tsk->sighand->siglock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002739
2740 return ret;
Oleg Nesterov53c8f9f2011-06-22 23:08:18 +02002741}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002742
Oleg Nesterov9a130492015-11-06 16:32:25 -08002743static inline void kernel_signal_stop(void)
2744{
2745 spin_lock_irq(&current->sighand->siglock);
2746 if (current->jobctl & JOBCTL_STOP_DEQUEUED)
2747 __set_current_state(TASK_STOPPED);
2748 spin_unlock_irq(&current->sighand->siglock);
2749
2750 schedule();
2751}
2752
Linus Torvalds1da177e2005-04-16 15:20:36 -07002753extern void release_task(struct task_struct * p);
2754extern int send_sig_info(int, struct siginfo *, struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002755extern int force_sigsegv(int, struct task_struct *);
2756extern int force_sig_info(int, struct siginfo *, struct task_struct *);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002757extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002758extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
Serge Hallynd178bc32011-09-26 10:45:18 -05002759extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2760 const struct cred *, u32);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002761extern int kill_pgrp(struct pid *pid, int sig, int priv);
2762extern int kill_pid(struct pid *pid, int sig, int priv);
Matthew Wilcoxc3de4b32007-02-09 08:11:47 -07002763extern int kill_proc_info(int, struct siginfo *, pid_t);
Oleg Nesterov86773472011-06-22 23:09:09 +02002764extern __must_check bool do_notify_parent(struct task_struct *, int);
Oleg Nesterova7f07652009-09-23 15:56:44 -07002765extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002766extern void force_sig(int, struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002767extern int send_sig(int, struct task_struct *, int);
Oleg Nesterov09faef12010-05-26 14:43:11 -07002768extern int zap_other_threads(struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002769extern struct sigqueue *sigqueue_alloc(void);
2770extern void sigqueue_free(struct sigqueue *);
Oleg Nesterovac5c2152008-04-30 00:52:57 -07002771extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
Oleg Nesterov9ac95f22006-02-09 22:41:50 +03002772extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002773
Andy Lutomirski7e781412016-08-02 14:05:36 -07002774#ifdef TIF_RESTORE_SIGMASK
2775/*
2776 * Legacy restore_sigmask accessors. These are inefficient on
2777 * SMP architectures because they require atomic operations.
2778 */
2779
2780/**
2781 * set_restore_sigmask() - make sure saved_sigmask processing gets done
2782 *
2783 * This sets TIF_RESTORE_SIGMASK and ensures that the arch signal code
2784 * will run before returning to user mode, to process the flag. For
2785 * all callers, TIF_SIGPENDING is already set or it's no harm to set
2786 * it. TIF_RESTORE_SIGMASK need not be in the set of bits that the
2787 * arch code will notice on return to user mode, in case those bits
2788 * are scarce. We set TIF_SIGPENDING here to ensure that the arch
2789 * signal code always gets run when TIF_RESTORE_SIGMASK is set.
2790 */
2791static inline void set_restore_sigmask(void)
2792{
2793 set_thread_flag(TIF_RESTORE_SIGMASK);
2794 WARN_ON(!test_thread_flag(TIF_SIGPENDING));
2795}
2796static inline void clear_restore_sigmask(void)
2797{
2798 clear_thread_flag(TIF_RESTORE_SIGMASK);
2799}
2800static inline bool test_restore_sigmask(void)
2801{
2802 return test_thread_flag(TIF_RESTORE_SIGMASK);
2803}
2804static inline bool test_and_clear_restore_sigmask(void)
2805{
2806 return test_and_clear_thread_flag(TIF_RESTORE_SIGMASK);
2807}
2808
2809#else /* TIF_RESTORE_SIGMASK */
2810
2811/* Higher-quality implementation, used if TIF_RESTORE_SIGMASK doesn't exist. */
2812static inline void set_restore_sigmask(void)
2813{
2814 current->restore_sigmask = true;
2815 WARN_ON(!test_thread_flag(TIF_SIGPENDING));
2816}
2817static inline void clear_restore_sigmask(void)
2818{
2819 current->restore_sigmask = false;
2820}
2821static inline bool test_restore_sigmask(void)
2822{
2823 return current->restore_sigmask;
2824}
2825static inline bool test_and_clear_restore_sigmask(void)
2826{
2827 if (!current->restore_sigmask)
2828 return false;
2829 current->restore_sigmask = false;
2830 return true;
2831}
2832#endif
2833
Al Viro51a7b442012-05-21 23:33:55 -04002834static inline void restore_saved_sigmask(void)
2835{
2836 if (test_and_clear_restore_sigmask())
Al Viro77097ae2012-04-27 13:58:59 -04002837 __set_current_blocked(&current->saved_sigmask);
Al Viro51a7b442012-05-21 23:33:55 -04002838}
2839
Al Virob7f9a112012-05-02 09:59:21 -04002840static inline sigset_t *sigmask_to_save(void)
2841{
2842 sigset_t *res = &current->blocked;
2843 if (unlikely(test_restore_sigmask()))
2844 res = &current->saved_sigmask;
2845 return res;
2846}
2847
Cedric Le Goater9ec52092006-10-02 02:19:00 -07002848static inline int kill_cad_pid(int sig, int priv)
2849{
2850 return kill_pid(cad_pid, sig, priv);
2851}
2852
Linus Torvalds1da177e2005-04-16 15:20:36 -07002853/* These can be the second arg to send_sig_info/send_group_sig_info. */
2854#define SEND_SIG_NOINFO ((struct siginfo *) 0)
2855#define SEND_SIG_PRIV ((struct siginfo *) 1)
2856#define SEND_SIG_FORCED ((struct siginfo *) 2)
2857
Sebastian Andrzej Siewior2a855dd2009-10-25 15:37:58 +01002858/*
2859 * True if we are on the alternate signal stack.
2860 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002861static inline int on_sig_stack(unsigned long sp)
2862{
Andy Lutomirskic876eea2016-05-03 10:31:49 -07002863 /*
2864 * If the signal stack is SS_AUTODISARM then, by construction, we
2865 * can't be on the signal stack unless user code deliberately set
2866 * SS_AUTODISARM when we were already on it.
2867 *
2868 * This improves reliability: if user state gets corrupted such that
2869 * the stack pointer points very close to the end of the signal stack,
2870 * then this check will enable the signal to be handled anyway.
2871 */
2872 if (current->sas_ss_flags & SS_AUTODISARM)
2873 return 0;
2874
Sebastian Andrzej Siewior2a855dd2009-10-25 15:37:58 +01002875#ifdef CONFIG_STACK_GROWSUP
2876 return sp >= current->sas_ss_sp &&
2877 sp - current->sas_ss_sp < current->sas_ss_size;
2878#else
2879 return sp > current->sas_ss_sp &&
2880 sp - current->sas_ss_sp <= current->sas_ss_size;
2881#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002882}
2883
2884static inline int sas_ss_flags(unsigned long sp)
2885{
Richard Weinberger72f15c02014-03-05 15:15:22 +01002886 if (!current->sas_ss_size)
2887 return SS_DISABLE;
2888
2889 return on_sig_stack(sp) ? SS_ONSTACK : 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002890}
2891
Stas Sergeev2a742132016-04-14 23:20:04 +03002892static inline void sas_ss_reset(struct task_struct *p)
2893{
2894 p->sas_ss_sp = 0;
2895 p->sas_ss_size = 0;
2896 p->sas_ss_flags = SS_DISABLE;
2897}
2898
Al Viro5a1b98d2012-11-06 13:28:21 -05002899static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2900{
2901 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2902#ifdef CONFIG_STACK_GROWSUP
2903 return current->sas_ss_sp;
2904#else
2905 return current->sas_ss_sp + current->sas_ss_size;
2906#endif
2907 return sp;
2908}
2909
Linus Torvalds1da177e2005-04-16 15:20:36 -07002910/*
2911 * Routines for handling mm_structs
2912 */
2913extern struct mm_struct * mm_alloc(void);
2914
2915/* mmdrop drops the mm and the page tables */
Harvey Harrisonb3c97522008-02-13 15:03:15 -08002916extern void __mmdrop(struct mm_struct *);
Oleg Nesterovd2005e32016-05-20 16:58:36 -07002917static inline void mmdrop(struct mm_struct *mm)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002918{
Ingo Molnar6fb43d72007-07-09 18:52:01 +02002919 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002920 __mmdrop(mm);
2921}
2922
Michal Hocko7283094ec2016-10-07 16:58:54 -07002923static inline void mmdrop_async_fn(struct work_struct *work)
2924{
2925 struct mm_struct *mm = container_of(work, struct mm_struct, async_put_work);
2926 __mmdrop(mm);
2927}
2928
2929static inline void mmdrop_async(struct mm_struct *mm)
2930{
2931 if (unlikely(atomic_dec_and_test(&mm->mm_count))) {
2932 INIT_WORK(&mm->async_put_work, mmdrop_async_fn);
2933 schedule_work(&mm->async_put_work);
2934 }
2935}
2936
Oleg Nesterovd2005e32016-05-20 16:58:36 -07002937static inline bool mmget_not_zero(struct mm_struct *mm)
2938{
2939 return atomic_inc_not_zero(&mm->mm_users);
2940}
2941
Linus Torvalds1da177e2005-04-16 15:20:36 -07002942/* mmput gets rid of the mappings and all user-space */
2943extern void mmput(struct mm_struct *);
Michal Hocko7ef949d72016-05-26 15:16:22 -07002944#ifdef CONFIG_MMU
2945/* same as above but performs the slow path from the async context. Can
Michal Hockoec8d7c12016-05-20 16:57:21 -07002946 * be called from the atomic context as well
2947 */
2948extern void mmput_async(struct mm_struct *);
Michal Hocko7ef949d72016-05-26 15:16:22 -07002949#endif
Michal Hockoec8d7c12016-05-20 16:57:21 -07002950
Linus Torvalds1da177e2005-04-16 15:20:36 -07002951/* Grab a reference to a task's mm, if it is not already going away */
2952extern struct mm_struct *get_task_mm(struct task_struct *task);
Christopher Yeoh8cdb8782012-02-02 11:34:09 +10302953/*
2954 * Grab a reference to a task's mm, if it is not already going away
2955 * and ptrace_may_access with the mode parameter passed to it
2956 * succeeds.
2957 */
2958extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
Thomas Gleixner394ff122021-02-01 10:01:35 +00002959/* Remove the current tasks stale references to the old mm_struct on exit() */
2960extern void exit_mm_release(struct task_struct *, struct mm_struct *);
2961/* Remove the current tasks stale references to the old mm_struct on exec() */
2962extern void exec_mm_release(struct task_struct *, struct mm_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002963
Josh Triplett3033f14a2015-06-25 15:01:19 -07002964#ifdef CONFIG_HAVE_COPY_THREAD_TLS
2965extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2966 struct task_struct *, unsigned long);
2967#else
Alexey Dobriyan6f2c55b2009-04-02 16:56:59 -07002968extern int copy_thread(unsigned long, unsigned long, unsigned long,
Al Viroafa86fc2012-10-22 22:51:14 -04002969 struct task_struct *);
Josh Triplett3033f14a2015-06-25 15:01:19 -07002970
2971/* Architectures that haven't opted into copy_thread_tls get the tls argument
2972 * via pt_regs, so ignore the tls argument passed via C. */
2973static inline int copy_thread_tls(
2974 unsigned long clone_flags, unsigned long sp, unsigned long arg,
2975 struct task_struct *p, unsigned long tls)
2976{
2977 return copy_thread(clone_flags, sp, arg, p);
2978}
2979#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002980extern void flush_thread(void);
Jiri Slaby5f56a5d2016-05-20 17:00:16 -07002981
2982#ifdef CONFIG_HAVE_EXIT_THREAD
Jiri Slabye6464692016-05-20 17:00:20 -07002983extern void exit_thread(struct task_struct *tsk);
Jiri Slaby5f56a5d2016-05-20 17:00:16 -07002984#else
Jiri Slabye6464692016-05-20 17:00:20 -07002985static inline void exit_thread(struct task_struct *tsk)
Jiri Slaby5f56a5d2016-05-20 17:00:16 -07002986{
2987}
2988#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002989
Linus Torvalds1da177e2005-04-16 15:20:36 -07002990extern void exit_files(struct task_struct *);
Oleg Nesterova7e53282006-03-28 16:11:27 -08002991extern void __cleanup_sighand(struct sighand_struct *);
Oleg Nesterovcbaffba2008-05-26 20:55:42 +04002992
Linus Torvalds1da177e2005-04-16 15:20:36 -07002993extern void exit_itimers(struct signal_struct *);
Oleg Nesterovcbaffba2008-05-26 20:55:42 +04002994extern void flush_itimer_signals(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002995
Joe Perches9402c952012-01-12 17:17:17 -08002996extern void do_group_exit(int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002997
Linus Torvaldsc4ad8f92014-02-05 12:54:53 -08002998extern int do_execve(struct filename *,
David Howellsd7627462010-08-17 23:52:56 +01002999 const char __user * const __user *,
Al Viroda3d4c52012-10-20 21:49:33 -04003000 const char __user * const __user *);
David Drysdale51f39a12014-12-12 16:57:29 -08003001extern int do_execveat(int, struct filename *,
3002 const char __user * const __user *,
3003 const char __user * const __user *,
3004 int);
Josh Triplett3033f14a2015-06-25 15:01:19 -07003005extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
Al Viroe80d6662012-10-22 23:10:08 -04003006extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
Ingo Molnar36c8b582006-07-03 00:25:41 -07003007struct task_struct *fork_idle(int);
Al Viro2aa3a7f2012-09-21 19:55:31 -04003008extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003009
Adrian Hunter82b89772014-05-28 11:45:04 +03003010extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
3011static inline void set_task_comm(struct task_struct *tsk, const char *from)
3012{
3013 __set_task_comm(tsk, from, false);
3014}
Arnd Bergmannb9dd1342017-12-14 15:32:41 -08003015extern char *__get_task_comm(char *to, size_t len, struct task_struct *tsk);
3016#define get_task_comm(buf, tsk) ({ \
3017 BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN); \
3018 __get_task_comm(buf, sizeof(buf), tsk); \
3019})
Linus Torvalds1da177e2005-04-16 15:20:36 -07003020
3021#ifdef CONFIG_SMP
Peter Zijlstra317f3942011-04-05 17:23:58 +02003022void scheduler_ipi(void);
Roland McGrath85ba2d82008-07-25 19:45:58 -07003023extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003024#else
Peter Zijlstra184748c2011-04-05 17:23:39 +02003025static inline void scheduler_ipi(void) { }
Roland McGrath85ba2d82008-07-25 19:45:58 -07003026static inline unsigned long wait_task_inactive(struct task_struct *p,
3027 long match_state)
3028{
3029 return 1;
3030}
Linus Torvalds1da177e2005-04-16 15:20:36 -07003031#endif
3032
Frederic Weisbeckerfafe8702015-05-06 18:04:24 +02003033#define tasklist_empty() \
3034 list_empty(&init_task.tasks)
3035
Jiri Pirko05725f72009-04-14 20:17:16 +02003036#define next_task(p) \
3037 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003038
3039#define for_each_process(p) \
3040 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
3041
Oleg Nesterov5bb459b2009-07-10 03:48:23 +02003042extern bool current_is_single_threaded(void);
David Howellsd84f4f92008-11-14 10:39:23 +11003043
Linus Torvalds1da177e2005-04-16 15:20:36 -07003044/*
3045 * Careful: do_each_thread/while_each_thread is a double loop so
3046 * 'break' will not work as expected - use goto instead.
3047 */
3048#define do_each_thread(g, t) \
3049 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
3050
3051#define while_each_thread(g, t) \
3052 while ((t = next_thread(t)) != g)
3053
Oleg Nesterov0c740d02014-01-21 15:49:56 -08003054#define __for_each_thread(signal, t) \
3055 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
3056
3057#define for_each_thread(p, t) \
3058 __for_each_thread((p)->signal, t)
3059
3060/* Careful: this is a double loop, 'break' won't work as expected. */
3061#define for_each_process_thread(p, t) \
3062 for_each_process(p) for_each_thread(p, t)
3063
Oleg Nesterov7e498272010-05-26 14:43:22 -07003064static inline int get_nr_threads(struct task_struct *tsk)
3065{
Oleg Nesterovb3ac0222010-05-26 14:43:24 -07003066 return tsk->signal->nr_threads;
Oleg Nesterov7e498272010-05-26 14:43:22 -07003067}
3068
Oleg Nesterov087806b2011-06-22 23:10:26 +02003069static inline bool thread_group_leader(struct task_struct *p)
3070{
3071 return p->exit_signal >= 0;
3072}
Linus Torvalds1da177e2005-04-16 15:20:36 -07003073
Eric W. Biederman0804ef42006-10-02 02:17:04 -07003074/* Do to the insanities of de_thread it is possible for a process
3075 * to have the pid of the thread group leader without actually being
3076 * the thread group leader. For iteration through the pids in proc
3077 * all we care about is that we have a task with the appropriate
3078 * pid, we don't actually care if we have the right task.
3079 */
Oleg Nesterove1403b82013-09-11 14:20:06 -07003080static inline bool has_group_leader_pid(struct task_struct *p)
Eric W. Biederman0804ef42006-10-02 02:17:04 -07003081{
Oleg Nesterove1403b82013-09-11 14:20:06 -07003082 return task_pid(p) == p->signal->leader_pid;
Eric W. Biederman0804ef42006-10-02 02:17:04 -07003083}
3084
Pavel Emelyanovbac0abd2007-10-18 23:40:18 -07003085static inline
Oleg Nesterove1403b82013-09-11 14:20:06 -07003086bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
Pavel Emelyanovbac0abd2007-10-18 23:40:18 -07003087{
Oleg Nesterove1403b82013-09-11 14:20:06 -07003088 return p1->signal == p2->signal;
Pavel Emelyanovbac0abd2007-10-18 23:40:18 -07003089}
3090
Ingo Molnar36c8b582006-07-03 00:25:41 -07003091static inline struct task_struct *next_thread(const struct task_struct *p)
Oleg Nesterov47e65322006-03-28 16:11:25 -08003092{
Jiri Pirko05725f72009-04-14 20:17:16 +02003093 return list_entry_rcu(p->thread_group.next,
3094 struct task_struct, thread_group);
Oleg Nesterov47e65322006-03-28 16:11:25 -08003095}
3096
Alexey Dobriyane8681712007-10-26 12:17:22 +04003097static inline int thread_group_empty(struct task_struct *p)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003098{
Oleg Nesterov47e65322006-03-28 16:11:25 -08003099 return list_empty(&p->thread_group);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003100}
3101
3102#define delay_group_leader(p) \
3103 (thread_group_leader(p) && !thread_group_empty(p))
3104
Linus Torvalds1da177e2005-04-16 15:20:36 -07003105/*
Eric W. Biederman260ea102006-06-23 02:05:18 -07003106 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
Jens Axboe22e2c502005-06-27 10:55:12 +02003107 * subscriptions and synchronises with wait4(). Also used in procfs. Also
Paul Menageddbcc7e2007-10-18 23:39:30 -07003108 * pins the final release of task.io_context. Also protects ->cpuset and
Alexander Mikhalitsyncc927c12021-11-19 16:43:21 -08003109 * ->cgroup.subsys[]. And ->vfork_done. And ->sysvshm.shm_clist.
Linus Torvalds1da177e2005-04-16 15:20:36 -07003110 *
3111 * Nests both inside and outside of read_lock(&tasklist_lock).
3112 * It must not be nested with write_lock_irq(&tasklist_lock),
3113 * neither inside nor outside.
3114 */
3115static inline void task_lock(struct task_struct *p)
3116{
3117 spin_lock(&p->alloc_lock);
3118}
3119
3120static inline void task_unlock(struct task_struct *p)
3121{
3122 spin_unlock(&p->alloc_lock);
3123}
3124
Namhyung Kimb8ed3742010-10-27 15:34:06 -07003125extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
Oleg Nesterovf63ee722006-03-28 16:11:13 -08003126 unsigned long *flags);
3127
Anton Vorontsov9388dc32012-02-09 20:45:19 +04003128static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
3129 unsigned long *flags)
3130{
3131 struct sighand_struct *ret;
3132
3133 ret = __lock_task_sighand(tsk, flags);
3134 (void)__cond_lock(&tsk->sighand->siglock, ret);
3135 return ret;
3136}
Namhyung Kimb8ed3742010-10-27 15:34:06 -07003137
Oleg Nesterovf63ee722006-03-28 16:11:13 -08003138static inline void unlock_task_sighand(struct task_struct *tsk,
3139 unsigned long *flags)
3140{
3141 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
3142}
3143
Tejun Heo7d7efec2015-05-13 16:35:16 -04003144/**
3145 * threadgroup_change_begin - mark the beginning of changes to a threadgroup
3146 * @tsk: task causing the changes
3147 *
3148 * All operations which modify a threadgroup - a new thread joining the
3149 * group, death of a member thread (the assertion of PF_EXITING) and
3150 * exec(2) dethreading the process and replacing the leader - are wrapped
3151 * by threadgroup_change_{begin|end}(). This is to provide a place which
3152 * subsystems needing threadgroup stability can hook into for
3153 * synchronization.
3154 */
Tejun Heo257058a2011-12-12 18:12:21 -08003155static inline void threadgroup_change_begin(struct task_struct *tsk)
Ben Blum4714d1d2011-05-26 16:25:18 -07003156{
Tejun Heo7d7efec2015-05-13 16:35:16 -04003157 might_sleep();
3158 cgroup_threadgroup_change_begin(tsk);
Ben Blum4714d1d2011-05-26 16:25:18 -07003159}
Tejun Heo7d7efec2015-05-13 16:35:16 -04003160
3161/**
3162 * threadgroup_change_end - mark the end of changes to a threadgroup
3163 * @tsk: task causing the changes
3164 *
3165 * See threadgroup_change_begin().
3166 */
Tejun Heo257058a2011-12-12 18:12:21 -08003167static inline void threadgroup_change_end(struct task_struct *tsk)
Ben Blum4714d1d2011-05-26 16:25:18 -07003168{
Tejun Heo7d7efec2015-05-13 16:35:16 -04003169 cgroup_threadgroup_change_end(tsk);
Ben Blum4714d1d2011-05-26 16:25:18 -07003170}
Tejun Heo77e4ef92011-12-12 18:12:21 -08003171
Andy Lutomirskic65eacb2016-09-13 14:29:24 -07003172#ifdef CONFIG_THREAD_INFO_IN_TASK
3173
3174static inline struct thread_info *task_thread_info(struct task_struct *task)
3175{
3176 return &task->thread_info;
3177}
Andy Lutomirskic6c314a2016-09-15 22:45:43 -07003178
3179/*
3180 * When accessing the stack of a non-current task that might exit, use
3181 * try_get_task_stack() instead. task_stack_page will return a pointer
3182 * that could get freed out from under you.
3183 */
Andy Lutomirskic65eacb2016-09-13 14:29:24 -07003184static inline void *task_stack_page(const struct task_struct *task)
3185{
3186 return task->stack;
3187}
Andy Lutomirskic6c314a2016-09-15 22:45:43 -07003188
Andy Lutomirskic65eacb2016-09-13 14:29:24 -07003189#define setup_thread_stack(new,old) do { } while(0)
Andy Lutomirskic6c314a2016-09-15 22:45:43 -07003190
Andy Lutomirskic65eacb2016-09-13 14:29:24 -07003191static inline unsigned long *end_of_stack(const struct task_struct *task)
3192{
3193 return task->stack;
3194}
3195
3196#elif !defined(__HAVE_THREAD_FUNCTIONS)
Al Virof0373602005-11-13 16:06:57 -08003197
Roman Zippelf7e42172007-05-09 02:35:17 -07003198#define task_thread_info(task) ((struct thread_info *)(task)->stack)
Andy Lutomirskic65eacb2016-09-13 14:29:24 -07003199#define task_stack_page(task) ((void *)(task)->stack)
Al Viroa1261f52005-11-13 16:06:55 -08003200
Al Viro10ebffd2005-11-13 16:06:56 -08003201static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
3202{
3203 *task_thread_info(p) = *task_thread_info(org);
3204 task_thread_info(p)->task = p;
3205}
3206
Chuck Ebbert6a402812014-09-20 10:17:51 -05003207/*
3208 * Return the address of the last usable long on the stack.
3209 *
3210 * When the stack grows down, this is just above the thread
3211 * info struct. Going any lower will corrupt the threadinfo.
3212 *
3213 * When the stack grows up, this is the highest address.
3214 * Beyond that position, we corrupt data on the next page.
3215 */
Al Viro10ebffd2005-11-13 16:06:56 -08003216static inline unsigned long *end_of_stack(struct task_struct *p)
3217{
Chuck Ebbert6a402812014-09-20 10:17:51 -05003218#ifdef CONFIG_STACK_GROWSUP
3219 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
3220#else
Roman Zippelf7e42172007-05-09 02:35:17 -07003221 return (unsigned long *)(task_thread_info(p) + 1);
Chuck Ebbert6a402812014-09-20 10:17:51 -05003222#endif
Al Viro10ebffd2005-11-13 16:06:56 -08003223}
3224
Al Virof0373602005-11-13 16:06:57 -08003225#endif
Andy Lutomirskic6c314a2016-09-15 22:45:43 -07003226
Andy Lutomirski68f24b082016-09-15 22:45:48 -07003227#ifdef CONFIG_THREAD_INFO_IN_TASK
3228static inline void *try_get_task_stack(struct task_struct *tsk)
3229{
3230 return atomic_inc_not_zero(&tsk->stack_refcount) ?
3231 task_stack_page(tsk) : NULL;
3232}
3233
3234extern void put_task_stack(struct task_struct *tsk);
3235#else
Andy Lutomirskic6c314a2016-09-15 22:45:43 -07003236static inline void *try_get_task_stack(struct task_struct *tsk)
3237{
3238 return task_stack_page(tsk);
3239}
3240
3241static inline void put_task_stack(struct task_struct *tsk) {}
Andy Lutomirski68f24b082016-09-15 22:45:48 -07003242#endif
Andy Lutomirskic6c314a2016-09-15 22:45:43 -07003243
Aaron Tomlina70857e2014-09-12 14:16:18 +01003244#define task_stack_end_corrupted(task) \
3245 (*(end_of_stack(task)) != STACK_END_MAGIC)
Al Virof0373602005-11-13 16:06:57 -08003246
FUJITA Tomonori8b05c7e2008-07-23 21:26:53 -07003247static inline int object_is_on_stack(void *obj)
3248{
3249 void *stack = task_stack_page(current);
3250
3251 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
3252}
3253
Linus Torvaldsb235bee2016-06-24 15:09:37 -07003254extern void thread_stack_cache_init(void);
Benjamin Herrenschmidt8c9843e2008-04-18 16:56:15 +10003255
Eric Sandeen7c9f8862008-04-22 16:38:23 -05003256#ifdef CONFIG_DEBUG_STACK_USAGE
3257static inline unsigned long stack_not_used(struct task_struct *p)
3258{
3259 unsigned long *n = end_of_stack(p);
3260
3261 do { /* Skip over canary */
Helge Deller6c31da32016-03-19 17:54:10 +01003262# ifdef CONFIG_STACK_GROWSUP
3263 n--;
3264# else
Eric Sandeen7c9f8862008-04-22 16:38:23 -05003265 n++;
Helge Deller6c31da32016-03-19 17:54:10 +01003266# endif
Eric Sandeen7c9f8862008-04-22 16:38:23 -05003267 } while (!*n);
3268
Helge Deller6c31da32016-03-19 17:54:10 +01003269# ifdef CONFIG_STACK_GROWSUP
3270 return (unsigned long)end_of_stack(p) - (unsigned long)n;
3271# else
Eric Sandeen7c9f8862008-04-22 16:38:23 -05003272 return (unsigned long)n - (unsigned long)end_of_stack(p);
Helge Deller6c31da32016-03-19 17:54:10 +01003273# endif
Eric Sandeen7c9f8862008-04-22 16:38:23 -05003274}
3275#endif
Aaron Tomlind4311ff2014-09-12 14:16:17 +01003276extern void set_task_stack_end_magic(struct task_struct *tsk);
Eric Sandeen7c9f8862008-04-22 16:38:23 -05003277
Linus Torvalds1da177e2005-04-16 15:20:36 -07003278/* set thread flags in other task's structures
3279 * - see asm/thread_info.h for TIF_xxxx flags available
3280 */
3281static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
3282{
Al Viroa1261f52005-11-13 16:06:55 -08003283 set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003284}
3285
3286static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
3287{
Al Viroa1261f52005-11-13 16:06:55 -08003288 clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003289}
3290
3291static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
3292{
Al Viroa1261f52005-11-13 16:06:55 -08003293 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003294}
3295
3296static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
3297{
Al Viroa1261f52005-11-13 16:06:55 -08003298 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003299}
3300
3301static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
3302{
Al Viroa1261f52005-11-13 16:06:55 -08003303 return test_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003304}
3305
3306static inline void set_tsk_need_resched(struct task_struct *tsk)
3307{
3308 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
3309}
3310
3311static inline void clear_tsk_need_resched(struct task_struct *tsk)
3312{
3313 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
3314}
3315
Gregory Haskins8ae121a2008-04-23 07:13:29 -04003316static inline int test_tsk_need_resched(struct task_struct *tsk)
3317{
3318 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
3319}
3320
Eric W. Biederman690cc3f2009-05-13 16:55:10 +00003321static inline int restart_syscall(void)
3322{
3323 set_tsk_thread_flag(current, TIF_SIGPENDING);
3324 return -ERESTARTNOINTR;
3325}
3326
Linus Torvalds1da177e2005-04-16 15:20:36 -07003327static inline int signal_pending(struct task_struct *p)
3328{
3329 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
3330}
Matthew Wilcoxf776d122007-12-06 11:15:50 -05003331
Roland McGrathd9588722009-09-23 15:57:04 -07003332static inline int __fatal_signal_pending(struct task_struct *p)
3333{
3334 return unlikely(sigismember(&p->pending.signal, SIGKILL));
3335}
Matthew Wilcoxf776d122007-12-06 11:15:50 -05003336
3337static inline int fatal_signal_pending(struct task_struct *p)
3338{
3339 return signal_pending(p) && __fatal_signal_pending(p);
3340}
3341
Oleg Nesterov16882c12008-06-08 21:20:41 +04003342static inline int signal_pending_state(long state, struct task_struct *p)
3343{
3344 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
3345 return 0;
3346 if (!signal_pending(p))
3347 return 0;
3348
Oleg Nesterov16882c12008-06-08 21:20:41 +04003349 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
3350}
3351
Linus Torvalds1da177e2005-04-16 15:20:36 -07003352/*
3353 * cond_resched() and cond_resched_lock(): latency reduction via
3354 * explicit rescheduling in places that are safe. The return
3355 * value indicates whether a reschedule was done in fact.
3356 * cond_resched_lock() will drop the spinlock before scheduling,
3357 * cond_resched_softirq() will enable bhs before scheduling.
3358 */
Peter Zijlstra35a773a2016-09-19 12:57:53 +02003359#ifndef CONFIG_PREEMPT
Linus Torvaldsc3921ab2008-05-11 16:04:48 -07003360extern int _cond_resched(void);
Peter Zijlstra35a773a2016-09-19 12:57:53 +02003361#else
3362static inline int _cond_resched(void) { return 0; }
3363#endif
Frederic Weisbecker6f80bd92009-07-16 15:44:29 +02003364
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02003365#define cond_resched() ({ \
Peter Zijlstra34274452014-09-24 10:18:56 +02003366 ___might_sleep(__FILE__, __LINE__, 0); \
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02003367 _cond_resched(); \
3368})
Frederic Weisbecker6f80bd92009-07-16 15:44:29 +02003369
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02003370extern int __cond_resched_lock(spinlock_t *lock);
3371
3372#define cond_resched_lock(lock) ({ \
Peter Zijlstra34274452014-09-24 10:18:56 +02003373 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02003374 __cond_resched_lock(lock); \
3375})
3376
3377extern int __cond_resched_softirq(void);
3378
Venkatesh Pallipadi75e10562010-10-04 17:03:16 -07003379#define cond_resched_softirq() ({ \
Peter Zijlstra34274452014-09-24 10:18:56 +02003380 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
Venkatesh Pallipadi75e10562010-10-04 17:03:16 -07003381 __cond_resched_softirq(); \
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02003382})
Linus Torvalds1da177e2005-04-16 15:20:36 -07003383
Simon Hormanf6f3c432013-05-22 14:50:31 +09003384static inline void cond_resched_rcu(void)
3385{
3386#if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
3387 rcu_read_unlock();
3388 cond_resched();
3389 rcu_read_lock();
3390#endif
3391}
3392
Vegard Nossumd1c6d142016-07-23 09:46:39 +02003393static inline unsigned long get_preempt_disable_ip(struct task_struct *p)
3394{
3395#ifdef CONFIG_DEBUG_PREEMPT
3396 return p->preempt_disable_ip;
3397#else
3398 return 0;
3399#endif
3400}
3401
Linus Torvalds1da177e2005-04-16 15:20:36 -07003402/*
3403 * Does a critical section need to be broken due to another
Nick Piggin95c354f2008-01-30 13:31:20 +01003404 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
3405 * but a general need for low latency)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003406 */
Nick Piggin95c354f2008-01-30 13:31:20 +01003407static inline int spin_needbreak(spinlock_t *lock)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003408{
Nick Piggin95c354f2008-01-30 13:31:20 +01003409#ifdef CONFIG_PREEMPT
3410 return spin_is_contended(lock);
3411#else
Linus Torvalds1da177e2005-04-16 15:20:36 -07003412 return 0;
Nick Piggin95c354f2008-01-30 13:31:20 +01003413#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07003414}
3415
Roland McGrath7bb44ad2007-05-23 13:57:44 -07003416/*
Thomas Gleixneree761f62013-03-21 22:49:32 +01003417 * Idle thread specific functions to determine the need_resched
Peter Zijlstra69dd0f82014-04-09 14:30:10 +02003418 * polling state.
Thomas Gleixneree761f62013-03-21 22:49:32 +01003419 */
Peter Zijlstra69dd0f82014-04-09 14:30:10 +02003420#ifdef TIF_POLLING_NRFLAG
Thomas Gleixneree761f62013-03-21 22:49:32 +01003421static inline int tsk_is_polling(struct task_struct *p)
3422{
3423 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
3424}
Peter Zijlstraea811742013-09-11 12:43:13 +02003425
3426static inline void __current_set_polling(void)
Thomas Gleixner3a98f872013-03-21 22:49:33 +01003427{
3428 set_thread_flag(TIF_POLLING_NRFLAG);
3429}
3430
Peter Zijlstraea811742013-09-11 12:43:13 +02003431static inline bool __must_check current_set_polling_and_test(void)
3432{
3433 __current_set_polling();
3434
3435 /*
3436 * Polling state must be visible before we test NEED_RESCHED,
Kirill Tkhai88751252014-06-29 00:03:57 +04003437 * paired by resched_curr()
Peter Zijlstraea811742013-09-11 12:43:13 +02003438 */
Peter Zijlstra4e857c52014-03-17 18:06:10 +01003439 smp_mb__after_atomic();
Peter Zijlstraea811742013-09-11 12:43:13 +02003440
3441 return unlikely(tif_need_resched());
3442}
3443
3444static inline void __current_clr_polling(void)
Thomas Gleixner3a98f872013-03-21 22:49:33 +01003445{
3446 clear_thread_flag(TIF_POLLING_NRFLAG);
3447}
Peter Zijlstraea811742013-09-11 12:43:13 +02003448
3449static inline bool __must_check current_clr_polling_and_test(void)
3450{
3451 __current_clr_polling();
3452
3453 /*
3454 * Polling state must be visible before we test NEED_RESCHED,
Kirill Tkhai88751252014-06-29 00:03:57 +04003455 * paired by resched_curr()
Peter Zijlstraea811742013-09-11 12:43:13 +02003456 */
Peter Zijlstra4e857c52014-03-17 18:06:10 +01003457 smp_mb__after_atomic();
Peter Zijlstraea811742013-09-11 12:43:13 +02003458
3459 return unlikely(tif_need_resched());
3460}
3461
Thomas Gleixneree761f62013-03-21 22:49:32 +01003462#else
3463static inline int tsk_is_polling(struct task_struct *p) { return 0; }
Peter Zijlstraea811742013-09-11 12:43:13 +02003464static inline void __current_set_polling(void) { }
3465static inline void __current_clr_polling(void) { }
3466
3467static inline bool __must_check current_set_polling_and_test(void)
3468{
3469 return unlikely(tif_need_resched());
3470}
3471static inline bool __must_check current_clr_polling_and_test(void)
3472{
3473 return unlikely(tif_need_resched());
3474}
Thomas Gleixneree761f62013-03-21 22:49:32 +01003475#endif
3476
Peter Zijlstra8cb75e02013-11-20 12:22:37 +01003477static inline void current_clr_polling(void)
3478{
3479 __current_clr_polling();
3480
3481 /*
3482 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3483 * Once the bit is cleared, we'll get IPIs with every new
3484 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3485 * fold.
3486 */
Kirill Tkhai88751252014-06-29 00:03:57 +04003487 smp_mb(); /* paired with resched_curr() */
Peter Zijlstra8cb75e02013-11-20 12:22:37 +01003488
3489 preempt_fold_need_resched();
3490}
3491
Peter Zijlstra75f93fe2013-09-27 17:30:03 +02003492static __always_inline bool need_resched(void)
3493{
3494 return unlikely(tif_need_resched());
3495}
3496
Thomas Gleixneree761f62013-03-21 22:49:32 +01003497/*
Frank Mayharf06febc2008-09-12 09:54:39 -07003498 * Thread group CPU time accounting.
3499 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +01003500void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
Peter Zijlstra4da94d492009-02-11 11:30:27 +01003501void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
Frank Mayharf06febc2008-09-12 09:54:39 -07003502
Frank Mayharf06febc2008-09-12 09:54:39 -07003503/*
Roland McGrath7bb44ad2007-05-23 13:57:44 -07003504 * Reevaluate whether the task has signals pending delivery.
3505 * Wake the task if so.
3506 * This is required every time the blocked sigset_t changes.
3507 * callers must hold sighand->siglock.
3508 */
3509extern void recalc_sigpending_and_wake(struct task_struct *t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003510extern void recalc_sigpending(void);
3511
Oleg Nesterov910ffdb2013-01-21 20:47:41 +01003512extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3513
3514static inline void signal_wake_up(struct task_struct *t, bool resume)
3515{
3516 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3517}
3518static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3519{
3520 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3521}
Linus Torvalds1da177e2005-04-16 15:20:36 -07003522
3523/*
3524 * Wrappers for p->thread_info->cpu access. No-op on UP.
3525 */
3526#ifdef CONFIG_SMP
3527
3528static inline unsigned int task_cpu(const struct task_struct *p)
3529{
Andy Lutomirskic65eacb2016-09-13 14:29:24 -07003530#ifdef CONFIG_THREAD_INFO_IN_TASK
3531 return p->cpu;
3532#else
Al Viroa1261f52005-11-13 16:06:55 -08003533 return task_thread_info(p)->cpu;
Andy Lutomirskic65eacb2016-09-13 14:29:24 -07003534#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07003535}
3536
Ingo Molnarb32e86b2013-10-07 11:29:30 +01003537static inline int task_node(const struct task_struct *p)
3538{
3539 return cpu_to_node(task_cpu(p));
3540}
3541
Ingo Molnarc65cc872007-07-09 18:51:58 +02003542extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003543
3544#else
3545
3546static inline unsigned int task_cpu(const struct task_struct *p)
3547{
3548 return 0;
3549}
3550
3551static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3552{
3553}
3554
3555#endif /* CONFIG_SMP */
3556
Rusty Russell96f874e2008-11-25 02:35:14 +10303557extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3558extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
Siddha, Suresh B5c45bf22006-06-27 02:54:42 -07003559
Dhaval Giani7c941432010-01-20 13:26:18 +01003560#ifdef CONFIG_CGROUP_SCHED
Yong Zhang07e06b02011-01-07 15:17:36 +08003561extern struct task_group root_task_group;
Peter Zijlstra8323f262012-06-22 13:36:05 +02003562#endif /* CONFIG_CGROUP_SCHED */
Srivatsa Vaddagiri9b5b7752007-10-15 17:00:09 +02003563
Dhaval Giani54e99122009-02-27 15:13:54 +05303564extern int task_can_switch_user(struct user_struct *up,
3565 struct task_struct *tsk);
3566
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08003567#ifdef CONFIG_TASK_XACCT
3568static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3569{
Andrea Righi940389b2008-07-28 00:48:12 +02003570 tsk->ioac.rchar += amt;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08003571}
3572
3573static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3574{
Andrea Righi940389b2008-07-28 00:48:12 +02003575 tsk->ioac.wchar += amt;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08003576}
3577
3578static inline void inc_syscr(struct task_struct *tsk)
3579{
Andrea Righi940389b2008-07-28 00:48:12 +02003580 tsk->ioac.syscr++;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08003581}
3582
3583static inline void inc_syscw(struct task_struct *tsk)
3584{
Andrea Righi940389b2008-07-28 00:48:12 +02003585 tsk->ioac.syscw++;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08003586}
3587#else
3588static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3589{
3590}
3591
3592static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3593{
3594}
3595
3596static inline void inc_syscr(struct task_struct *tsk)
3597{
3598}
3599
3600static inline void inc_syscw(struct task_struct *tsk)
3601{
3602}
3603#endif
3604
Dave Hansen82455252008-02-04 22:28:59 -08003605#ifndef TASK_SIZE_OF
3606#define TASK_SIZE_OF(tsk) TASK_SIZE
3607#endif
3608
Oleg Nesterovf98bafa2014-06-04 16:07:34 -07003609#ifdef CONFIG_MEMCG
Balbir Singhcf475ad2008-04-29 01:00:16 -07003610extern void mm_update_next_owner(struct mm_struct *mm);
Balbir Singhcf475ad2008-04-29 01:00:16 -07003611#else
3612static inline void mm_update_next_owner(struct mm_struct *mm)
3613{
3614}
Oleg Nesterovf98bafa2014-06-04 16:07:34 -07003615#endif /* CONFIG_MEMCG */
Balbir Singhcf475ad2008-04-29 01:00:16 -07003616
Jiri Slaby3e10e712009-11-19 17:16:37 +01003617static inline unsigned long task_rlimit(const struct task_struct *tsk,
3618 unsigned int limit)
3619{
Jason Low316c1608d2015-04-28 13:00:20 -07003620 return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
Jiri Slaby3e10e712009-11-19 17:16:37 +01003621}
3622
3623static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3624 unsigned int limit)
3625{
Jason Low316c1608d2015-04-28 13:00:20 -07003626 return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
Jiri Slaby3e10e712009-11-19 17:16:37 +01003627}
3628
3629static inline unsigned long rlimit(unsigned int limit)
3630{
3631 return task_rlimit(current, limit);
3632}
3633
3634static inline unsigned long rlimit_max(unsigned int limit)
3635{
3636 return task_rlimit_max(current, limit);
3637}
3638
Rafael J. Wysocki58919e82016-08-16 22:14:55 +02003639#define SCHED_CPUFREQ_RT (1U << 0)
3640#define SCHED_CPUFREQ_DL (1U << 1)
Rafael J. Wysocki8c34ab12016-09-09 23:59:33 +02003641#define SCHED_CPUFREQ_IOWAIT (1U << 2)
Rafael J. Wysocki58919e82016-08-16 22:14:55 +02003642
3643#define SCHED_CPUFREQ_RT_DL (SCHED_CPUFREQ_RT | SCHED_CPUFREQ_DL)
3644
Rafael J. Wysockiadaf9fc2016-03-10 20:44:47 +01003645#ifdef CONFIG_CPU_FREQ
3646struct update_util_data {
Rafael J. Wysocki58919e82016-08-16 22:14:55 +02003647 void (*func)(struct update_util_data *data, u64 time, unsigned int flags);
Rafael J. Wysockiadaf9fc2016-03-10 20:44:47 +01003648};
3649
Rafael J. Wysocki0bed6122016-04-02 01:08:43 +02003650void cpufreq_add_update_util_hook(int cpu, struct update_util_data *data,
Rafael J. Wysocki58919e82016-08-16 22:14:55 +02003651 void (*func)(struct update_util_data *data, u64 time,
3652 unsigned int flags));
Rafael J. Wysocki0bed6122016-04-02 01:08:43 +02003653void cpufreq_remove_update_util_hook(int cpu);
Rafael J. Wysockiadaf9fc2016-03-10 20:44:47 +01003654#endif /* CONFIG_CPU_FREQ */
3655
Linus Torvalds1da177e2005-04-16 15:20:36 -07003656#endif