blob: f8497059f88cd635f59c552c50f9ecedc8ba5cf6 [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>
Thomas Gleixner00d1a392013-09-17 18:53:09 +000028#include <linux/preempt_mask.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>
36#include <linux/signal.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070037#include <linux/compiler.h>
38#include <linux/completion.h>
39#include <linux/pid.h>
40#include <linux/percpu.h>
41#include <linux/topology.h>
Peter Zijlstra3e26c142007-10-16 23:25:50 -070042#include <linux/proportions.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070043#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>
Andrew Morton7c3ab7382006-12-10 02:19:19 -080053#include <linux/task_io_accounting.h>
Arjan van de Ven97455122008-01-25 21:08:34 +010054#include <linux/latencytop.h>
David Howells9e2b2dc2008-08-13 16:20:04 +010055#include <linux/cred.h>
Peter Zijlstrafa14ff42011-09-12 13:06:17 +020056#include <linux/llist.h>
Eric W. Biederman7b44ab92011-11-16 23:20:58 -080057#include <linux/uidgid.h>
Ming Lei21caf2f2013-02-22 16:34:08 -080058#include <linux/gfp.h>
David Woodhousea3b67142006-04-25 14:54:40 +010059
60#include <asm/processor.h>
H. J. Lu36d57ac2005-09-06 15:16:49 -070061
Dario Faggiolid50dde52013-11-07 14:43:36 +010062#define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
63
64/*
65 * Extended scheduling parameters data structure.
66 *
67 * This is needed because the original struct sched_param can not be
68 * altered without introducing ABI issues with legacy applications
69 * (e.g., in sched_getparam()).
70 *
71 * However, the possibility of specifying more than just a priority for
72 * the tasks may be useful for a wide variety of application fields, e.g.,
73 * multimedia, streaming, automation and control, and many others.
74 *
75 * This variant (sched_attr) is meant at describing a so-called
76 * sporadic time-constrained task. In such model a task is specified by:
77 * - the activation period or minimum instance inter-arrival time;
78 * - the maximum (or average, depending on the actual scheduling
79 * discipline) computation time of all instances, a.k.a. runtime;
80 * - the deadline (relative to the actual activation time) of each
81 * instance.
82 * Very briefly, a periodic (sporadic) task asks for the execution of
83 * some specific computation --which is typically called an instance--
84 * (at most) every period. Moreover, each instance typically lasts no more
85 * than the runtime and must be completed by time instant t equal to
86 * the instance activation time + the deadline.
87 *
88 * This is reflected by the actual fields of the sched_attr structure:
89 *
90 * @size size of the structure, for fwd/bwd compat.
91 *
92 * @sched_policy task's scheduling policy
93 * @sched_flags for customizing the scheduler behaviour
94 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
95 * @sched_priority task's static priority (SCHED_FIFO/RR)
96 * @sched_deadline representative of the task's deadline
97 * @sched_runtime representative of the task's runtime
98 * @sched_period representative of the task's period
99 *
100 * Given this task model, there are a multiplicity of scheduling algorithms
101 * and policies, that can be used to ensure all the tasks will make their
102 * timing constraints.
Dario Faggioliaab03e02013-11-28 11:14:43 +0100103 *
104 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
105 * only user of this new interface. More information about the algorithm
106 * available in the scheduling class file or in Documentation/.
Dario Faggiolid50dde52013-11-07 14:43:36 +0100107 */
108struct sched_attr {
109 u32 size;
110
111 u32 sched_policy;
112 u64 sched_flags;
113
114 /* SCHED_NORMAL, SCHED_BATCH */
115 s32 sched_nice;
116
117 /* SCHED_FIFO, SCHED_RR */
118 u32 sched_priority;
119
120 /* SCHED_DEADLINE */
121 u64 sched_runtime;
122 u64 sched_deadline;
123 u64 sched_period;
124};
125
Linus Torvalds1da177e2005-04-16 15:20:36 -0700126struct exec_domain;
Ingo Molnarc87e2832006-06-27 02:54:58 -0700127struct futex_pi_state;
Alexey Dobriyan286100a2008-01-25 21:08:34 +0100128struct robust_list_head;
Akinobu Mitabddd87c2010-02-23 08:55:42 +0100129struct bio_list;
Al Viro5ad4e532009-03-29 19:50:06 -0400130struct fs_struct;
Ingo Molnarcdd6c482009-09-21 12:02:48 +0200131struct perf_event_context;
Jens Axboe73c10102011-03-08 13:19:51 +0100132struct blk_plug;
Linus Torvaldsc4ad8f92014-02-05 12:54:53 -0800133struct filename;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700134
Davidlohr Bueso615d6e82014-04-07 15:37:25 -0700135#define VMACACHE_BITS 2
136#define VMACACHE_SIZE (1U << VMACACHE_BITS)
137#define VMACACHE_MASK (VMACACHE_SIZE - 1)
138
Linus Torvalds1da177e2005-04-16 15:20:36 -0700139/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700140 * List of flags we want to share for kernel threads,
141 * if only because they are not used by them anyway.
142 */
143#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
144
145/*
146 * These are the constant used to fake the fixed-point load-average
147 * counting. Some notes:
148 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
149 * a load-average precision of 10 bits integer + 11 bits fractional
150 * - if you want to count load-averages more often, you need more
151 * precision, or rounding will get you. With 2-second counting freq,
152 * the EXP_n values would be 1981, 2034 and 2043 if still using only
153 * 11 bit fractions.
154 */
155extern unsigned long avenrun[]; /* Load averages */
Thomas Gleixner2d024942009-05-02 20:08:52 +0200156extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700157
158#define FSHIFT 11 /* nr of bits of precision */
159#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
Linus Torvalds0c2043a2007-10-07 16:17:38 -0700160#define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700161#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
162#define EXP_5 2014 /* 1/exp(5sec/5min) */
163#define EXP_15 2037 /* 1/exp(5sec/15min) */
164
165#define CALC_LOAD(load,exp,n) \
166 load *= exp; \
167 load += n*(FIXED_1-exp); \
168 load >>= FSHIFT;
169
170extern unsigned long total_forks;
171extern int nr_threads;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700172DECLARE_PER_CPU(unsigned long, process_counts);
173extern int nr_processes(void);
174extern unsigned long nr_running(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700175extern unsigned long nr_iowait(void);
Peter Zijlstra8c215bd2010-07-01 09:07:17 +0200176extern unsigned long nr_iowait_cpu(int cpu);
Arjan van de Ven69d25872009-09-21 17:04:08 -0700177extern unsigned long this_cpu_load(void);
178
179
Peter Zijlstra0f004f52010-11-30 19:48:45 +0100180extern void calc_global_load(unsigned long ticks);
Peter Zijlstra5aaa0b72012-05-17 17:15:29 +0200181extern void update_cpu_load_nohz(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700182
Steven Rostedt7e49fcc2009-01-22 19:01:40 -0500183extern unsigned long get_parent_ip(unsigned long addr);
184
Paul E. McKenneyb637a322012-09-19 16:58:38 -0700185extern void dump_cpu_task(int cpu);
186
Ingo Molnar43ae34c2007-07-09 18:52:00 +0200187struct seq_file;
188struct cfs_rq;
Ingo Molnar4cf86d72007-10-15 17:00:14 +0200189struct task_group;
Ingo Molnar43ae34c2007-07-09 18:52:00 +0200190#ifdef CONFIG_SCHED_DEBUG
191extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
192extern void proc_sched_set_task(struct task_struct *p);
193extern void
Ingo Molnar5cef9ec2007-08-09 11:16:47 +0200194print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
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 */
213#define EXIT_ZOMBIE 16
214#define EXIT_DEAD 32
215/* in tsk->state again */
Mike Galbraithaf927232007-10-15 17:00:13 +0200216#define TASK_DEAD 64
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500217#define TASK_WAKEKILL 128
Peter Zijlstrae9c84312009-09-15 14:43:03 +0200218#define TASK_WAKING 256
Thomas Gleixnerf2530dc2013-04-09 09:33:34 +0200219#define TASK_PARKED 512
220#define TASK_STATE_MAX 1024
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500221
Thomas Gleixnerf2530dc2013-04-09 09:33:34 +0200222#define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
Peter Zijlstra73342152009-12-17 13:16:27 +0100223
Peter Zijlstrae1781532009-12-17 13:16:30 +0100224extern char ___assert_task_state[1 - 2*!!(
225 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500226
227/* Convenience macros for the sake of set_task_state */
228#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
229#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
230#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700231
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500232/* Convenience macros for the sake of wake_up */
233#define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500234#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500235
236/* get_task_state() */
237#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500238 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
Oleg Nesterov74e37202014-01-23 15:55:35 -0800239 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500240
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500241#define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
242#define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500243#define task_is_stopped_or_traced(task) \
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500244 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500245#define task_contributes_to_load(task) \
Nathan Lynche3c8ca82009-04-08 19:45:12 -0500246 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
Tejun Heo376fede2011-11-21 12:32:24 -0800247 (task->flags & PF_FROZEN) == 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700248
249#define __set_task_state(tsk, state_value) \
250 do { (tsk)->state = (state_value); } while (0)
251#define set_task_state(tsk, state_value) \
252 set_mb((tsk)->state, (state_value))
253
Andrew Morton498d0c52005-09-13 01:25:14 -0700254/*
255 * set_current_state() includes a barrier so that the write of current->state
256 * is correctly serialised wrt the caller's subsequent test of whether to
257 * actually sleep:
258 *
259 * set_current_state(TASK_UNINTERRUPTIBLE);
260 * if (do_i_need_to_sleep())
261 * schedule();
262 *
263 * If the caller does not need such serialisation then use __set_current_state()
264 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700265#define __set_current_state(state_value) \
266 do { current->state = (state_value); } while (0)
267#define set_current_state(state_value) \
268 set_mb(current->state, (state_value))
269
270/* Task command name length */
271#define TASK_COMM_LEN 16
272
Linus Torvalds1da177e2005-04-16 15:20:36 -0700273#include <linux/spinlock.h>
274
275/*
276 * This serializes "schedule()" and also protects
277 * the run-queue from deletions/modifications (but
278 * _adding_ to the beginning of the run-queue has
279 * a separate lock).
280 */
281extern rwlock_t tasklist_lock;
282extern spinlock_t mmlist_lock;
283
Ingo Molnar36c8b582006-07-03 00:25:41 -0700284struct task_struct;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700285
Paul E. McKenneydb1466b2010-03-03 07:46:56 -0800286#ifdef CONFIG_PROVE_RCU
287extern int lockdep_tasklist_lock_is_held(void);
288#endif /* #ifdef CONFIG_PROVE_RCU */
289
Linus Torvalds1da177e2005-04-16 15:20:36 -0700290extern void sched_init(void);
291extern void sched_init_smp(void);
Harvey Harrison2d07b252008-02-15 09:56:34 -0800292extern asmlinkage void schedule_tail(struct task_struct *prev);
Ingo Molnar36c8b582006-07-03 00:25:41 -0700293extern void init_idle(struct task_struct *idle, int cpu);
Ingo Molnar1df21052007-07-09 18:51:58 +0200294extern void init_idle_bootup_task(struct task_struct *idle);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700295
Andrew Morton89f19f02009-09-19 11:55:44 -0700296extern int runqueue_is_locked(int cpu);
Ingo Molnar017730c2008-05-12 21:20:52 +0200297
Frederic Weisbecker3451d022011-08-10 23:21:01 +0200298#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
Alex Shic1cc0172012-09-10 15:10:58 +0800299extern void nohz_balance_enter_idle(int cpu);
Suresh Siddha69e1e812011-12-01 17:07:33 -0800300extern void set_cpu_sd_state_idle(void);
Viresh Kumar6201b4d2014-03-18 16:26:07 +0530301extern int get_nohz_timer_target(int pinned);
Siddha, Suresh B46cb4b72007-05-08 00:32:51 -0700302#else
Alex Shic1cc0172012-09-10 15:10:58 +0800303static inline void nohz_balance_enter_idle(int cpu) { }
Peter Zijlstrafdaabd82011-12-06 12:47:55 +0100304static inline void set_cpu_sd_state_idle(void) { }
Viresh Kumar6201b4d2014-03-18 16:26:07 +0530305static inline int get_nohz_timer_target(int pinned)
306{
307 return smp_processor_id();
308}
Siddha, Suresh B46cb4b72007-05-08 00:32:51 -0700309#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700310
Ingo Molnare59e2ae2006-12-06 20:35:59 -0800311/*
Ingo Molnar39bc89f2007-04-25 20:50:03 -0700312 * Only dump TASK_* tasks. (0 for all tasks)
Ingo Molnare59e2ae2006-12-06 20:35:59 -0800313 */
314extern void show_state_filter(unsigned long state_filter);
315
316static inline void show_state(void)
317{
Ingo Molnar39bc89f2007-04-25 20:50:03 -0700318 show_state_filter(0);
Ingo Molnare59e2ae2006-12-06 20:35:59 -0800319}
320
Linus Torvalds1da177e2005-04-16 15:20:36 -0700321extern void show_regs(struct pt_regs *);
322
323/*
324 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
325 * task), SP is the stack pointer of the first frame that should be shown in the back
326 * trace (or NULL if the entire call-chain of the task should be shown).
327 */
328extern void show_stack(struct task_struct *task, unsigned long *sp);
329
330void io_schedule(void);
331long io_schedule_timeout(long timeout);
332
333extern void cpu_init (void);
334extern void trap_init(void);
335extern void update_process_times(int user);
336extern void scheduler_tick(void);
337
Ingo Molnar82a1fcb2008-01-25 21:08:02 +0100338extern void sched_show_task(struct task_struct *p);
339
Frederic Weisbecker19cc36c2010-05-13 02:30:49 +0200340#ifdef CONFIG_LOCKUP_DETECTOR
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700341extern void touch_softlockup_watchdog(void);
Jason Wesseld6ad3e22010-01-27 16:25:22 -0600342extern void touch_softlockup_watchdog_sync(void);
Jeremy Fitzhardinge04c91672007-05-08 00:28:05 -0700343extern void touch_all_softlockup_watchdogs(void);
Don Zickus332fbdb2010-05-07 17:11:45 -0400344extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
345 void __user *buffer,
346 size_t *lenp, loff_t *ppos);
Ingo Molnar9c44bc02008-05-12 21:21:04 +0200347extern unsigned int softlockup_panic;
Peter Zijlstra004417a2010-11-25 18:38:29 +0100348void lockup_detector_init(void);
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700349#else
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700350static inline void touch_softlockup_watchdog(void)
351{
352}
Jason Wesseld6ad3e22010-01-27 16:25:22 -0600353static inline void touch_softlockup_watchdog_sync(void)
354{
355}
Jeremy Fitzhardinge04c91672007-05-08 00:28:05 -0700356static inline void touch_all_softlockup_watchdogs(void)
357{
358}
Peter Zijlstra004417a2010-11-25 18:38:29 +0100359static inline void lockup_detector_init(void)
360{
361}
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700362#endif
363
Marcelo Tosatti8b414522013-10-11 21:39:26 -0300364#ifdef CONFIG_DETECT_HUNG_TASK
365void reset_hung_task_detector(void);
366#else
367static inline void reset_hung_task_detector(void)
368{
369}
370#endif
371
Linus Torvalds1da177e2005-04-16 15:20:36 -0700372/* Attach to any functions which should be ignored in wchan output. */
373#define __sched __attribute__((__section__(".sched.text")))
Ingo Molnardeaf2222007-11-28 15:52:56 +0100374
375/* Linker adds these: start and end of __sched functions */
376extern char __sched_text_start[], __sched_text_end[];
377
Linus Torvalds1da177e2005-04-16 15:20:36 -0700378/* Is this address in the __sched functions? */
379extern int in_sched_functions(unsigned long addr);
380
381#define MAX_SCHEDULE_TIMEOUT LONG_MAX
Harvey Harrisonb3c97522008-02-13 15:03:15 -0800382extern signed long schedule_timeout(signed long timeout);
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700383extern signed long schedule_timeout_interruptible(signed long timeout);
Matthew Wilcox294d5cc2007-12-06 11:59:46 -0500384extern signed long schedule_timeout_killable(signed long timeout);
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700385extern signed long schedule_timeout_uninterruptible(signed long timeout);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700386asmlinkage void schedule(void);
Thomas Gleixnerc5491ea2011-03-21 12:09:35 +0100387extern void schedule_preempt_disabled(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700388
Serge E. Hallynab516012006-10-02 02:18:06 -0700389struct nsproxy;
Cedric Le Goateracce2922007-07-15 23:40:59 -0700390struct user_namespace;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700391
David Howellsefc1a3b2010-01-15 17:01:35 -0800392#ifdef CONFIG_MMU
393extern void arch_pick_mmap_layout(struct mm_struct *mm);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700394extern unsigned long
395arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
396 unsigned long, unsigned long);
397extern unsigned long
398arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
399 unsigned long len, unsigned long pgoff,
400 unsigned long flags);
David Howellsefc1a3b2010-01-15 17:01:35 -0800401#else
402static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
403#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700404
Kees Cookd049f742013-11-12 15:11:17 -0800405#define SUID_DUMP_DISABLE 0 /* No setuid dumping */
406#define SUID_DUMP_USER 1 /* Dump as user of process */
407#define SUID_DUMP_ROOT 2 /* Dump as root */
408
Kawai, Hidehiro6c5d5232007-07-19 01:48:27 -0700409/* mm flags */
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700410
Oleg Nesterov7288e112014-01-23 15:55:32 -0800411/* for SUID_DUMP_* above */
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700412#define MMF_DUMPABLE_BITS 2
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700413#define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700414
Oleg Nesterov942be382014-01-23 15:55:34 -0800415extern void set_dumpable(struct mm_struct *mm, int value);
416/*
417 * This returns the actual value of the suid_dumpable flag. For things
418 * that are using this for checking for privilege transitions, it must
419 * test against SUID_DUMP_USER rather than treating it as a boolean
420 * value.
421 */
422static inline int __get_dumpable(unsigned long mm_flags)
423{
424 return mm_flags & MMF_DUMPABLE_MASK;
425}
426
427static inline int get_dumpable(struct mm_struct *mm)
428{
429 return __get_dumpable(mm->flags);
430}
431
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700432/* coredump filter bits */
433#define MMF_DUMP_ANON_PRIVATE 2
434#define MMF_DUMP_ANON_SHARED 3
435#define MMF_DUMP_MAPPED_PRIVATE 4
436#define MMF_DUMP_MAPPED_SHARED 5
Roland McGrath82df3972007-10-16 23:27:02 -0700437#define MMF_DUMP_ELF_HEADERS 6
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700438#define MMF_DUMP_HUGETLB_PRIVATE 7
439#define MMF_DUMP_HUGETLB_SHARED 8
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700440
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700441#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700442#define MMF_DUMP_FILTER_BITS 7
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700443#define MMF_DUMP_FILTER_MASK \
444 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
445#define MMF_DUMP_FILTER_DEFAULT \
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700446 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
Roland McGrath656eb2c2008-10-18 20:28:23 -0700447 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
448
449#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
450# define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
451#else
452# define MMF_DUMP_MASK_DEFAULT_ELF 0
453#endif
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700454 /* leave room for more dump flags */
455#define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
Andrea Arcangeliba761492011-01-13 15:46:58 -0800456#define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
Konstantin Khlebnikovbafb2822012-06-07 14:21:11 -0700457#define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700458
Oleg Nesterov9f68f6722012-08-19 16:15:09 +0200459#define MMF_HAS_UPROBES 19 /* has uprobes */
460#define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
Oleg Nesterovf8ac4ec2012-08-08 17:11:42 +0200461
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700462#define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
Kawai, Hidehiro6c5d5232007-07-19 01:48:27 -0700463
Linus Torvalds1da177e2005-04-16 15:20:36 -0700464struct sighand_struct {
465 atomic_t count;
466 struct k_sigaction action[_NSIG];
467 spinlock_t siglock;
Davide Libenzib8fceee2007-09-20 12:40:16 -0700468 wait_queue_head_t signalfd_wqh;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700469};
470
KaiGai Kohei0e464812006-06-25 05:49:24 -0700471struct pacct_struct {
KaiGai Koheif6ec29a2006-06-25 05:49:25 -0700472 int ac_flag;
473 long ac_exitcode;
KaiGai Kohei0e464812006-06-25 05:49:24 -0700474 unsigned long ac_mem;
KaiGai Kohei77787bf2006-06-25 05:49:26 -0700475 cputime_t ac_utime, ac_stime;
476 unsigned long ac_minflt, ac_majflt;
KaiGai Kohei0e464812006-06-25 05:49:24 -0700477};
478
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200479struct cpu_itimer {
480 cputime_t expires;
481 cputime_t incr;
Stanislaw Gruszka8356b5f2009-07-29 12:15:27 +0200482 u32 error;
483 u32 incr_error;
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200484};
485
Frank Mayharf06febc2008-09-12 09:54:39 -0700486/**
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +0100487 * struct cputime - snaphsot of system and user cputime
488 * @utime: time spent in user mode
489 * @stime: time spent in system mode
490 *
491 * Gathers a generic snapshot of user and system time.
492 */
493struct cputime {
494 cputime_t utime;
495 cputime_t stime;
496};
497
498/**
Frank Mayharf06febc2008-09-12 09:54:39 -0700499 * struct task_cputime - collected CPU time counts
500 * @utime: time spent in user mode, in &cputime_t units
501 * @stime: time spent in kernel mode, in &cputime_t units
502 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
Ingo Molnar5ce73a42008-09-14 17:11:46 +0200503 *
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +0100504 * This is an extension of struct cputime that includes the total runtime
505 * spent by the task from the scheduler point of view.
506 *
507 * As a result, this structure groups together three kinds of CPU time
508 * that are tracked for threads and thread groups. Most things considering
Frank Mayharf06febc2008-09-12 09:54:39 -0700509 * CPU time want to group these counts together and treat all three
510 * of them in parallel.
511 */
512struct task_cputime {
513 cputime_t utime;
514 cputime_t stime;
515 unsigned long long sum_exec_runtime;
516};
517/* Alternate field names when used to cache expirations. */
518#define prof_exp stime
519#define virt_exp utime
520#define sched_exp sum_exec_runtime
521
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100522#define INIT_CPUTIME \
523 (struct task_cputime) { \
Martin Schwidefsky64861632011-12-15 14:56:09 +0100524 .utime = 0, \
525 .stime = 0, \
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100526 .sum_exec_runtime = 0, \
527 }
528
Peter Zijlstraa233f112013-09-23 19:04:26 +0200529#ifdef CONFIG_PREEMPT_COUNT
530#define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
531#else
532#define PREEMPT_DISABLED PREEMPT_ENABLED
533#endif
534
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200535/*
536 * Disable preemption until the scheduler is running.
537 * Reset by start_kernel()->sched_init()->init_idle().
Peter Zijlstrad86ee482009-07-10 14:57:57 +0200538 *
539 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
540 * before the scheduler is active -- see should_resched().
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200541 */
Peter Zijlstraa233f112013-09-23 19:04:26 +0200542#define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200543
Frank Mayharf06febc2008-09-12 09:54:39 -0700544/**
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100545 * struct thread_group_cputimer - thread group interval timer counts
546 * @cputime: thread group interval timers.
547 * @running: non-zero when there are timers running and
548 * @cputime receives updates.
549 * @lock: lock for fields in this struct.
Frank Mayharf06febc2008-09-12 09:54:39 -0700550 *
551 * This structure contains the version of task_cputime, above, that is
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100552 * used for thread group CPU timer calculations.
Frank Mayharf06febc2008-09-12 09:54:39 -0700553 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100554struct thread_group_cputimer {
555 struct task_cputime cputime;
556 int running;
Thomas Gleixneree30a7b2009-07-25 18:56:56 +0200557 raw_spinlock_t lock;
Frank Mayharf06febc2008-09-12 09:54:39 -0700558};
Frank Mayharf06febc2008-09-12 09:54:39 -0700559
Ben Blum4714d1d2011-05-26 16:25:18 -0700560#include <linux/rwsem.h>
Mike Galbraith5091faa2010-11-30 14:18:03 +0100561struct autogroup;
562
Linus Torvalds1da177e2005-04-16 15:20:36 -0700563/*
Jonathan Neuschäfere815f0a2011-03-21 20:24:47 +0100564 * NOTE! "signal_struct" does not have its own
Linus Torvalds1da177e2005-04-16 15:20:36 -0700565 * locking, because a shared signal_struct always
566 * implies a shared sighand_struct, so locking
567 * sighand_struct is always a proper superset of
568 * the locking of signal_struct.
569 */
570struct signal_struct {
Oleg Nesterovea6d2902010-05-26 14:43:16 -0700571 atomic_t sigcnt;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700572 atomic_t live;
Oleg Nesterovb3ac0222010-05-26 14:43:24 -0700573 int nr_threads;
Oleg Nesterov0c740d02014-01-21 15:49:56 -0800574 struct list_head thread_head;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700575
576 wait_queue_head_t wait_chldexit; /* for wait4() */
577
578 /* current thread group signal load-balancing target: */
Ingo Molnar36c8b582006-07-03 00:25:41 -0700579 struct task_struct *curr_target;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700580
581 /* shared signal handling: */
582 struct sigpending shared_pending;
583
584 /* thread group exit support */
585 int group_exit_code;
586 /* overloaded:
587 * - notify group_exit_task when ->count is equal to notify_count
588 * - everyone except group_exit_task is stopped during signal delivery
589 * of fatal signals, group_exit_task processes the signal.
590 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700591 int notify_count;
Richard Kennedy07dd20e2008-08-01 13:18:04 +0100592 struct task_struct *group_exit_task;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700593
594 /* thread group stop support, overloads group_exit_code too */
595 int group_stop_count;
596 unsigned int flags; /* see SIGNAL_* flags below */
597
Lennart Poetteringebec18a2012-03-23 15:01:54 -0700598 /*
599 * PR_SET_CHILD_SUBREAPER marks a process, like a service
600 * manager, to re-parent orphan (double-forking) child processes
601 * to this process instead of 'init'. The service manager is
602 * able to receive SIGCHLD signals and is able to investigate
603 * the process until it calls wait(). All children of this
604 * process will inherit a flag if they should look for a
605 * child_subreaper process at exit.
606 */
607 unsigned int is_child_subreaper:1;
608 unsigned int has_child_subreaper:1;
609
Linus Torvalds1da177e2005-04-16 15:20:36 -0700610 /* POSIX.1b Interval Timers */
Pavel Emelyanov5ed67f02013-03-11 13:12:21 +0400611 int posix_timer_id;
612 struct list_head posix_timers;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700613
614 /* ITIMER_REAL timer for the process */
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800615 struct hrtimer real_timer;
Oleg Nesterovfea9d172008-02-08 04:19:19 -0800616 struct pid *leader_pid;
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800617 ktime_t it_real_incr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700618
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200619 /*
620 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
621 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
622 * values are defined to 0 and 1 respectively
623 */
624 struct cpu_itimer it[2];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700625
Frank Mayharf06febc2008-09-12 09:54:39 -0700626 /*
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100627 * Thread group totals for process CPU timers.
628 * See thread_group_cputimer(), et al, for details.
Frank Mayharf06febc2008-09-12 09:54:39 -0700629 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100630 struct thread_group_cputimer cputimer;
Frank Mayharf06febc2008-09-12 09:54:39 -0700631
632 /* Earliest-expiration cache. */
633 struct task_cputime cputime_expires;
634
635 struct list_head cpu_timers[3];
636
Eric W. Biedermanab521dc2007-02-12 00:53:00 -0800637 struct pid *tty_old_pgrp;
Cedric Le Goater1ec320a2006-12-08 02:37:55 -0800638
Linus Torvalds1da177e2005-04-16 15:20:36 -0700639 /* boolean value for session group leader */
640 int leader;
641
642 struct tty_struct *tty; /* NULL if no tty */
643
Mike Galbraith5091faa2010-11-30 14:18:03 +0100644#ifdef CONFIG_SCHED_AUTOGROUP
645 struct autogroup *autogroup;
646#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700647 /*
648 * Cumulative resource counters for dead threads in the group,
649 * and for reaped dead child processes forked by this group.
650 * Live threads maintain their own counters and add to these
651 * in __exit_signal, except for the group leader.
652 */
Peter Zijlstra32bd6712009-02-05 12:24:15 +0100653 cputime_t utime, stime, cutime, cstime;
Laurent Vivier9ac52312007-10-15 17:00:19 +0200654 cputime_t gtime;
655 cputime_t cgtime;
Frederic Weisbecker9fbc42e2013-02-25 17:25:39 +0100656#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +0100657 struct cputime prev_cputime;
Hidetoshi Seto0cf55e12009-12-02 17:28:07 +0900658#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700659 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
660 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
Eric Dumazet6eaeeab2007-05-10 22:22:37 -0700661 unsigned long inblock, oublock, cinblock, coublock;
Jiri Pirko1f102062009-09-22 16:44:10 -0700662 unsigned long maxrss, cmaxrss;
Andrea Righi940389b2008-07-28 00:48:12 +0200663 struct task_io_accounting ioac;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700664
665 /*
Peter Zijlstra32bd6712009-02-05 12:24:15 +0100666 * Cumulative ns of schedule CPU time fo dead threads in the
667 * group, not including a zombie group leader, (This only differs
668 * from jiffies_to_ns(utime + stime) if sched_clock uses something
669 * other than jiffies.)
670 */
671 unsigned long long sum_sched_runtime;
672
673 /*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700674 * We don't bother to synchronize most readers of this at all,
675 * because there is no reader checking a limit that actually needs
676 * to get both rlim_cur and rlim_max atomically, and either one
677 * alone is a single word that can safely be read normally.
678 * getrlimit/setrlimit use task_lock(current->group_leader) to
679 * protect this instead of the siglock, because they really
680 * have no need to disable irqs.
681 */
682 struct rlimit rlim[RLIM_NLIMITS];
683
KaiGai Kohei0e464812006-06-25 05:49:24 -0700684#ifdef CONFIG_BSD_PROCESS_ACCT
685 struct pacct_struct pacct; /* per-process accounting information */
686#endif
Shailabh Nagarad4ecbc2006-07-14 00:24:44 -0700687#ifdef CONFIG_TASKSTATS
Shailabh Nagarad4ecbc2006-07-14 00:24:44 -0700688 struct taskstats *stats;
689#endif
Miloslav Trmac522ed772007-07-15 23:40:56 -0700690#ifdef CONFIG_AUDIT
691 unsigned audit_tty;
Richard Guy Briggs46e959e2013-05-03 14:03:50 -0400692 unsigned audit_tty_log_passwd;
Miloslav Trmac522ed772007-07-15 23:40:56 -0700693 struct tty_audit_buf *tty_audit_buf;
694#endif
Ben Blum4714d1d2011-05-26 16:25:18 -0700695#ifdef CONFIG_CGROUPS
696 /*
Tejun Heo77e4ef92011-12-12 18:12:21 -0800697 * group_rwsem prevents new tasks from entering the threadgroup and
698 * member tasks from exiting,a more specifically, setting of
699 * PF_EXITING. fork and exit paths are protected with this rwsem
700 * using threadgroup_change_begin/end(). Users which require
701 * threadgroup to remain stable should use threadgroup_[un]lock()
702 * which also takes care of exec path. Currently, cgroup is the
703 * only user.
Ben Blum4714d1d2011-05-26 16:25:18 -0700704 */
Tejun Heo257058a2011-12-12 18:12:21 -0800705 struct rw_semaphore group_rwsem;
Ben Blum4714d1d2011-05-26 16:25:18 -0700706#endif
KOSAKI Motohiro28b83c52009-09-21 17:03:13 -0700707
David Rientjese1e12d22012-12-11 16:02:56 -0800708 oom_flags_t oom_flags;
David Rientjesa9c58b902012-12-11 16:02:54 -0800709 short oom_score_adj; /* OOM kill score adjustment */
710 short oom_score_adj_min; /* OOM kill score adjustment min value.
711 * Only settable by CAP_SYS_RESOURCE. */
KOSAKI Motohiro9b1bf122010-10-27 15:34:08 -0700712
713 struct mutex cred_guard_mutex; /* guard against foreign influences on
714 * credential calculations
715 * (notably. ptrace) */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700716};
717
718/*
719 * Bits in flags field of signal_struct.
720 */
721#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
Oleg Nesterovee77f072011-04-01 20:12:38 +0200722#define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
723#define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
Oleg Nesterov403bad72013-04-30 15:28:10 -0700724#define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
Oleg Nesterove4420552008-04-30 00:52:44 -0700725/*
726 * Pending notifications to parent.
727 */
728#define SIGNAL_CLD_STOPPED 0x00000010
729#define SIGNAL_CLD_CONTINUED 0x00000020
730#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700731
Oleg Nesterovfae5fa42008-04-30 00:53:03 -0700732#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
733
Oleg Nesteroved5d2ca2008-02-04 22:27:24 -0800734/* If true, all threads except ->group_exit_task have pending SIGKILL */
735static inline int signal_group_exit(const struct signal_struct *sig)
736{
737 return (sig->flags & SIGNAL_GROUP_EXIT) ||
738 (sig->group_exit_task != NULL);
739}
740
Linus Torvalds1da177e2005-04-16 15:20:36 -0700741/*
742 * Some day this will be a full-fledged user tracking system..
743 */
744struct user_struct {
745 atomic_t __count; /* reference count */
746 atomic_t processes; /* How many processes does this user have? */
747 atomic_t files; /* How many open files does this user have? */
748 atomic_t sigpending; /* How many pending signals does this user have? */
Amy Griffis2d9048e2006-06-01 13:10:59 -0700749#ifdef CONFIG_INOTIFY_USER
Robert Love0eeca282005-07-12 17:06:03 -0400750 atomic_t inotify_watches; /* How many inotify watches does this user have? */
751 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
752#endif
Eric Paris4afeff82010-10-28 17:21:58 -0400753#ifdef CONFIG_FANOTIFY
754 atomic_t fanotify_listeners;
755#endif
Davide Libenzi7ef99642008-12-01 13:13:55 -0800756#ifdef CONFIG_EPOLL
Robin Holt52bd19f72011-01-12 17:00:01 -0800757 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
Davide Libenzi7ef99642008-12-01 13:13:55 -0800758#endif
Alexey Dobriyan970a8642007-10-16 23:30:09 -0700759#ifdef CONFIG_POSIX_MQUEUE
Linus Torvalds1da177e2005-04-16 15:20:36 -0700760 /* protected by mq_lock */
761 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
Alexey Dobriyan970a8642007-10-16 23:30:09 -0700762#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700763 unsigned long locked_shm; /* How many pages of mlocked shm ? */
764
765#ifdef CONFIG_KEYS
766 struct key *uid_keyring; /* UID specific keyring */
767 struct key *session_keyring; /* UID's default session keyring */
768#endif
769
770 /* Hash table maintenance information */
Pavel Emelyanov735de222007-09-18 22:46:44 -0700771 struct hlist_node uidhash_node;
Eric W. Biederman7b44ab92011-11-16 23:20:58 -0800772 kuid_t uid;
Srivatsa Vaddagiri24e377a2007-10-15 17:00:09 +0200773
Ingo Molnarcdd6c482009-09-21 12:02:48 +0200774#ifdef CONFIG_PERF_EVENTS
Peter Zijlstra789f90f2009-05-15 15:19:27 +0200775 atomic_long_t locked_vm;
776#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700777};
778
Kay Sieverseb41d942007-11-02 13:47:53 +0100779extern int uids_sysfs_init(void);
Dhaval Giani5cb350b2007-10-15 17:00:14 +0200780
Eric W. Biederman7b44ab92011-11-16 23:20:58 -0800781extern struct user_struct *find_user(kuid_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700782
783extern struct user_struct root_user;
784#define INIT_USER (&root_user)
785
David Howellsb6dff3e2008-11-14 10:39:16 +1100786
Linus Torvalds1da177e2005-04-16 15:20:36 -0700787struct backing_dev_info;
788struct reclaim_state;
789
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700790#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700791struct sched_info {
792 /* cumulative counters */
Ingo Molnar2d723762007-10-15 17:00:12 +0200793 unsigned long pcount; /* # of times run on this cpu */
Ken Chen9c2c4802008-12-16 23:41:22 -0800794 unsigned long long run_delay; /* time spent waiting on a runqueue */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700795
796 /* timestamps */
Balbir Singh172ba842007-07-09 18:52:00 +0200797 unsigned long long last_arrival,/* when we last ran on a cpu */
798 last_queued; /* when we were last queued to run */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700799};
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700800#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700801
Shailabh Nagarca74e922006-07-14 00:24:36 -0700802#ifdef CONFIG_TASK_DELAY_ACCT
803struct task_delay_info {
804 spinlock_t lock;
805 unsigned int flags; /* Private per-task flags */
806
807 /* For each stat XXX, add following, aligned appropriately
808 *
809 * struct timespec XXX_start, XXX_end;
810 * u64 XXX_delay;
811 * u32 XXX_count;
812 *
813 * Atomicity of updates to XXX_delay, XXX_count protected by
814 * single lock above (split into XXX_lock if contention is an issue).
815 */
Shailabh Nagar0ff92242006-07-14 00:24:37 -0700816
817 /*
818 * XXX_count is incremented on every XXX operation, the delay
819 * associated with the operation is added to XXX_delay.
820 * XXX_delay contains the accumulated delay time in nanoseconds.
821 */
822 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
823 u64 blkio_delay; /* wait for sync block io completion */
824 u64 swapin_delay; /* wait for swapin block io completion */
825 u32 blkio_count; /* total count of the number of sync block */
826 /* io operations performed */
827 u32 swapin_count; /* total count of the number of swapin block */
828 /* io operations performed */
Keika Kobayashi873b4772008-07-25 01:48:52 -0700829
830 struct timespec freepages_start, freepages_end;
831 u64 freepages_delay; /* wait for memory reclaim */
832 u32 freepages_count; /* total count of memory reclaim */
Shailabh Nagarca74e922006-07-14 00:24:36 -0700833};
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700834#endif /* CONFIG_TASK_DELAY_ACCT */
835
836static inline int sched_info_on(void)
837{
838#ifdef CONFIG_SCHEDSTATS
839 return 1;
840#elif defined(CONFIG_TASK_DELAY_ACCT)
841 extern int delayacct_on;
842 return delayacct_on;
843#else
844 return 0;
Shailabh Nagarca74e922006-07-14 00:24:36 -0700845#endif
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700846}
Shailabh Nagarca74e922006-07-14 00:24:36 -0700847
Ingo Molnard15bcfd2007-07-09 18:51:57 +0200848enum cpu_idle_type {
849 CPU_IDLE,
850 CPU_NOT_IDLE,
851 CPU_NEWLY_IDLE,
852 CPU_MAX_IDLE_TYPES
Linus Torvalds1da177e2005-04-16 15:20:36 -0700853};
854
855/*
Nikhil Rao1399fa72011-05-18 10:09:39 -0700856 * Increase resolution of cpu_power calculations
857 */
858#define SCHED_POWER_SHIFT 10
859#define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700860
Nikhil Rao1399fa72011-05-18 10:09:39 -0700861/*
862 * sched-domains (multiprocessor balancing) declarations:
863 */
Peter Williams2dd73a42006-06-27 02:54:34 -0700864#ifdef CONFIG_SMP
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200865#define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
866#define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
867#define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
868#define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
Peter Zijlstrac88d5912009-09-10 13:50:02 +0200869#define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200870#define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200871#define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200872#define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
873#define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
Michael Neuling532cb4c2010-06-08 14:57:02 +1000874#define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200875#define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
Peter Zijlstrae3589f62011-07-15 10:35:52 +0200876#define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
Mel Gorman3a7053b2013-10-07 11:29:00 +0100877#define SD_NUMA 0x4000 /* cross-node balancing */
Siddha, Suresh B5c45bf22006-06-27 02:54:42 -0700878
Michael Neuling532cb4c2010-06-08 14:57:02 +1000879extern int __weak arch_sd_sibiling_asym_packing(void);
880
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +0900881struct sched_domain_attr {
882 int relax_domain_level;
883};
884
885#define SD_ATTR_INIT (struct sched_domain_attr) { \
886 .relax_domain_level = -1, \
887}
888
Peter Zijlstra60495e72011-04-07 14:10:04 +0200889extern int sched_domain_level_max;
890
Li Zefan5e6521e2013-03-05 16:06:23 +0800891struct sched_group;
892
Linus Torvalds1da177e2005-04-16 15:20:36 -0700893struct sched_domain {
894 /* These fields must be setup */
895 struct sched_domain *parent; /* top domain must be null terminated */
Siddha, Suresh B1a848872006-10-03 01:14:08 -0700896 struct sched_domain *child; /* bottom domain must be null terminated */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700897 struct sched_group *groups; /* the balancing groups of the domain */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700898 unsigned long min_interval; /* Minimum balance interval ms */
899 unsigned long max_interval; /* Maximum balance interval ms */
900 unsigned int busy_factor; /* less balancing by factor if busy */
901 unsigned int imbalance_pct; /* No balance until over watermark */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700902 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
Nick Piggin78979862005-06-25 14:57:13 -0700903 unsigned int busy_idx;
904 unsigned int idle_idx;
905 unsigned int newidle_idx;
906 unsigned int wake_idx;
Nick Piggin147cbb42005-06-25 14:57:19 -0700907 unsigned int forkexec_idx;
Peter Zijlstraa52bfd732009-09-01 10:34:35 +0200908 unsigned int smt_gain;
Vincent Guittot25f55d92013-04-23 16:59:02 +0200909
910 int nohz_idle; /* NOHZ IDLE status */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700911 int flags; /* See SD_* */
Peter Zijlstra60495e72011-04-07 14:10:04 +0200912 int level;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700913
914 /* Runtime fields. */
915 unsigned long last_balance; /* init to jiffies. units in jiffies */
916 unsigned int balance_interval; /* initialise to 1. units in ms. */
917 unsigned int nr_balance_failed; /* initialise to 0 */
918
Jason Lowf48627e2013-09-13 11:26:53 -0700919 /* idle_balance() stats */
Jason Low9bd721c2013-09-13 11:26:52 -0700920 u64 max_newidle_lb_cost;
Jason Lowf48627e2013-09-13 11:26:53 -0700921 unsigned long next_decay_max_lb_cost;
Peter Zijlstra2398f2c2008-06-27 13:41:35 +0200922
Linus Torvalds1da177e2005-04-16 15:20:36 -0700923#ifdef CONFIG_SCHEDSTATS
924 /* load_balance() stats */
Ken Chen480b9432007-10-18 21:32:56 +0200925 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
926 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
927 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
928 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
929 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
930 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
931 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
932 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700933
934 /* Active load balancing */
Ken Chen480b9432007-10-18 21:32:56 +0200935 unsigned int alb_count;
936 unsigned int alb_failed;
937 unsigned int alb_pushed;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700938
Nick Piggin68767a02005-06-25 14:57:20 -0700939 /* SD_BALANCE_EXEC stats */
Ken Chen480b9432007-10-18 21:32:56 +0200940 unsigned int sbe_count;
941 unsigned int sbe_balanced;
942 unsigned int sbe_pushed;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700943
Nick Piggin68767a02005-06-25 14:57:20 -0700944 /* SD_BALANCE_FORK stats */
Ken Chen480b9432007-10-18 21:32:56 +0200945 unsigned int sbf_count;
946 unsigned int sbf_balanced;
947 unsigned int sbf_pushed;
Nick Piggin68767a02005-06-25 14:57:20 -0700948
Linus Torvalds1da177e2005-04-16 15:20:36 -0700949 /* try_to_wake_up() stats */
Ken Chen480b9432007-10-18 21:32:56 +0200950 unsigned int ttwu_wake_remote;
951 unsigned int ttwu_move_affine;
952 unsigned int ttwu_move_balance;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700953#endif
Ingo Molnara5d8c342008-10-09 11:35:51 +0200954#ifdef CONFIG_SCHED_DEBUG
955 char *name;
956#endif
Peter Zijlstradce840a2011-04-07 14:09:50 +0200957 union {
958 void *private; /* used during construction */
959 struct rcu_head rcu; /* used during destruction */
960 };
Rusty Russell6c99e9a2008-11-25 02:35:04 +1030961
Peter Zijlstra669c55e2010-04-16 14:59:29 +0200962 unsigned int span_weight;
Ingo Molnar4200efd2009-05-19 09:22:19 +0200963 /*
964 * Span of all CPUs in this domain.
965 *
966 * NOTE: this field is variable length. (Allocated dynamically
967 * by attaching extra space to the end of the structure,
968 * depending on how many CPUs the kernel has booted up with)
Ingo Molnar4200efd2009-05-19 09:22:19 +0200969 */
970 unsigned long span[0];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700971};
972
Rusty Russell758b2cd2008-11-25 02:35:04 +1030973static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
974{
Rusty Russell6c99e9a2008-11-25 02:35:04 +1030975 return to_cpumask(sd->span);
Rusty Russell758b2cd2008-11-25 02:35:04 +1030976}
977
Rusty Russellacc3f5d2009-11-03 14:53:40 +1030978extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +0900979 struct sched_domain_attr *dattr_new);
Paul Jackson029190c2007-10-18 23:40:20 -0700980
Rusty Russellacc3f5d2009-11-03 14:53:40 +1030981/* Allocate an array of sched domains, for partition_sched_domains(). */
982cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
983void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
984
Peter Zijlstra39be3502012-01-26 12:44:34 +0100985bool cpus_share_cache(int this_cpu, int that_cpu);
986
Ingo Molnar1b427c12008-07-18 14:01:39 +0200987#else /* CONFIG_SMP */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700988
Ingo Molnar1b427c12008-07-18 14:01:39 +0200989struct sched_domain_attr;
990
991static inline void
Rusty Russellacc3f5d2009-11-03 14:53:40 +1030992partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
Ingo Molnar1b427c12008-07-18 14:01:39 +0200993 struct sched_domain_attr *dattr_new)
Con Kolivasd02c7a82007-07-26 13:40:43 +0200994{
Con Kolivasd02c7a82007-07-26 13:40:43 +0200995}
Peter Zijlstra39be3502012-01-26 12:44:34 +0100996
997static inline bool cpus_share_cache(int this_cpu, int that_cpu)
998{
999 return true;
1000}
1001
Ingo Molnar1b427c12008-07-18 14:01:39 +02001002#endif /* !CONFIG_SMP */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001003
Peter Zijlstra47fe38f2009-09-02 13:49:18 +02001004
Linus Torvalds1da177e2005-04-16 15:20:36 -07001005struct io_context; /* See blkdev.h */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001006
Linus Torvalds1da177e2005-04-16 15:20:36 -07001007
Chen, Kenneth W383f2832005-09-09 13:02:02 -07001008#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
Ingo Molnar36c8b582006-07-03 00:25:41 -07001009extern void prefetch_stack(struct task_struct *t);
Chen, Kenneth W383f2832005-09-09 13:02:02 -07001010#else
1011static inline void prefetch_stack(struct task_struct *t) { }
1012#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001013
1014struct audit_context; /* See audit.c */
1015struct mempolicy;
Jens Axboeb92ce552006-04-11 13:52:07 +02001016struct pipe_inode_info;
Serge E. Hallyn4865ecf2006-10-02 02:18:14 -07001017struct uts_namespace;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001018
Ingo Molnar20b8a592007-07-09 18:51:58 +02001019struct load_weight {
Peter Zijlstra9dbdb152013-11-18 18:27:06 +01001020 unsigned long weight;
1021 u32 inv_weight;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001022};
1023
Paul Turner9d85f212012-10-04 13:18:29 +02001024struct sched_avg {
1025 /*
1026 * These sums represent an infinite geometric series and so are bound
Kamalesh Babulal239003e2013-06-27 11:34:09 +05301027 * above by 1024/(1-y). Thus we only need a u32 to store them for all
Paul Turner9d85f212012-10-04 13:18:29 +02001028 * choices of y < 1-2^(-32)*1024.
1029 */
1030 u32 runnable_avg_sum, runnable_avg_period;
1031 u64 last_runnable_update;
Paul Turner9ee474f2012-10-04 13:18:30 +02001032 s64 decay_count;
Paul Turner2dac7542012-10-04 13:18:30 +02001033 unsigned long load_avg_contrib;
Paul Turner9d85f212012-10-04 13:18:29 +02001034};
1035
Ingo Molnar94c18222007-08-02 17:41:40 +02001036#ifdef CONFIG_SCHEDSTATS
Lucas De Marchi41acab82010-03-10 23:37:45 -03001037struct sched_statistics {
Ingo Molnar94c18222007-08-02 17:41:40 +02001038 u64 wait_start;
1039 u64 wait_max;
Arjan van de Ven6d082592008-01-25 21:08:35 +01001040 u64 wait_count;
1041 u64 wait_sum;
Arjan van de Ven8f0dfc32009-07-20 11:26:58 -07001042 u64 iowait_count;
1043 u64 iowait_sum;
Ingo Molnar94c18222007-08-02 17:41:40 +02001044
1045 u64 sleep_start;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001046 u64 sleep_max;
Ingo Molnar94c18222007-08-02 17:41:40 +02001047 s64 sum_sleep_runtime;
1048
1049 u64 block_start;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001050 u64 block_max;
1051 u64 exec_max;
Ingo Molnareba1ed42007-10-15 17:00:02 +02001052 u64 slice_max;
Ingo Molnarcc367732007-10-15 17:00:18 +02001053
Ingo Molnarcc367732007-10-15 17:00:18 +02001054 u64 nr_migrations_cold;
1055 u64 nr_failed_migrations_affine;
1056 u64 nr_failed_migrations_running;
1057 u64 nr_failed_migrations_hot;
1058 u64 nr_forced_migrations;
Ingo Molnarcc367732007-10-15 17:00:18 +02001059
1060 u64 nr_wakeups;
1061 u64 nr_wakeups_sync;
1062 u64 nr_wakeups_migrate;
1063 u64 nr_wakeups_local;
1064 u64 nr_wakeups_remote;
1065 u64 nr_wakeups_affine;
1066 u64 nr_wakeups_affine_attempts;
1067 u64 nr_wakeups_passive;
1068 u64 nr_wakeups_idle;
Lucas De Marchi41acab82010-03-10 23:37:45 -03001069};
1070#endif
1071
1072struct sched_entity {
1073 struct load_weight load; /* for load-balancing */
1074 struct rb_node run_node;
1075 struct list_head group_node;
1076 unsigned int on_rq;
1077
1078 u64 exec_start;
1079 u64 sum_exec_runtime;
1080 u64 vruntime;
1081 u64 prev_sum_exec_runtime;
1082
Lucas De Marchi41acab82010-03-10 23:37:45 -03001083 u64 nr_migrations;
1084
Lucas De Marchi41acab82010-03-10 23:37:45 -03001085#ifdef CONFIG_SCHEDSTATS
1086 struct sched_statistics statistics;
Ingo Molnar94c18222007-08-02 17:41:40 +02001087#endif
1088
Ingo Molnar20b8a592007-07-09 18:51:58 +02001089#ifdef CONFIG_FAIR_GROUP_SCHED
Peter Zijlstrafed14d42012-02-11 06:05:00 +01001090 int depth;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001091 struct sched_entity *parent;
1092 /* rq on which this entity is (to be) queued: */
1093 struct cfs_rq *cfs_rq;
1094 /* rq "owned" by this entity/group: */
1095 struct cfs_rq *my_q;
1096#endif
Clark Williams8bd75c72013-02-07 09:47:07 -06001097
Alex Shi141965c2013-06-26 13:05:39 +08001098#ifdef CONFIG_SMP
Paul Turnerf4e26b12012-10-04 13:18:32 +02001099 /* Per-entity load-tracking */
Paul Turner9d85f212012-10-04 13:18:29 +02001100 struct sched_avg avg;
1101#endif
Ingo Molnar20b8a592007-07-09 18:51:58 +02001102};
Ingo Molnar70b97a72006-07-03 00:25:42 -07001103
Peter Zijlstrafa717062008-01-25 21:08:27 +01001104struct sched_rt_entity {
1105 struct list_head run_list;
Peter Zijlstra78f2c7d2008-01-25 21:08:27 +01001106 unsigned long timeout;
Ying Xue57d2aa02012-07-17 15:03:43 +08001107 unsigned long watchdog_stamp;
Richard Kennedybee367e2008-08-01 13:24:08 +01001108 unsigned int time_slice;
Peter Zijlstra6f505b12008-01-25 21:08:30 +01001109
Peter Zijlstra58d6c2d2008-04-19 19:45:00 +02001110 struct sched_rt_entity *back;
Peter Zijlstra052f1dc2008-02-13 15:45:40 +01001111#ifdef CONFIG_RT_GROUP_SCHED
Peter Zijlstra6f505b12008-01-25 21:08:30 +01001112 struct sched_rt_entity *parent;
1113 /* rq on which this entity is (to be) queued: */
1114 struct rt_rq *rt_rq;
1115 /* rq "owned" by this entity/group: */
1116 struct rt_rq *my_q;
1117#endif
Peter Zijlstrafa717062008-01-25 21:08:27 +01001118};
1119
Dario Faggioliaab03e02013-11-28 11:14:43 +01001120struct sched_dl_entity {
1121 struct rb_node rb_node;
1122
1123 /*
1124 * Original scheduling parameters. Copied here from sched_attr
1125 * during sched_setscheduler2(), they will remain the same until
1126 * the next sched_setscheduler2().
1127 */
1128 u64 dl_runtime; /* maximum runtime for each instance */
1129 u64 dl_deadline; /* relative deadline of each instance */
Harald Gustafsson755378a2013-11-07 14:43:40 +01001130 u64 dl_period; /* separation of two instances (period) */
Dario Faggioli332ac172013-11-07 14:43:45 +01001131 u64 dl_bw; /* dl_runtime / dl_deadline */
Dario Faggioliaab03e02013-11-28 11:14:43 +01001132
1133 /*
1134 * Actual scheduling parameters. Initialized with the values above,
1135 * they are continously updated during task execution. Note that
1136 * the remaining runtime could be < 0 in case we are in overrun.
1137 */
1138 s64 runtime; /* remaining runtime for this instance */
1139 u64 deadline; /* absolute deadline for this instance */
1140 unsigned int flags; /* specifying the scheduler behaviour */
1141
1142 /*
1143 * Some bool flags:
1144 *
1145 * @dl_throttled tells if we exhausted the runtime. If so, the
1146 * task has to wait for a replenishment to be performed at the
1147 * next firing of dl_timer.
1148 *
1149 * @dl_new tells if a new instance arrived. If so we must
1150 * start executing it with full runtime and reset its absolute
1151 * deadline;
Dario Faggioli2d3d8912013-11-07 14:43:44 +01001152 *
1153 * @dl_boosted tells if we are boosted due to DI. If so we are
1154 * outside bandwidth enforcement mechanism (but only until we
1155 * exit the critical section).
Dario Faggioliaab03e02013-11-28 11:14:43 +01001156 */
Dario Faggioli2d3d8912013-11-07 14:43:44 +01001157 int dl_throttled, dl_new, dl_boosted;
Dario Faggioliaab03e02013-11-28 11:14:43 +01001158
1159 /*
1160 * Bandwidth enforcement timer. Each -deadline task has its
1161 * own bandwidth to be enforced, thus we need one timer per task.
1162 */
1163 struct hrtimer dl_timer;
1164};
Clark Williams8bd75c72013-02-07 09:47:07 -06001165
Paul E. McKenney86848962009-08-27 15:00:12 -07001166struct rcu_node;
1167
Peter Zijlstra8dc85d5472010-09-02 16:50:03 +02001168enum perf_event_task_context {
1169 perf_invalid_context = -1,
1170 perf_hw_context = 0,
Peter Zijlstra89a1e182010-09-07 17:34:50 +02001171 perf_sw_context,
Peter Zijlstra8dc85d5472010-09-02 16:50:03 +02001172 perf_nr_task_contexts,
1173};
1174
Linus Torvalds1da177e2005-04-16 15:20:36 -07001175struct task_struct {
1176 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
Roman Zippelf7e42172007-05-09 02:35:17 -07001177 void *stack;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001178 atomic_t usage;
William Cohen97dc32c2007-05-08 00:23:41 -07001179 unsigned int flags; /* per process flags, defined below */
1180 unsigned int ptrace;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001181
Peter Williams2dd73a42006-06-27 02:54:34 -07001182#ifdef CONFIG_SMP
Peter Zijlstrafa14ff42011-09-12 13:06:17 +02001183 struct llist_node wake_entry;
Peter Zijlstra3ca7a442011-04-05 17:23:40 +02001184 int on_cpu;
Michael Wang62470412013-07-04 12:55:51 +08001185 struct task_struct *last_wakee;
1186 unsigned long wakee_flips;
1187 unsigned long wakee_flip_decay_ts;
Peter Zijlstraac66f542013-10-07 11:29:16 +01001188
1189 int wake_cpu;
Nick Piggin4866cde2005-06-25 14:57:23 -07001190#endif
Peter Zijlstrafd2f4412011-04-05 17:23:44 +02001191 int on_rq;
Ingo Molnar50e645a2007-07-09 18:52:00 +02001192
Ingo Molnarb29739f2006-06-27 02:54:51 -07001193 int prio, static_prio, normal_prio;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001194 unsigned int rt_priority;
Ingo Molnar5522d5d2007-10-15 17:00:12 +02001195 const struct sched_class *sched_class;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001196 struct sched_entity se;
Peter Zijlstrafa717062008-01-25 21:08:27 +01001197 struct sched_rt_entity rt;
Peter Zijlstra8323f262012-06-22 13:36:05 +02001198#ifdef CONFIG_CGROUP_SCHED
1199 struct task_group *sched_task_group;
1200#endif
Dario Faggioliaab03e02013-11-28 11:14:43 +01001201 struct sched_dl_entity dl;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001202
Avi Kivitye107be32007-07-26 13:40:43 +02001203#ifdef CONFIG_PREEMPT_NOTIFIERS
1204 /* list of struct preempt_notifier: */
1205 struct hlist_head preempt_notifiers;
1206#endif
1207
Alexey Dobriyan6c5c9342006-09-29 01:59:40 -07001208#ifdef CONFIG_BLK_DEV_IO_TRACE
Jens Axboe2056a782006-03-23 20:00:26 +01001209 unsigned int btrace_seq;
Alexey Dobriyan6c5c9342006-09-29 01:59:40 -07001210#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001211
William Cohen97dc32c2007-05-08 00:23:41 -07001212 unsigned int policy;
Peter Zijlstra29baa742012-04-23 12:11:21 +02001213 int nr_cpus_allowed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001214 cpumask_t cpus_allowed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001215
Paul E. McKenneya57eb942010-06-29 16:49:16 -07001216#ifdef CONFIG_PREEMPT_RCU
Paul E. McKenneye260be62008-01-25 21:08:24 +01001217 int rcu_read_lock_nesting;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001218 char rcu_read_unlock_special;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001219 struct list_head rcu_node_entry;
Paul E. McKenneya57eb942010-06-29 16:49:16 -07001220#endif /* #ifdef CONFIG_PREEMPT_RCU */
1221#ifdef CONFIG_TREE_PREEMPT_RCU
1222 struct rcu_node *rcu_blocked_node;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001223#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
Paul E. McKenney24278d12010-09-27 17:25:23 -07001224#ifdef CONFIG_RCU_BOOST
1225 struct rt_mutex *rcu_boost_mutex;
1226#endif /* #ifdef CONFIG_RCU_BOOST */
Paul E. McKenneye260be62008-01-25 21:08:24 +01001227
Chandra Seetharaman52f17b62006-07-14 00:24:38 -07001228#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001229 struct sched_info sched_info;
1230#endif
1231
1232 struct list_head tasks;
Dario Faggioli806c09a2010-11-30 19:51:33 +01001233#ifdef CONFIG_SMP
Gregory Haskins917b6272008-12-29 09:39:53 -05001234 struct plist_node pushable_tasks;
Juri Lelli1baca4c2013-11-07 14:43:38 +01001235 struct rb_node pushable_dl_tasks;
Dario Faggioli806c09a2010-11-30 19:51:33 +01001236#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001237
1238 struct mm_struct *mm, *active_mm;
Jiri Kosina4471a672011-04-14 15:22:09 -07001239#ifdef CONFIG_COMPAT_BRK
1240 unsigned brk_randomized:1;
1241#endif
Davidlohr Bueso615d6e82014-04-07 15:37:25 -07001242 /* per-thread vma caching */
1243 u32 vmacache_seqnum;
1244 struct vm_area_struct *vmacache[VMACACHE_SIZE];
KAMEZAWA Hiroyuki34e55232010-03-05 13:41:40 -08001245#if defined(SPLIT_RSS_COUNTING)
1246 struct task_rss_stat rss_stat;
1247#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001248/* task state */
William Cohen97dc32c2007-05-08 00:23:41 -07001249 int exit_state;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001250 int exit_code, exit_signal;
1251 int pdeath_signal; /* The signal sent when the parent dies */
Tejun Heoa8f072c2011-06-02 11:13:59 +02001252 unsigned int jobctl; /* JOBCTL_*, siglock protected */
Andrei Epure9b89f6b2013-04-11 20:30:29 +03001253
1254 /* Used for emulating ABI behavior of previous Linux versions */
William Cohen97dc32c2007-05-08 00:23:41 -07001255 unsigned int personality;
Andrei Epure9b89f6b2013-04-11 20:30:29 +03001256
Kentaro Takedaf9ce1f12009-02-05 17:18:11 +09001257 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1258 * execve */
Arjan van de Ven8f0dfc32009-07-20 11:26:58 -07001259 unsigned in_iowait:1;
1260
Andy Lutomirski259e5e62012-04-12 16:47:50 -05001261 /* task may not gain privileges */
1262 unsigned no_new_privs:1;
Lennart Poetteringca94c442009-06-15 17:17:47 +02001263
1264 /* Revert to default priority/policy when forking */
1265 unsigned sched_reset_on_fork:1;
Peter Zijlstraa8e4f2e2011-04-05 17:23:49 +02001266 unsigned sched_contributes_to_load:1;
Lennart Poetteringca94c442009-06-15 17:17:47 +02001267
Linus Torvalds1da177e2005-04-16 15:20:36 -07001268 pid_t pid;
1269 pid_t tgid;
Arjan van de Ven0a4254052006-09-26 10:52:38 +02001270
Hiroshi Shimamoto13145622009-08-18 15:06:02 +09001271#ifdef CONFIG_CC_STACKPROTECTOR
Arjan van de Ven0a4254052006-09-26 10:52:38 +02001272 /* Canary value for the -fstack-protector gcc feature */
1273 unsigned long stack_canary;
Hiroshi Shimamoto13145622009-08-18 15:06:02 +09001274#endif
Oleg Nesterov4d1d61a2012-05-11 10:59:08 +10001275 /*
Linus Torvalds1da177e2005-04-16 15:20:36 -07001276 * pointers to (original) parent process, youngest child, younger sibling,
Oleg Nesterov4d1d61a2012-05-11 10:59:08 +10001277 * older sibling, respectively. (p->father can be replaced with
Roland McGrathf4700212008-03-24 18:36:23 -07001278 * p->real_parent->pid)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001279 */
Kees Cookabd63bc2011-12-14 14:39:26 -08001280 struct task_struct __rcu *real_parent; /* real parent process */
1281 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001282 /*
Roland McGrathf4700212008-03-24 18:36:23 -07001283 * children/sibling forms the list of my natural children
Linus Torvalds1da177e2005-04-16 15:20:36 -07001284 */
1285 struct list_head children; /* list of my children */
1286 struct list_head sibling; /* linkage in my parent's children list */
1287 struct task_struct *group_leader; /* threadgroup leader */
1288
Roland McGrathf4700212008-03-24 18:36:23 -07001289 /*
1290 * ptraced is the list of tasks this task is using ptrace on.
1291 * This includes both natural children and PTRACE_ATTACH targets.
1292 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1293 */
1294 struct list_head ptraced;
1295 struct list_head ptrace_entry;
1296
Linus Torvalds1da177e2005-04-16 15:20:36 -07001297 /* PID/PID hash table linkage. */
Eric W. Biederman92476d72006-03-31 02:31:42 -08001298 struct pid_link pids[PIDTYPE_MAX];
Oleg Nesterov47e65322006-03-28 16:11:25 -08001299 struct list_head thread_group;
Oleg Nesterov0c740d02014-01-21 15:49:56 -08001300 struct list_head thread_node;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001301
1302 struct completion *vfork_done; /* for vfork() */
1303 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1304 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1305
Michael Neulingc66f08b2007-10-18 03:06:34 -07001306 cputime_t utime, stime, utimescaled, stimescaled;
Laurent Vivier9ac52312007-10-15 17:00:19 +02001307 cputime_t gtime;
Frederic Weisbecker9fbc42e2013-02-25 17:25:39 +01001308#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +01001309 struct cputime prev_cputime;
Hidetoshi Setod99ca3b2009-12-02 17:26:47 +09001310#endif
Frederic Weisbecker6a616712012-12-16 20:00:34 +01001311#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1312 seqlock_t vtime_seqlock;
1313 unsigned long long vtime_snap;
1314 enum {
1315 VTIME_SLEEPING = 0,
1316 VTIME_USER,
1317 VTIME_SYS,
1318 } vtime_snap_whence;
1319#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001320 unsigned long nvcsw, nivcsw; /* context switch counts */
Tomas Janousek924b42d2007-07-15 23:39:42 -07001321 struct timespec start_time; /* monotonic time */
1322 struct timespec real_start_time; /* boot based time */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001323/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1324 unsigned long min_flt, maj_flt;
1325
Frank Mayharf06febc2008-09-12 09:54:39 -07001326 struct task_cputime cputime_expires;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001327 struct list_head cpu_timers[3];
1328
1329/* process credentials */
Arnd Bergmann1b0ba1c2010-02-24 19:45:09 +01001330 const struct cred __rcu *real_cred; /* objective and real subjective task
David Howells3b11a1d2008-11-14 10:39:26 +11001331 * credentials (COW) */
Arnd Bergmann1b0ba1c2010-02-24 19:45:09 +01001332 const struct cred __rcu *cred; /* effective (overridable) subjective task
David Howells3b11a1d2008-11-14 10:39:26 +11001333 * credentials (COW) */
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -07001334 char comm[TASK_COMM_LEN]; /* executable name excluding path
1335 - access with [gs]et_task_comm (which lock
1336 it with task_lock())
Linus Torvalds221af7f2010-01-28 22:14:42 -08001337 - initialized normally by setup_new_exec */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001338/* file system info */
1339 int link_count, total_link_count;
Alexey Dobriyan3d5b6fc2006-09-29 01:59:40 -07001340#ifdef CONFIG_SYSVIPC
Linus Torvalds1da177e2005-04-16 15:20:36 -07001341/* ipc stuff */
1342 struct sysv_sem sysvsem;
Alexey Dobriyan3d5b6fc2006-09-29 01:59:40 -07001343#endif
Mandeep Singh Bainese162b392009-01-15 11:08:40 -08001344#ifdef CONFIG_DETECT_HUNG_TASK
Ingo Molnar82a1fcb2008-01-25 21:08:02 +01001345/* hung task detection */
Ingo Molnar82a1fcb2008-01-25 21:08:02 +01001346 unsigned long last_switch_count;
1347#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001348/* CPU-specific state of this task */
1349 struct thread_struct thread;
1350/* filesystem information */
1351 struct fs_struct *fs;
1352/* open file information */
1353 struct files_struct *files;
Serge E. Hallyn1651e142006-10-02 02:18:08 -07001354/* namespaces */
Serge E. Hallynab516012006-10-02 02:18:06 -07001355 struct nsproxy *nsproxy;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001356/* signal handlers */
1357 struct signal_struct *signal;
1358 struct sighand_struct *sighand;
1359
1360 sigset_t blocked, real_blocked;
Roland McGrathf3de2722008-04-30 00:53:09 -07001361 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001362 struct sigpending pending;
1363
1364 unsigned long sas_ss_sp;
1365 size_t sas_ss_size;
1366 int (*notifier)(void *priv);
1367 void *notifier_data;
1368 sigset_t *notifier_mask;
Al Viro67d12142012-06-27 11:07:19 +04001369 struct callback_head *task_works;
Oleg Nesterove73f8952012-05-11 10:59:07 +10001370
Linus Torvalds1da177e2005-04-16 15:20:36 -07001371 struct audit_context *audit_context;
Al Virobfef93a2008-01-10 04:53:18 -05001372#ifdef CONFIG_AUDITSYSCALL
Eric W. Biedermane1760bd2012-09-10 22:39:43 -07001373 kuid_t loginuid;
Eric Paris4746ec52008-01-08 10:06:53 -05001374 unsigned int sessionid;
Al Virobfef93a2008-01-10 04:53:18 -05001375#endif
Will Drewry932eceb2012-04-12 16:47:54 -05001376 struct seccomp seccomp;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001377
1378/* Thread group tracking */
1379 u32 parent_exec_id;
1380 u32 self_exec_id;
Miao Xie58568d22009-06-16 15:31:49 -07001381/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1382 * mempolicy */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001383 spinlock_t alloc_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001384
Ingo Molnarb29739f2006-06-27 02:54:51 -07001385 /* Protection of the PI data structures: */
Thomas Gleixner1d615482009-11-17 14:54:03 +01001386 raw_spinlock_t pi_lock;
Ingo Molnarb29739f2006-06-27 02:54:51 -07001387
Ingo Molnar23f78d4a2006-06-27 02:54:53 -07001388#ifdef CONFIG_RT_MUTEXES
1389 /* PI waiters blocked on a rt_mutex held by this task */
Peter Zijlstrafb00aca2013-11-07 14:43:43 +01001390 struct rb_root pi_waiters;
1391 struct rb_node *pi_waiters_leftmost;
Ingo Molnar23f78d4a2006-06-27 02:54:53 -07001392 /* Deadlock detection and priority inheritance handling */
1393 struct rt_mutex_waiter *pi_blocked_on;
Dario Faggioli2d3d8912013-11-07 14:43:44 +01001394 /* Top pi_waiters task */
1395 struct task_struct *pi_top_task;
Ingo Molnar23f78d4a2006-06-27 02:54:53 -07001396#endif
1397
Ingo Molnar408894e2006-01-09 15:59:20 -08001398#ifdef CONFIG_DEBUG_MUTEXES
1399 /* mutex deadlock detection */
1400 struct mutex_waiter *blocked_on;
1401#endif
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001402#ifdef CONFIG_TRACE_IRQFLAGS
1403 unsigned int irq_events;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001404 unsigned long hardirq_enable_ip;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001405 unsigned long hardirq_disable_ip;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001406 unsigned int hardirq_enable_event;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001407 unsigned int hardirq_disable_event;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001408 int hardirqs_enabled;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001409 int hardirq_context;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001410 unsigned long softirq_disable_ip;
1411 unsigned long softirq_enable_ip;
1412 unsigned int softirq_disable_event;
1413 unsigned int softirq_enable_event;
1414 int softirqs_enabled;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001415 int softirq_context;
1416#endif
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001417#ifdef CONFIG_LOCKDEP
Peter Zijlstrabdb94412008-02-25 23:02:48 +01001418# define MAX_LOCK_DEPTH 48UL
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001419 u64 curr_chain_key;
1420 int lockdep_depth;
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001421 unsigned int lockdep_recursion;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001422 struct held_lock held_locks[MAX_LOCK_DEPTH];
Nick Piggincf40bd12009-01-21 08:12:39 +01001423 gfp_t lockdep_reclaim_gfp;
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001424#endif
Ingo Molnar408894e2006-01-09 15:59:20 -08001425
Linus Torvalds1da177e2005-04-16 15:20:36 -07001426/* journalling filesystem info */
1427 void *journal_info;
1428
Neil Brownd89d8792007-05-01 09:53:42 +02001429/* stacked block device info */
Akinobu Mitabddd87c2010-02-23 08:55:42 +01001430 struct bio_list *bio_list;
Neil Brownd89d8792007-05-01 09:53:42 +02001431
Jens Axboe73c10102011-03-08 13:19:51 +01001432#ifdef CONFIG_BLOCK
1433/* stack plugging */
1434 struct blk_plug *plug;
1435#endif
1436
Linus Torvalds1da177e2005-04-16 15:20:36 -07001437/* VM state */
1438 struct reclaim_state *reclaim_state;
1439
Linus Torvalds1da177e2005-04-16 15:20:36 -07001440 struct backing_dev_info *backing_dev_info;
1441
1442 struct io_context *io_context;
1443
1444 unsigned long ptrace_message;
1445 siginfo_t *last_siginfo; /* For ptrace use. */
Andrew Morton7c3ab7382006-12-10 02:19:19 -08001446 struct task_io_accounting ioac;
Jay Lan8f0ab512006-09-30 23:28:59 -07001447#if defined(CONFIG_TASK_XACCT)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001448 u64 acct_rss_mem1; /* accumulated rss usage */
1449 u64 acct_vm_mem1; /* accumulated virtual memory usage */
Jonathan Lim49b5cf32008-07-25 01:48:40 -07001450 cputime_t acct_timexpd; /* stime + utime since last update */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001451#endif
1452#ifdef CONFIG_CPUSETS
Miao Xie58568d22009-06-16 15:31:49 -07001453 nodemask_t mems_allowed; /* Protected by alloc_lock */
Mel Gormancc9a6c82012-03-21 16:34:11 -07001454 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
Paul Jackson825a46a2006-03-24 03:16:03 -08001455 int cpuset_mem_spread_rotor;
Jack Steiner6adef3e2010-05-26 14:42:49 -07001456 int cpuset_slab_spread_rotor;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001457#endif
Paul Menageddbcc7e2007-10-18 23:39:30 -07001458#ifdef CONFIG_CGROUPS
Paul Menage817929e2007-10-18 23:39:36 -07001459 /* Control Group info protected by css_set_lock */
Arnd Bergmann2c392b82010-02-24 19:41:39 +01001460 struct css_set __rcu *cgroups;
Paul Menage817929e2007-10-18 23:39:36 -07001461 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1462 struct list_head cg_list;
Paul Menageddbcc7e2007-10-18 23:39:30 -07001463#endif
Alexey Dobriyan42b2dd02007-10-16 23:27:30 -07001464#ifdef CONFIG_FUTEX
Ingo Molnar0771dfe2006-03-27 01:16:22 -08001465 struct robust_list_head __user *robust_list;
Ingo Molnar34f192c2006-03-27 01:16:24 -08001466#ifdef CONFIG_COMPAT
1467 struct compat_robust_list_head __user *compat_robust_list;
1468#endif
Ingo Molnarc87e2832006-06-27 02:54:58 -07001469 struct list_head pi_state_list;
1470 struct futex_pi_state *pi_state_cache;
Alexey Dobriyan42b2dd02007-10-16 23:27:30 -07001471#endif
Ingo Molnarcdd6c482009-09-21 12:02:48 +02001472#ifdef CONFIG_PERF_EVENTS
Peter Zijlstra8dc85d5472010-09-02 16:50:03 +02001473 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
Ingo Molnarcdd6c482009-09-21 12:02:48 +02001474 struct mutex perf_event_mutex;
1475 struct list_head perf_event_list;
Paul Mackerrasa63eaf32009-05-22 14:17:31 +10001476#endif
Thomas Gleixner8f47b182014-02-07 20:58:39 +01001477#ifdef CONFIG_DEBUG_PREEMPT
1478 unsigned long preempt_disable_ip;
1479#endif
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001480#ifdef CONFIG_NUMA
Miao Xie58568d22009-06-16 15:31:49 -07001481 struct mempolicy *mempolicy; /* Protected by alloc_lock */
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001482 short il_next;
Eric Dumazet207205a2011-03-22 16:30:44 -07001483 short pref_node_fork;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001484#endif
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001485#ifdef CONFIG_NUMA_BALANCING
1486 int numa_scan_seq;
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001487 unsigned int numa_scan_period;
Mel Gorman598f0ec2013-10-07 11:28:55 +01001488 unsigned int numa_scan_period_max;
Rik van Rielde1c9ce2013-10-07 11:29:39 +01001489 int numa_preferred_nid;
Mel Gorman6b9a7462013-10-07 11:29:11 +01001490 unsigned long numa_migrate_retry;
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001491 u64 node_stamp; /* migration stamp */
Rik van Riel7e2703e2014-01-27 17:03:45 -05001492 u64 last_task_numa_placement;
1493 u64 last_sum_exec_runtime;
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001494 struct callback_head numa_work;
Mel Gormanf809ca92013-10-07 11:28:57 +01001495
Peter Zijlstra8c8a7432013-10-07 11:29:21 +01001496 struct list_head numa_entry;
1497 struct numa_group *numa_group;
1498
Mel Gorman745d6142013-10-07 11:28:59 +01001499 /*
1500 * Exponential decaying average of faults on a per-node basis.
1501 * Scheduling placement decisions are made based on the these counts.
1502 * The values remain static for the duration of a PTE scan
1503 */
Rik van Rielff1df892014-01-27 17:03:41 -05001504 unsigned long *numa_faults_memory;
Mel Gorman83e1d2c2013-10-07 11:29:27 +01001505 unsigned long total_numa_faults;
Mel Gorman745d6142013-10-07 11:28:59 +01001506
1507 /*
1508 * numa_faults_buffer records faults per node during the current
Rik van Rielff1df892014-01-27 17:03:41 -05001509 * scan window. When the scan completes, the counts in
1510 * numa_faults_memory decay and these values are copied.
Mel Gorman745d6142013-10-07 11:28:59 +01001511 */
Rik van Rielff1df892014-01-27 17:03:41 -05001512 unsigned long *numa_faults_buffer_memory;
Mel Gorman745d6142013-10-07 11:28:59 +01001513
Rik van Riel04bb2f92013-10-07 11:29:36 +01001514 /*
Rik van Riel50ec8a42014-01-27 17:03:42 -05001515 * Track the nodes the process was running on when a NUMA hinting
1516 * fault was incurred.
1517 */
1518 unsigned long *numa_faults_cpu;
1519 unsigned long *numa_faults_buffer_cpu;
1520
1521 /*
Rik van Riel04bb2f92013-10-07 11:29:36 +01001522 * numa_faults_locality tracks if faults recorded during the last
1523 * scan window were remote/local. The task scan period is adapted
1524 * based on the locality of the faults with different weights
1525 * depending on whether they were shared or private faults
1526 */
1527 unsigned long numa_faults_locality[2];
1528
Ingo Molnarb32e86b2013-10-07 11:29:30 +01001529 unsigned long numa_pages_migrated;
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001530#endif /* CONFIG_NUMA_BALANCING */
1531
Ingo Molnare56d0902006-01-08 01:01:37 -08001532 struct rcu_head rcu;
Jens Axboeb92ce552006-04-11 13:52:07 +02001533
1534 /*
1535 * cache last used pipe for splice
1536 */
1537 struct pipe_inode_info *splice_pipe;
Eric Dumazet5640f762012-09-23 23:04:42 +00001538
1539 struct page_frag task_frag;
1540
Shailabh Nagarca74e922006-07-14 00:24:36 -07001541#ifdef CONFIG_TASK_DELAY_ACCT
1542 struct task_delay_info *delays;
1543#endif
Akinobu Mitaf4f154f2006-12-08 02:39:47 -08001544#ifdef CONFIG_FAULT_INJECTION
1545 int make_it_fail;
1546#endif
Wu Fengguang9d823e82011-06-11 18:10:12 -06001547 /*
1548 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1549 * balance_dirty_pages() for some dirty throttling pause
1550 */
1551 int nr_dirtied;
1552 int nr_dirtied_pause;
Wu Fengguang83712352011-06-11 19:25:42 -06001553 unsigned long dirty_paused_when; /* start of a write-and-pause period */
Wu Fengguang9d823e82011-06-11 18:10:12 -06001554
Arjan van de Ven97455122008-01-25 21:08:34 +01001555#ifdef CONFIG_LATENCYTOP
1556 int latency_record_count;
1557 struct latency_record latency_record[LT_SAVECOUNT];
1558#endif
Arjan van de Ven69766752008-09-01 15:52:40 -07001559 /*
1560 * time slack values; these are used to round up poll() and
1561 * select() etc timeout values. These are in nanoseconds.
1562 */
1563 unsigned long timer_slack_ns;
1564 unsigned long default_timer_slack_ns;
David Millerf8d570a2008-11-06 00:37:40 -08001565
Frederic Weisbeckerfb526072008-11-25 21:07:04 +01001566#ifdef CONFIG_FUNCTION_GRAPH_TRACER
Daniel Mack3ad2f3fb2010-02-03 08:01:28 +08001567 /* Index of current stored address in ret_stack */
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001568 int curr_ret_stack;
1569 /* Stack of return addresses for return function tracing */
1570 struct ftrace_ret_stack *ret_stack;
Steven Rostedt8aef2d22009-03-24 01:10:15 -04001571 /* time stamp for last schedule */
1572 unsigned long long ftrace_timestamp;
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001573 /*
1574 * Number of functions that haven't been traced
1575 * because of depth overrun.
1576 */
1577 atomic_t trace_overrun;
Frederic Weisbecker380c4b12008-12-06 03:43:41 +01001578 /* Pause for the tracing */
1579 atomic_t tracing_graph_pause;
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001580#endif
Steven Rostedtea4e2bc2008-12-03 15:36:57 -05001581#ifdef CONFIG_TRACING
1582 /* state flags for use by tracers */
1583 unsigned long trace;
Steven Rostedtb1cff0a2011-05-25 14:27:43 -04001584 /* bitmask and counter of trace recursion */
Steven Rostedt261842b2009-04-16 21:41:52 -04001585 unsigned long trace_recursion;
1586#endif /* CONFIG_TRACING */
Andrew Mortonc255a452012-07-31 16:43:02 -07001587#ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
KAMEZAWA Hiroyuki569b8462009-12-15 16:47:03 -08001588 struct memcg_batch_info {
1589 int do_batch; /* incremented when batch uncharge started */
1590 struct mem_cgroup *memcg; /* target memcg of uncharge */
Johannes Weiner7ffd4ca2011-03-23 16:42:35 -07001591 unsigned long nr_pages; /* uncharged usage */
1592 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
KAMEZAWA Hiroyuki569b8462009-12-15 16:47:03 -08001593 } memcg_batch;
Glauber Costa0e9d92f2012-12-18 14:22:42 -08001594 unsigned int memcg_kmem_skip_account;
Johannes Weiner519e5242013-09-12 15:13:42 -07001595 struct memcg_oom_info {
Johannes Weiner49426422013-10-16 13:46:59 -07001596 struct mem_cgroup *memcg;
1597 gfp_t gfp_mask;
1598 int order;
Johannes Weiner519e5242013-09-12 15:13:42 -07001599 unsigned int may_oom:1;
1600 } memcg_oom;
KAMEZAWA Hiroyuki569b8462009-12-15 16:47:03 -08001601#endif
Srikar Dronamraju0326f5a2012-03-13 23:30:11 +05301602#ifdef CONFIG_UPROBES
1603 struct uprobe_task *utask;
Srikar Dronamraju0326f5a2012-03-13 23:30:11 +05301604#endif
Kent Overstreetcafe5632013-03-23 16:11:31 -07001605#if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1606 unsigned int sequential_io;
1607 unsigned int sequential_io_avg;
1608#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001609};
1610
Rusty Russell76e6eee2009-03-12 14:35:43 -06001611/* Future-safe accessor for struct task_struct's cpus_allowed. */
Rusty Russella4636812009-12-17 11:43:29 -06001612#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
Rusty Russell76e6eee2009-03-12 14:35:43 -06001613
Peter Zijlstra6688cc02013-10-07 11:29:24 +01001614#define TNF_MIGRATED 0x01
1615#define TNF_NO_GROUP 0x02
Rik van Rieldabe1d92013-10-07 11:29:34 +01001616#define TNF_SHARED 0x04
Rik van Riel04bb2f92013-10-07 11:29:36 +01001617#define TNF_FAULT_LOCAL 0x08
Peter Zijlstra6688cc02013-10-07 11:29:24 +01001618
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001619#ifdef CONFIG_NUMA_BALANCING
Peter Zijlstra6688cc02013-10-07 11:29:24 +01001620extern void task_numa_fault(int last_node, int node, int pages, int flags);
Mel Gormane29cf082013-10-07 11:29:22 +01001621extern pid_t task_numa_group_id(struct task_struct *p);
Mel Gorman1a687c22012-11-22 11:16:36 +00001622extern void set_numabalancing_state(bool enabled);
Rik van Riel82727012013-10-07 11:29:28 +01001623extern void task_numa_free(struct task_struct *p);
Rik van Riel10f39042014-01-27 17:03:44 -05001624extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1625 int src_nid, int dst_cpu);
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001626#else
Mel Gormanac8e8952013-10-07 11:29:03 +01001627static inline void task_numa_fault(int last_node, int node, int pages,
Peter Zijlstra6688cc02013-10-07 11:29:24 +01001628 int flags)
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001629{
1630}
Mel Gormane29cf082013-10-07 11:29:22 +01001631static inline pid_t task_numa_group_id(struct task_struct *p)
1632{
1633 return 0;
1634}
Mel Gorman1a687c22012-11-22 11:16:36 +00001635static inline void set_numabalancing_state(bool enabled)
1636{
1637}
Rik van Riel82727012013-10-07 11:29:28 +01001638static inline void task_numa_free(struct task_struct *p)
1639{
1640}
Rik van Riel10f39042014-01-27 17:03:44 -05001641static inline bool should_numa_migrate_memory(struct task_struct *p,
1642 struct page *page, int src_nid, int dst_cpu)
1643{
1644 return true;
1645}
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001646#endif
1647
Alexey Dobriyane8681712007-10-26 12:17:22 +04001648static inline struct pid *task_pid(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001649{
1650 return task->pids[PIDTYPE_PID].pid;
1651}
1652
Alexey Dobriyane8681712007-10-26 12:17:22 +04001653static inline struct pid *task_tgid(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001654{
1655 return task->group_leader->pids[PIDTYPE_PID].pid;
1656}
1657
Oleg Nesterov6dda81f2009-04-02 16:58:35 -07001658/*
1659 * Without tasklist or rcu lock it is not safe to dereference
1660 * the result of task_pgrp/task_session even if task == current,
1661 * we can race with another thread doing sys_setsid/sys_setpgid.
1662 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04001663static inline struct pid *task_pgrp(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001664{
1665 return task->group_leader->pids[PIDTYPE_PGID].pid;
1666}
1667
Alexey Dobriyane8681712007-10-26 12:17:22 +04001668static inline struct pid *task_session(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001669{
1670 return task->group_leader->pids[PIDTYPE_SID].pid;
1671}
1672
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001673struct pid_namespace;
1674
1675/*
1676 * the helpers to get the task's different pids as they are seen
1677 * from various namespaces
1678 *
1679 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
Eric W. Biederman44c4e1b2008-02-08 04:19:15 -08001680 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1681 * current.
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001682 * task_xid_nr_ns() : id seen from the ns specified;
1683 *
1684 * set_task_vxid() : assigns a virtual id to a task;
1685 *
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001686 * see also pid_nr() etc in include/linux/pid.h
1687 */
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001688pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1689 struct pid_namespace *ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001690
Alexey Dobriyane8681712007-10-26 12:17:22 +04001691static inline pid_t task_pid_nr(struct task_struct *tsk)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001692{
1693 return tsk->pid;
1694}
1695
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001696static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1697 struct pid_namespace *ns)
1698{
1699 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1700}
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001701
1702static inline pid_t task_pid_vnr(struct task_struct *tsk)
1703{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001704 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001705}
1706
1707
Alexey Dobriyane8681712007-10-26 12:17:22 +04001708static inline pid_t task_tgid_nr(struct task_struct *tsk)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001709{
1710 return tsk->tgid;
1711}
1712
Pavel Emelyanov2f2a3a42007-10-18 23:40:19 -07001713pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001714
1715static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1716{
1717 return pid_vnr(task_tgid(tsk));
1718}
1719
1720
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001721static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1722 struct pid_namespace *ns)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001723{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001724 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001725}
1726
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001727static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1728{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001729 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001730}
1731
1732
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001733static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1734 struct pid_namespace *ns)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001735{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001736 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001737}
1738
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001739static inline pid_t task_session_vnr(struct task_struct *tsk)
1740{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001741 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001742}
1743
Oleg Nesterov1b0f7ffd2009-04-02 16:58:39 -07001744/* obsolete, do not use */
1745static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1746{
1747 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1748}
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001749
Linus Torvalds1da177e2005-04-16 15:20:36 -07001750/**
1751 * pid_alive - check that a task structure is not stale
1752 * @p: Task structure to be checked.
1753 *
1754 * Test if a process is not yet dead (at most zombie state)
1755 * If pid_alive fails, then pointers within the task structure
1756 * can be stale and must not be dereferenced.
Yacine Belkadie69f6182013-07-12 20:45:47 +02001757 *
1758 * Return: 1 if the process is alive. 0 otherwise.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001759 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04001760static inline int pid_alive(struct task_struct *p)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001761{
Eric W. Biederman92476d72006-03-31 02:31:42 -08001762 return p->pids[PIDTYPE_PID].pid != NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001763}
1764
Sukadev Bhattiproluf400e192006-09-29 02:00:07 -07001765/**
Serge E. Hallynb460cbc2007-10-18 23:39:52 -07001766 * is_global_init - check if a task structure is init
Henne32602592006-10-06 00:44:01 -07001767 * @tsk: Task structure to be checked.
1768 *
1769 * Check if a task structure is the first user space task the kernel created.
Yacine Belkadie69f6182013-07-12 20:45:47 +02001770 *
1771 * Return: 1 if the task structure is init. 0 otherwise.
Sukadev Bhattiproluf400e192006-09-29 02:00:07 -07001772 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04001773static inline int is_global_init(struct task_struct *tsk)
Pavel Emelyanovb461cc02007-10-18 23:40:09 -07001774{
1775 return tsk->pid == 1;
1776}
Serge E. Hallynb460cbc2007-10-18 23:39:52 -07001777
Cedric Le Goater9ec52092006-10-02 02:19:00 -07001778extern struct pid *cad_pid;
1779
Linus Torvalds1da177e2005-04-16 15:20:36 -07001780extern void free_task(struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001781#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
Ingo Molnare56d0902006-01-08 01:01:37 -08001782
Andrew Morton158d9eb2006-03-31 02:31:34 -08001783extern void __put_task_struct(struct task_struct *t);
Ingo Molnare56d0902006-01-08 01:01:37 -08001784
1785static inline void put_task_struct(struct task_struct *t)
1786{
1787 if (atomic_dec_and_test(&t->usage))
Eric W. Biederman8c7904a2006-03-31 02:31:37 -08001788 __put_task_struct(t);
Ingo Molnare56d0902006-01-08 01:01:37 -08001789}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001790
Frederic Weisbecker6a616712012-12-16 20:00:34 +01001791#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1792extern void task_cputime(struct task_struct *t,
1793 cputime_t *utime, cputime_t *stime);
1794extern void task_cputime_scaled(struct task_struct *t,
1795 cputime_t *utimescaled, cputime_t *stimescaled);
1796extern cputime_t task_gtime(struct task_struct *t);
1797#else
Frederic Weisbecker6fac4822012-11-13 14:20:55 +01001798static inline void task_cputime(struct task_struct *t,
1799 cputime_t *utime, cputime_t *stime)
1800{
1801 if (utime)
1802 *utime = t->utime;
1803 if (stime)
1804 *stime = t->stime;
1805}
1806
1807static inline void task_cputime_scaled(struct task_struct *t,
1808 cputime_t *utimescaled,
1809 cputime_t *stimescaled)
1810{
1811 if (utimescaled)
1812 *utimescaled = t->utimescaled;
1813 if (stimescaled)
1814 *stimescaled = t->stimescaled;
1815}
Frederic Weisbecker6a616712012-12-16 20:00:34 +01001816
1817static inline cputime_t task_gtime(struct task_struct *t)
1818{
1819 return t->gtime;
1820}
1821#endif
Frederic Weisbeckere80d0a1a2012-11-21 16:26:44 +01001822extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1823extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
Balbir Singh49048622008-09-05 18:12:23 +02001824
Linus Torvalds1da177e2005-04-16 15:20:36 -07001825/*
1826 * Per process flags
1827 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001828#define PF_EXITING 0x00000004 /* getting shut down */
Alexey Kuznetsov778e9a92007-06-08 13:47:00 -07001829#define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
Laurent Vivier94886b82007-10-15 17:00:19 +02001830#define PF_VCPU 0x00000010 /* I'm a virtual CPU */
Tejun Heo21aa9af2010-06-08 21:40:37 +02001831#define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001832#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
Andi Kleen4db96cf2009-09-16 11:50:14 +02001833#define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001834#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1835#define PF_DUMPCORE 0x00000200 /* dumped core */
1836#define PF_SIGNALED 0x00000400 /* killed by a signal */
1837#define PF_MEMALLOC 0x00000800 /* Allocating memory */
Vasiliy Kulikov72fa5992011-08-08 19:02:04 +04001838#define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001839#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
Tejun Heo774a1222013-01-15 18:52:51 -08001840#define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001841#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1842#define PF_FROZEN 0x00010000 /* frozen for system suspend */
1843#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1844#define PF_KSWAPD 0x00040000 /* I am kswapd */
Ming Lei21caf2f2013-02-22 16:34:08 -08001845#define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001846#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
Oleg Nesterov246bb0b2008-07-25 01:47:38 -07001847#define PF_KTHREAD 0x00200000 /* I am a kernel thread */
Jens Axboeb31dc662006-06-13 08:26:10 +02001848#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1849#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1850#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1851#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
Tejun Heo14a40ff2013-03-19 13:45:20 -07001852#define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
Andi Kleen4db96cf2009-09-16 11:50:14 +02001853#define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
Thomas Gleixner61a87122006-06-27 02:54:56 -07001854#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
Tejun Heo58a69cb2011-02-16 09:25:31 +01001855#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
Colin Cross2b44c4d2013-07-24 17:41:33 -07001856#define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001857
1858/*
1859 * Only the _current_ task can read/write to tsk->flags, but other
1860 * tasks can access tsk->flags in readonly mode for example
1861 * with tsk_used_math (like during threaded core dumping).
1862 * There is however an exception to this rule during ptrace
1863 * or during fork: the ptracer task is allowed to write to the
1864 * child->flags of its traced child (same goes for fork, the parent
1865 * can write to the child->flags), because we're guaranteed the
1866 * child is not running and in turn not changing child->flags
1867 * at the same time the parent does it.
1868 */
1869#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1870#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1871#define clear_used_math() clear_stopped_child_used_math(current)
1872#define set_used_math() set_stopped_child_used_math(current)
1873#define conditional_stopped_child_used_math(condition, child) \
1874 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1875#define conditional_used_math(condition) \
1876 conditional_stopped_child_used_math(condition, current)
1877#define copy_to_stopped_child_used_math(child) \
1878 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1879/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1880#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1881#define used_math() tsk_used_math(current)
1882
Ming Lei21caf2f2013-02-22 16:34:08 -08001883/* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1884static inline gfp_t memalloc_noio_flags(gfp_t flags)
1885{
1886 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1887 flags &= ~__GFP_IO;
1888 return flags;
1889}
1890
1891static inline unsigned int memalloc_noio_save(void)
1892{
1893 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1894 current->flags |= PF_MEMALLOC_NOIO;
1895 return flags;
1896}
1897
1898static inline void memalloc_noio_restore(unsigned int flags)
1899{
1900 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1901}
1902
Tejun Heoe5c1902e2011-03-23 10:37:00 +01001903/*
Tejun Heoa8f072c2011-06-02 11:13:59 +02001904 * task->jobctl flags
Tejun Heoe5c1902e2011-03-23 10:37:00 +01001905 */
Tejun Heoa8f072c2011-06-02 11:13:59 +02001906#define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
Tejun Heoe5c1902e2011-03-23 10:37:00 +01001907
Tejun Heoa8f072c2011-06-02 11:13:59 +02001908#define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1909#define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1910#define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
Tejun Heo73ddff22011-06-14 11:20:14 +02001911#define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
Tejun Heofb1d9102011-06-14 11:20:17 +02001912#define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
Tejun Heoa8f072c2011-06-02 11:13:59 +02001913#define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
Tejun Heo544b2c92011-06-14 11:20:18 +02001914#define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
Tejun Heoa8f072c2011-06-02 11:13:59 +02001915
1916#define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1917#define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1918#define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
Tejun Heo73ddff22011-06-14 11:20:14 +02001919#define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
Tejun Heofb1d9102011-06-14 11:20:17 +02001920#define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
Tejun Heoa8f072c2011-06-02 11:13:59 +02001921#define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
Tejun Heo544b2c92011-06-14 11:20:18 +02001922#define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
Tejun Heoa8f072c2011-06-02 11:13:59 +02001923
Tejun Heofb1d9102011-06-14 11:20:17 +02001924#define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
Tejun Heo73ddff22011-06-14 11:20:14 +02001925#define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
Tejun Heo3759a0d2011-06-02 11:14:00 +02001926
Tejun Heo7dd3db52011-06-02 11:14:00 +02001927extern bool task_set_jobctl_pending(struct task_struct *task,
1928 unsigned int mask);
Tejun Heo73ddff22011-06-14 11:20:14 +02001929extern void task_clear_jobctl_trapping(struct task_struct *task);
Tejun Heo3759a0d2011-06-02 11:14:00 +02001930extern void task_clear_jobctl_pending(struct task_struct *task,
1931 unsigned int mask);
Tejun Heo39efa3e2011-03-23 10:37:00 +01001932
Paul E. McKenneya57eb942010-06-29 16:49:16 -07001933#ifdef CONFIG_PREEMPT_RCU
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001934
1935#define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
Paul E. McKenney1aa03f12012-01-11 17:25:17 -08001936#define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001937
1938static inline void rcu_copy_process(struct task_struct *p)
1939{
1940 p->rcu_read_lock_nesting = 0;
1941 p->rcu_read_unlock_special = 0;
Paul E. McKenneya57eb942010-06-29 16:49:16 -07001942#ifdef CONFIG_TREE_PREEMPT_RCU
Paul E. McKenneydd5d19b2009-08-27 14:58:16 -07001943 p->rcu_blocked_node = NULL;
Paul E. McKenney24278d12010-09-27 17:25:23 -07001944#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1945#ifdef CONFIG_RCU_BOOST
1946 p->rcu_boost_mutex = NULL;
1947#endif /* #ifdef CONFIG_RCU_BOOST */
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001948 INIT_LIST_HEAD(&p->rcu_node_entry);
1949}
1950
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001951#else
1952
1953static inline void rcu_copy_process(struct task_struct *p)
1954{
1955}
1956
1957#endif
1958
Mel Gorman907aed42012-07-31 16:44:07 -07001959static inline void tsk_restore_flags(struct task_struct *task,
1960 unsigned long orig_flags, unsigned long flags)
1961{
1962 task->flags &= ~flags;
1963 task->flags |= orig_flags & flags;
1964}
1965
Linus Torvalds1da177e2005-04-16 15:20:36 -07001966#ifdef CONFIG_SMP
KOSAKI Motohiro1e1b6c52011-05-19 15:08:58 +09001967extern void do_set_cpus_allowed(struct task_struct *p,
1968 const struct cpumask *new_mask);
1969
Mike Traviscd8ba7c2008-03-26 14:23:49 -07001970extern int set_cpus_allowed_ptr(struct task_struct *p,
Rusty Russell96f874e22008-11-25 02:35:14 +10301971 const struct cpumask *new_mask);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001972#else
KOSAKI Motohiro1e1b6c52011-05-19 15:08:58 +09001973static inline void do_set_cpus_allowed(struct task_struct *p,
1974 const struct cpumask *new_mask)
1975{
1976}
Mike Traviscd8ba7c2008-03-26 14:23:49 -07001977static inline int set_cpus_allowed_ptr(struct task_struct *p,
Rusty Russell96f874e22008-11-25 02:35:14 +10301978 const struct cpumask *new_mask)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001979{
Rusty Russell96f874e22008-11-25 02:35:14 +10301980 if (!cpumask_test_cpu(0, new_mask))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001981 return -EINVAL;
1982 return 0;
1983}
1984#endif
Rusty Russelle0ad9552009-09-24 09:34:38 -06001985
Frederic Weisbecker3451d022011-08-10 23:21:01 +02001986#ifdef CONFIG_NO_HZ_COMMON
Peter Zijlstra5167e8d2012-06-22 15:52:09 +02001987void calc_load_enter_idle(void);
1988void calc_load_exit_idle(void);
1989#else
1990static inline void calc_load_enter_idle(void) { }
1991static inline void calc_load_exit_idle(void) { }
Frederic Weisbecker3451d022011-08-10 23:21:01 +02001992#endif /* CONFIG_NO_HZ_COMMON */
Peter Zijlstra5167e8d2012-06-22 15:52:09 +02001993
Rusty Russelle0ad9552009-09-24 09:34:38 -06001994#ifndef CONFIG_CPUMASK_OFFSTACK
Mike Traviscd8ba7c2008-03-26 14:23:49 -07001995static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1996{
1997 return set_cpus_allowed_ptr(p, &new_mask);
1998}
Rusty Russelle0ad9552009-09-24 09:34:38 -06001999#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002000
Ingo Molnarb3425012009-02-26 20:20:29 +01002001/*
Peter Zijlstrac6763292010-05-25 10:48:51 +02002002 * Do not use outside of architecture code which knows its limitations.
2003 *
2004 * sched_clock() has no promise of monotonicity or bounded drift between
2005 * CPUs, use (which you should not) requires disabling IRQs.
2006 *
2007 * Please use one of the three interfaces below.
Ingo Molnarb3425012009-02-26 20:20:29 +01002008 */
Mike Frysinger1bbfa6f2009-12-09 20:07:03 -05002009extern unsigned long long notrace sched_clock(void);
Peter Zijlstrac6763292010-05-25 10:48:51 +02002010/*
Hiroshi Shimamoto489a71b2012-04-02 17:00:44 +09002011 * See the comment in kernel/sched/clock.c
Peter Zijlstrac6763292010-05-25 10:48:51 +02002012 */
2013extern u64 cpu_clock(int cpu);
2014extern u64 local_clock(void);
2015extern u64 sched_clock_cpu(int cpu);
2016
Ingo Molnare436d802007-07-19 21:28:35 +02002017
Peter Zijlstrac1955a32008-08-11 08:59:03 +02002018extern void sched_clock_init(void);
Peter Zijlstrac1955a32008-08-11 08:59:03 +02002019
Peter Zijlstra3e51f332008-05-03 18:29:28 +02002020#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
Peter Zijlstra3e51f332008-05-03 18:29:28 +02002021static inline void sched_clock_tick(void)
2022{
2023}
2024
2025static inline void sched_clock_idle_sleep_event(void)
2026{
2027}
2028
2029static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2030{
2031}
2032#else
Peter Zijlstrac6763292010-05-25 10:48:51 +02002033/*
2034 * Architectures can set this to 1 if they have specified
2035 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2036 * but then during bootup it turns out that sched_clock()
2037 * is reliable after all:
2038 */
Peter Zijlstra35af99e2013-11-28 19:38:42 +01002039extern int sched_clock_stable(void);
2040extern void set_sched_clock_stable(void);
2041extern void clear_sched_clock_stable(void);
Peter Zijlstrac6763292010-05-25 10:48:51 +02002042
Peter Zijlstra3e51f332008-05-03 18:29:28 +02002043extern void sched_clock_tick(void);
2044extern void sched_clock_idle_sleep_event(void);
2045extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2046#endif
2047
Venkatesh Pallipadib52bfee2010-10-04 17:03:19 -07002048#ifdef CONFIG_IRQ_TIME_ACCOUNTING
2049/*
2050 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2051 * The reason for this explicit opt-in is not to have perf penalty with
2052 * slow sched_clocks.
2053 */
2054extern void enable_sched_clock_irqtime(void);
2055extern void disable_sched_clock_irqtime(void);
2056#else
2057static inline void enable_sched_clock_irqtime(void) {}
2058static inline void disable_sched_clock_irqtime(void) {}
2059#endif
2060
Ingo Molnar36c8b582006-07-03 00:25:41 -07002061extern unsigned long long
Ingo Molnar41b86e92007-07-09 18:51:58 +02002062task_sched_runtime(struct task_struct *task);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002063
2064/* sched_exec is called by processes performing an exec */
2065#ifdef CONFIG_SMP
2066extern void sched_exec(void);
2067#else
2068#define sched_exec() {}
2069#endif
2070
Ingo Molnar2aa44d02007-08-23 15:18:02 +02002071extern void sched_clock_idle_sleep_event(void);
2072extern void sched_clock_idle_wakeup_event(u64 delta_ns);
Ingo Molnarbb29ab22007-07-09 18:51:59 +02002073
Linus Torvalds1da177e2005-04-16 15:20:36 -07002074#ifdef CONFIG_HOTPLUG_CPU
2075extern void idle_task_exit(void);
2076#else
2077static inline void idle_task_exit(void) {}
2078#endif
2079
Frederic Weisbecker3451d022011-08-10 23:21:01 +02002080#if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
Frederic Weisbecker1c200912011-08-10 23:21:01 +02002081extern void wake_up_nohz_cpu(int cpu);
Thomas Gleixner06d83082008-03-22 09:20:24 +01002082#else
Frederic Weisbecker1c200912011-08-10 23:21:01 +02002083static inline void wake_up_nohz_cpu(int cpu) { }
Thomas Gleixner06d83082008-03-22 09:20:24 +01002084#endif
2085
Frederic Weisbeckerce831b32013-04-20 15:15:35 +02002086#ifdef CONFIG_NO_HZ_FULL
2087extern bool sched_can_stop_tick(void);
Frederic Weisbecker265f22a2013-05-03 03:39:05 +02002088extern u64 scheduler_tick_max_deferment(void);
Frederic Weisbeckerce831b32013-04-20 15:15:35 +02002089#else
2090static inline bool sched_can_stop_tick(void) { return false; }
Ingo Molnarbf0f6f22007-07-09 18:51:58 +02002091#endif
Ingo Molnarbf0f6f22007-07-09 18:51:58 +02002092
Mike Galbraith5091faa2010-11-30 14:18:03 +01002093#ifdef CONFIG_SCHED_AUTOGROUP
Mike Galbraith5091faa2010-11-30 14:18:03 +01002094extern void sched_autogroup_create_attach(struct task_struct *p);
2095extern void sched_autogroup_detach(struct task_struct *p);
2096extern void sched_autogroup_fork(struct signal_struct *sig);
2097extern void sched_autogroup_exit(struct signal_struct *sig);
2098#ifdef CONFIG_PROC_FS
2099extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
Hiroshi Shimamoto2e5b5b32012-02-23 17:41:27 +09002100extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
Mike Galbraith5091faa2010-11-30 14:18:03 +01002101#endif
2102#else
2103static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2104static inline void sched_autogroup_detach(struct task_struct *p) { }
2105static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2106static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2107#endif
2108
Mike Galbraithd95f4122011-02-01 09:50:51 -05002109extern bool yield_to(struct task_struct *p, bool preempt);
Ingo Molnar36c8b582006-07-03 00:25:41 -07002110extern void set_user_nice(struct task_struct *p, long nice);
2111extern int task_prio(const struct task_struct *p);
Dongsheng Yangd0ea0262014-01-27 22:00:45 -05002112/**
2113 * task_nice - return the nice value of a given task.
2114 * @p: the task in question.
2115 *
2116 * Return: The nice value [ -20 ... 0 ... 19 ].
2117 */
2118static inline int task_nice(const struct task_struct *p)
2119{
2120 return PRIO_TO_NICE((p)->static_prio);
2121}
Ingo Molnar36c8b582006-07-03 00:25:41 -07002122extern int can_nice(const struct task_struct *p, const int nice);
2123extern int task_curr(const struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002124extern int idle_cpu(int cpu);
KOSAKI Motohirofe7de492010-10-20 16:01:12 -07002125extern int sched_setscheduler(struct task_struct *, int,
2126 const struct sched_param *);
Rusty Russell961ccdd2008-06-23 13:55:38 +10002127extern int sched_setscheduler_nocheck(struct task_struct *, int,
KOSAKI Motohirofe7de492010-10-20 16:01:12 -07002128 const struct sched_param *);
Dario Faggiolid50dde52013-11-07 14:43:36 +01002129extern int sched_setattr(struct task_struct *,
2130 const struct sched_attr *);
Ingo Molnar36c8b582006-07-03 00:25:41 -07002131extern struct task_struct *idle_task(int cpu);
Paul E. McKenneyc4f30602011-11-10 12:41:56 -08002132/**
2133 * is_idle_task - is the specified task an idle task?
Randy Dunlapfa757282012-01-21 11:03:13 -08002134 * @p: the task in question.
Yacine Belkadie69f6182013-07-12 20:45:47 +02002135 *
2136 * Return: 1 if @p is an idle task. 0 otherwise.
Paul E. McKenneyc4f30602011-11-10 12:41:56 -08002137 */
Paul E. McKenney7061ca32011-12-20 08:20:46 -08002138static inline bool is_idle_task(const struct task_struct *p)
Paul E. McKenneyc4f30602011-11-10 12:41:56 -08002139{
2140 return p->pid == 0;
2141}
Ingo Molnar36c8b582006-07-03 00:25:41 -07002142extern struct task_struct *curr_task(int cpu);
2143extern void set_curr_task(int cpu, struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002144
2145void yield(void);
2146
2147/*
2148 * The default (Linux) execution domain.
2149 */
2150extern struct exec_domain default_exec_domain;
2151
2152union thread_union {
2153 struct thread_info thread_info;
2154 unsigned long stack[THREAD_SIZE/sizeof(long)];
2155};
2156
2157#ifndef __HAVE_ARCH_KSTACK_END
2158static inline int kstack_end(void *addr)
2159{
2160 /* Reliable end of stack detection:
2161 * Some APM bios versions misalign the stack
2162 */
2163 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2164}
2165#endif
2166
2167extern union thread_union init_thread_union;
2168extern struct task_struct init_task;
2169
2170extern struct mm_struct init_mm;
2171
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002172extern struct pid_namespace init_pid_ns;
2173
2174/*
2175 * find a task by one of its numerical ids
2176 *
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002177 * find_task_by_pid_ns():
2178 * finds a task by its pid in the specified namespace
Pavel Emelyanov228ebcb2007-10-18 23:40:16 -07002179 * find_task_by_vpid():
2180 * finds a task by its virtual pid
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002181 *
Pavel Emelyanove49859e2008-07-25 01:48:36 -07002182 * see also find_vpid() etc in include/linux/pid.h
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002183 */
2184
Pavel Emelyanov228ebcb2007-10-18 23:40:16 -07002185extern struct task_struct *find_task_by_vpid(pid_t nr);
2186extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2187 struct pid_namespace *ns);
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002188
Linus Torvalds1da177e2005-04-16 15:20:36 -07002189/* per-UID process charging. */
Eric W. Biederman7b44ab92011-11-16 23:20:58 -08002190extern struct user_struct * alloc_uid(kuid_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002191static inline struct user_struct *get_uid(struct user_struct *u)
2192{
2193 atomic_inc(&u->__count);
2194 return u;
2195}
2196extern void free_uid(struct user_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002197
2198#include <asm/current.h>
2199
Torben Hohnf0af911a92011-01-27 15:59:10 +01002200extern void xtime_update(unsigned long ticks);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002201
Harvey Harrisonb3c97522008-02-13 15:03:15 -08002202extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2203extern int wake_up_process(struct task_struct *tsk);
Samir Bellabes3e51e3e2011-05-11 18:18:05 +02002204extern void wake_up_new_task(struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002205#ifdef CONFIG_SMP
2206 extern void kick_process(struct task_struct *tsk);
2207#else
2208 static inline void kick_process(struct task_struct *tsk) { }
2209#endif
Dario Faggioliaab03e02013-11-28 11:14:43 +01002210extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
Ingo Molnarad46c2c2007-07-09 18:52:00 +02002211extern void sched_dead(struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002212
Linus Torvalds1da177e2005-04-16 15:20:36 -07002213extern void proc_caches_init(void);
2214extern void flush_signals(struct task_struct *);
David Howells3bcac022009-04-29 13:45:05 +01002215extern void __flush_signals(struct task_struct *);
Oleg Nesterov10ab8252007-05-09 02:34:37 -07002216extern void ignore_signals(struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002217extern void flush_signal_handlers(struct task_struct *, int force_default);
2218extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2219
2220static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2221{
2222 unsigned long flags;
2223 int ret;
2224
2225 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2226 ret = dequeue_signal(tsk, mask, info);
2227 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2228
2229 return ret;
Oleg Nesterov53c8f9f2011-06-22 23:08:18 +02002230}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002231
2232extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2233 sigset_t *mask);
2234extern void unblock_all_signals(void);
2235extern void release_task(struct task_struct * p);
2236extern int send_sig_info(int, struct siginfo *, struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002237extern int force_sigsegv(int, struct task_struct *);
2238extern int force_sig_info(int, struct siginfo *, struct task_struct *);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002239extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002240extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
Serge Hallynd178bc32011-09-26 10:45:18 -05002241extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2242 const struct cred *, u32);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002243extern int kill_pgrp(struct pid *pid, int sig, int priv);
2244extern int kill_pid(struct pid *pid, int sig, int priv);
Matthew Wilcoxc3de4b32007-02-09 08:11:47 -07002245extern int kill_proc_info(int, struct siginfo *, pid_t);
Oleg Nesterov86773472011-06-22 23:09:09 +02002246extern __must_check bool do_notify_parent(struct task_struct *, int);
Oleg Nesterova7f07652009-09-23 15:56:44 -07002247extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002248extern void force_sig(int, struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002249extern int send_sig(int, struct task_struct *, int);
Oleg Nesterov09faef12010-05-26 14:43:11 -07002250extern int zap_other_threads(struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002251extern struct sigqueue *sigqueue_alloc(void);
2252extern void sigqueue_free(struct sigqueue *);
Oleg Nesterovac5c2152008-04-30 00:52:57 -07002253extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
Oleg Nesterov9ac95f22006-02-09 22:41:50 +03002254extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002255
Al Viro51a7b442012-05-21 23:33:55 -04002256static inline void restore_saved_sigmask(void)
2257{
2258 if (test_and_clear_restore_sigmask())
Al Viro77097ae2012-04-27 13:58:59 -04002259 __set_current_blocked(&current->saved_sigmask);
Al Viro51a7b442012-05-21 23:33:55 -04002260}
2261
Al Virob7f9a112012-05-02 09:59:21 -04002262static inline sigset_t *sigmask_to_save(void)
2263{
2264 sigset_t *res = &current->blocked;
2265 if (unlikely(test_restore_sigmask()))
2266 res = &current->saved_sigmask;
2267 return res;
2268}
2269
Cedric Le Goater9ec52092006-10-02 02:19:00 -07002270static inline int kill_cad_pid(int sig, int priv)
2271{
2272 return kill_pid(cad_pid, sig, priv);
2273}
2274
Linus Torvalds1da177e2005-04-16 15:20:36 -07002275/* These can be the second arg to send_sig_info/send_group_sig_info. */
2276#define SEND_SIG_NOINFO ((struct siginfo *) 0)
2277#define SEND_SIG_PRIV ((struct siginfo *) 1)
2278#define SEND_SIG_FORCED ((struct siginfo *) 2)
2279
Sebastian Andrzej Siewior2a855dd2009-10-25 15:37:58 +01002280/*
2281 * True if we are on the alternate signal stack.
2282 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002283static inline int on_sig_stack(unsigned long sp)
2284{
Sebastian Andrzej Siewior2a855dd2009-10-25 15:37:58 +01002285#ifdef CONFIG_STACK_GROWSUP
2286 return sp >= current->sas_ss_sp &&
2287 sp - current->sas_ss_sp < current->sas_ss_size;
2288#else
2289 return sp > current->sas_ss_sp &&
2290 sp - current->sas_ss_sp <= current->sas_ss_size;
2291#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002292}
2293
2294static inline int sas_ss_flags(unsigned long sp)
2295{
2296 return (current->sas_ss_size == 0 ? SS_DISABLE
2297 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2298}
2299
Al Viro5a1b98d2012-11-06 13:28:21 -05002300static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2301{
2302 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2303#ifdef CONFIG_STACK_GROWSUP
2304 return current->sas_ss_sp;
2305#else
2306 return current->sas_ss_sp + current->sas_ss_size;
2307#endif
2308 return sp;
2309}
2310
Linus Torvalds1da177e2005-04-16 15:20:36 -07002311/*
2312 * Routines for handling mm_structs
2313 */
2314extern struct mm_struct * mm_alloc(void);
2315
2316/* mmdrop drops the mm and the page tables */
Harvey Harrisonb3c97522008-02-13 15:03:15 -08002317extern void __mmdrop(struct mm_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002318static inline void mmdrop(struct mm_struct * mm)
2319{
Ingo Molnar6fb43d72007-07-09 18:52:01 +02002320 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002321 __mmdrop(mm);
2322}
2323
2324/* mmput gets rid of the mappings and all user-space */
2325extern void mmput(struct mm_struct *);
2326/* Grab a reference to a task's mm, if it is not already going away */
2327extern struct mm_struct *get_task_mm(struct task_struct *task);
Christopher Yeoh8cdb8782012-02-02 11:34:09 +10302328/*
2329 * Grab a reference to a task's mm, if it is not already going away
2330 * and ptrace_may_access with the mode parameter passed to it
2331 * succeeds.
2332 */
2333extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002334/* Remove the current tasks stale references to the old mm_struct */
2335extern void mm_release(struct task_struct *, struct mm_struct *);
2336
Alexey Dobriyan6f2c55b2009-04-02 16:56:59 -07002337extern int copy_thread(unsigned long, unsigned long, unsigned long,
Al Viroafa86fc2012-10-22 22:51:14 -04002338 struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002339extern void flush_thread(void);
2340extern void exit_thread(void);
2341
Linus Torvalds1da177e2005-04-16 15:20:36 -07002342extern void exit_files(struct task_struct *);
Oleg Nesterova7e53282006-03-28 16:11:27 -08002343extern void __cleanup_sighand(struct sighand_struct *);
Oleg Nesterovcbaffba2008-05-26 20:55:42 +04002344
Linus Torvalds1da177e2005-04-16 15:20:36 -07002345extern void exit_itimers(struct signal_struct *);
Oleg Nesterovcbaffba2008-05-26 20:55:42 +04002346extern void flush_itimer_signals(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002347
Joe Perches9402c952012-01-12 17:17:17 -08002348extern void do_group_exit(int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002349
Linus Torvalds1da177e2005-04-16 15:20:36 -07002350extern int allow_signal(int);
2351extern int disallow_signal(int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002352
Linus Torvaldsc4ad8f92014-02-05 12:54:53 -08002353extern int do_execve(struct filename *,
David Howellsd7627462010-08-17 23:52:56 +01002354 const char __user * const __user *,
Al Viroda3d4c52012-10-20 21:49:33 -04002355 const char __user * const __user *);
Al Viroe80d6662012-10-22 23:10:08 -04002356extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
Ingo Molnar36c8b582006-07-03 00:25:41 -07002357struct task_struct *fork_idle(int);
Al Viro2aa3a7f2012-09-21 19:55:31 -04002358extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002359
Oleg Nesterov23aebe12014-04-07 15:38:39 -07002360extern void set_task_comm(struct task_struct *tsk, const char *from);
Andrew Morton59714d62008-02-04 22:27:21 -08002361extern char *get_task_comm(char *to, struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002362
2363#ifdef CONFIG_SMP
Peter Zijlstra317f3942011-04-05 17:23:58 +02002364void scheduler_ipi(void);
Roland McGrath85ba2d82008-07-25 19:45:58 -07002365extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002366#else
Peter Zijlstra184748c2011-04-05 17:23:39 +02002367static inline void scheduler_ipi(void) { }
Roland McGrath85ba2d82008-07-25 19:45:58 -07002368static inline unsigned long wait_task_inactive(struct task_struct *p,
2369 long match_state)
2370{
2371 return 1;
2372}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002373#endif
2374
Jiri Pirko05725f72009-04-14 20:17:16 +02002375#define next_task(p) \
2376 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002377
2378#define for_each_process(p) \
2379 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2380
Oleg Nesterov5bb459b2009-07-10 03:48:23 +02002381extern bool current_is_single_threaded(void);
David Howellsd84f4f92008-11-14 10:39:23 +11002382
Linus Torvalds1da177e2005-04-16 15:20:36 -07002383/*
2384 * Careful: do_each_thread/while_each_thread is a double loop so
2385 * 'break' will not work as expected - use goto instead.
2386 */
2387#define do_each_thread(g, t) \
2388 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2389
2390#define while_each_thread(g, t) \
2391 while ((t = next_thread(t)) != g)
2392
Oleg Nesterov0c740d02014-01-21 15:49:56 -08002393#define __for_each_thread(signal, t) \
2394 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2395
2396#define for_each_thread(p, t) \
2397 __for_each_thread((p)->signal, t)
2398
2399/* Careful: this is a double loop, 'break' won't work as expected. */
2400#define for_each_process_thread(p, t) \
2401 for_each_process(p) for_each_thread(p, t)
2402
Oleg Nesterov7e498272010-05-26 14:43:22 -07002403static inline int get_nr_threads(struct task_struct *tsk)
2404{
Oleg Nesterovb3ac0222010-05-26 14:43:24 -07002405 return tsk->signal->nr_threads;
Oleg Nesterov7e498272010-05-26 14:43:22 -07002406}
2407
Oleg Nesterov087806b2011-06-22 23:10:26 +02002408static inline bool thread_group_leader(struct task_struct *p)
2409{
2410 return p->exit_signal >= 0;
2411}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002412
Eric W. Biederman0804ef42006-10-02 02:17:04 -07002413/* Do to the insanities of de_thread it is possible for a process
2414 * to have the pid of the thread group leader without actually being
2415 * the thread group leader. For iteration through the pids in proc
2416 * all we care about is that we have a task with the appropriate
2417 * pid, we don't actually care if we have the right task.
2418 */
Oleg Nesterove1403b82013-09-11 14:20:06 -07002419static inline bool has_group_leader_pid(struct task_struct *p)
Eric W. Biederman0804ef42006-10-02 02:17:04 -07002420{
Oleg Nesterove1403b82013-09-11 14:20:06 -07002421 return task_pid(p) == p->signal->leader_pid;
Eric W. Biederman0804ef42006-10-02 02:17:04 -07002422}
2423
Pavel Emelyanovbac0abd2007-10-18 23:40:18 -07002424static inline
Oleg Nesterove1403b82013-09-11 14:20:06 -07002425bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
Pavel Emelyanovbac0abd2007-10-18 23:40:18 -07002426{
Oleg Nesterove1403b82013-09-11 14:20:06 -07002427 return p1->signal == p2->signal;
Pavel Emelyanovbac0abd2007-10-18 23:40:18 -07002428}
2429
Ingo Molnar36c8b582006-07-03 00:25:41 -07002430static inline struct task_struct *next_thread(const struct task_struct *p)
Oleg Nesterov47e65322006-03-28 16:11:25 -08002431{
Jiri Pirko05725f72009-04-14 20:17:16 +02002432 return list_entry_rcu(p->thread_group.next,
2433 struct task_struct, thread_group);
Oleg Nesterov47e65322006-03-28 16:11:25 -08002434}
2435
Alexey Dobriyane8681712007-10-26 12:17:22 +04002436static inline int thread_group_empty(struct task_struct *p)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002437{
Oleg Nesterov47e65322006-03-28 16:11:25 -08002438 return list_empty(&p->thread_group);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002439}
2440
2441#define delay_group_leader(p) \
2442 (thread_group_leader(p) && !thread_group_empty(p))
2443
Linus Torvalds1da177e2005-04-16 15:20:36 -07002444/*
Eric W. Biederman260ea102006-06-23 02:05:18 -07002445 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
Jens Axboe22e2c502005-06-27 10:55:12 +02002446 * subscriptions and synchronises with wait4(). Also used in procfs. Also
Paul Menageddbcc7e2007-10-18 23:39:30 -07002447 * pins the final release of task.io_context. Also protects ->cpuset and
Oleg Nesterovd68b46f2012-03-05 14:59:13 -08002448 * ->cgroup.subsys[]. And ->vfork_done.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002449 *
2450 * Nests both inside and outside of read_lock(&tasklist_lock).
2451 * It must not be nested with write_lock_irq(&tasklist_lock),
2452 * neither inside nor outside.
2453 */
2454static inline void task_lock(struct task_struct *p)
2455{
2456 spin_lock(&p->alloc_lock);
2457}
2458
2459static inline void task_unlock(struct task_struct *p)
2460{
2461 spin_unlock(&p->alloc_lock);
2462}
2463
Namhyung Kimb8ed3742010-10-27 15:34:06 -07002464extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
Oleg Nesterovf63ee722006-03-28 16:11:13 -08002465 unsigned long *flags);
2466
Anton Vorontsov9388dc32012-02-09 20:45:19 +04002467static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2468 unsigned long *flags)
2469{
2470 struct sighand_struct *ret;
2471
2472 ret = __lock_task_sighand(tsk, flags);
2473 (void)__cond_lock(&tsk->sighand->siglock, ret);
2474 return ret;
2475}
Namhyung Kimb8ed3742010-10-27 15:34:06 -07002476
Oleg Nesterovf63ee722006-03-28 16:11:13 -08002477static inline void unlock_task_sighand(struct task_struct *tsk,
2478 unsigned long *flags)
2479{
2480 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2481}
2482
Ben Blum4714d1d2011-05-26 16:25:18 -07002483#ifdef CONFIG_CGROUPS
Tejun Heo257058a2011-12-12 18:12:21 -08002484static inline void threadgroup_change_begin(struct task_struct *tsk)
Ben Blum4714d1d2011-05-26 16:25:18 -07002485{
Tejun Heo257058a2011-12-12 18:12:21 -08002486 down_read(&tsk->signal->group_rwsem);
Ben Blum4714d1d2011-05-26 16:25:18 -07002487}
Tejun Heo257058a2011-12-12 18:12:21 -08002488static inline void threadgroup_change_end(struct task_struct *tsk)
Ben Blum4714d1d2011-05-26 16:25:18 -07002489{
Tejun Heo257058a2011-12-12 18:12:21 -08002490 up_read(&tsk->signal->group_rwsem);
Ben Blum4714d1d2011-05-26 16:25:18 -07002491}
Tejun Heo77e4ef92011-12-12 18:12:21 -08002492
2493/**
2494 * threadgroup_lock - lock threadgroup
2495 * @tsk: member task of the threadgroup to lock
2496 *
2497 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2498 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
Oleg Nesterove56fb282013-04-30 15:28:20 -07002499 * change ->group_leader/pid. This is useful for cases where the threadgroup
2500 * needs to stay stable across blockable operations.
Tejun Heo77e4ef92011-12-12 18:12:21 -08002501 *
2502 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2503 * synchronization. While held, no new task will be added to threadgroup
2504 * and no existing live task will have its PF_EXITING set.
2505 *
Oleg Nesterove56fb282013-04-30 15:28:20 -07002506 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2507 * sub-thread becomes a new leader.
Tejun Heo77e4ef92011-12-12 18:12:21 -08002508 */
Tejun Heo257058a2011-12-12 18:12:21 -08002509static inline void threadgroup_lock(struct task_struct *tsk)
Ben Blum4714d1d2011-05-26 16:25:18 -07002510{
Tejun Heo257058a2011-12-12 18:12:21 -08002511 down_write(&tsk->signal->group_rwsem);
Ben Blum4714d1d2011-05-26 16:25:18 -07002512}
Tejun Heo77e4ef92011-12-12 18:12:21 -08002513
2514/**
2515 * threadgroup_unlock - unlock threadgroup
2516 * @tsk: member task of the threadgroup to unlock
2517 *
2518 * Reverse threadgroup_lock().
2519 */
Tejun Heo257058a2011-12-12 18:12:21 -08002520static inline void threadgroup_unlock(struct task_struct *tsk)
Ben Blum4714d1d2011-05-26 16:25:18 -07002521{
Tejun Heo257058a2011-12-12 18:12:21 -08002522 up_write(&tsk->signal->group_rwsem);
Ben Blum4714d1d2011-05-26 16:25:18 -07002523}
2524#else
Tejun Heo257058a2011-12-12 18:12:21 -08002525static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2526static inline void threadgroup_change_end(struct task_struct *tsk) {}
2527static inline void threadgroup_lock(struct task_struct *tsk) {}
2528static inline void threadgroup_unlock(struct task_struct *tsk) {}
Ben Blum4714d1d2011-05-26 16:25:18 -07002529#endif
2530
Al Virof0373602005-11-13 16:06:57 -08002531#ifndef __HAVE_THREAD_FUNCTIONS
2532
Roman Zippelf7e42172007-05-09 02:35:17 -07002533#define task_thread_info(task) ((struct thread_info *)(task)->stack)
2534#define task_stack_page(task) ((task)->stack)
Al Viroa1261f542005-11-13 16:06:55 -08002535
Al Viro10ebffd2005-11-13 16:06:56 -08002536static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2537{
2538 *task_thread_info(p) = *task_thread_info(org);
2539 task_thread_info(p)->task = p;
2540}
2541
2542static inline unsigned long *end_of_stack(struct task_struct *p)
2543{
Roman Zippelf7e42172007-05-09 02:35:17 -07002544 return (unsigned long *)(task_thread_info(p) + 1);
Al Viro10ebffd2005-11-13 16:06:56 -08002545}
2546
Al Virof0373602005-11-13 16:06:57 -08002547#endif
2548
FUJITA Tomonori8b05c7e2008-07-23 21:26:53 -07002549static inline int object_is_on_stack(void *obj)
2550{
2551 void *stack = task_stack_page(current);
2552
2553 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2554}
2555
Benjamin Herrenschmidt8c9843e2008-04-18 16:56:15 +10002556extern void thread_info_cache_init(void);
2557
Eric Sandeen7c9f8862008-04-22 16:38:23 -05002558#ifdef CONFIG_DEBUG_STACK_USAGE
2559static inline unsigned long stack_not_used(struct task_struct *p)
2560{
2561 unsigned long *n = end_of_stack(p);
2562
2563 do { /* Skip over canary */
2564 n++;
2565 } while (!*n);
2566
2567 return (unsigned long)n - (unsigned long)end_of_stack(p);
2568}
2569#endif
2570
Linus Torvalds1da177e2005-04-16 15:20:36 -07002571/* set thread flags in other task's structures
2572 * - see asm/thread_info.h for TIF_xxxx flags available
2573 */
2574static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2575{
Al Viroa1261f542005-11-13 16:06:55 -08002576 set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002577}
2578
2579static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2580{
Al Viroa1261f542005-11-13 16:06:55 -08002581 clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002582}
2583
2584static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2585{
Al Viroa1261f542005-11-13 16:06:55 -08002586 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002587}
2588
2589static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2590{
Al Viroa1261f542005-11-13 16:06:55 -08002591 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002592}
2593
2594static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2595{
Al Viroa1261f542005-11-13 16:06:55 -08002596 return test_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002597}
2598
2599static inline void set_tsk_need_resched(struct task_struct *tsk)
2600{
2601 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2602}
2603
2604static inline void clear_tsk_need_resched(struct task_struct *tsk)
2605{
2606 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2607}
2608
Gregory Haskins8ae121a2008-04-23 07:13:29 -04002609static inline int test_tsk_need_resched(struct task_struct *tsk)
2610{
2611 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2612}
2613
Eric W. Biederman690cc3f2009-05-13 16:55:10 +00002614static inline int restart_syscall(void)
2615{
2616 set_tsk_thread_flag(current, TIF_SIGPENDING);
2617 return -ERESTARTNOINTR;
2618}
2619
Linus Torvalds1da177e2005-04-16 15:20:36 -07002620static inline int signal_pending(struct task_struct *p)
2621{
2622 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2623}
Matthew Wilcoxf776d122007-12-06 11:15:50 -05002624
Roland McGrathd9588722009-09-23 15:57:04 -07002625static inline int __fatal_signal_pending(struct task_struct *p)
2626{
2627 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2628}
Matthew Wilcoxf776d122007-12-06 11:15:50 -05002629
2630static inline int fatal_signal_pending(struct task_struct *p)
2631{
2632 return signal_pending(p) && __fatal_signal_pending(p);
2633}
2634
Oleg Nesterov16882c12008-06-08 21:20:41 +04002635static inline int signal_pending_state(long state, struct task_struct *p)
2636{
2637 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2638 return 0;
2639 if (!signal_pending(p))
2640 return 0;
2641
Oleg Nesterov16882c12008-06-08 21:20:41 +04002642 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2643}
2644
Linus Torvalds1da177e2005-04-16 15:20:36 -07002645/*
2646 * cond_resched() and cond_resched_lock(): latency reduction via
2647 * explicit rescheduling in places that are safe. The return
2648 * value indicates whether a reschedule was done in fact.
2649 * cond_resched_lock() will drop the spinlock before scheduling,
2650 * cond_resched_softirq() will enable bhs before scheduling.
2651 */
Linus Torvaldsc3921ab2008-05-11 16:04:48 -07002652extern int _cond_resched(void);
Frederic Weisbecker6f80bd92009-07-16 15:44:29 +02002653
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002654#define cond_resched() ({ \
2655 __might_sleep(__FILE__, __LINE__, 0); \
2656 _cond_resched(); \
2657})
Frederic Weisbecker6f80bd92009-07-16 15:44:29 +02002658
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002659extern int __cond_resched_lock(spinlock_t *lock);
2660
Frederic Weisbeckerbdd4e852011-06-08 01:13:27 +02002661#ifdef CONFIG_PREEMPT_COUNT
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002662#define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
Herbert Xu02b67cc32008-01-25 21:08:28 +01002663#else
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002664#define PREEMPT_LOCK_OFFSET 0
Herbert Xu02b67cc32008-01-25 21:08:28 +01002665#endif
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002666
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002667#define cond_resched_lock(lock) ({ \
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002668 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002669 __cond_resched_lock(lock); \
2670})
2671
2672extern int __cond_resched_softirq(void);
2673
Venkatesh Pallipadi75e10562010-10-04 17:03:16 -07002674#define cond_resched_softirq() ({ \
2675 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2676 __cond_resched_softirq(); \
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002677})
Linus Torvalds1da177e2005-04-16 15:20:36 -07002678
Simon Hormanf6f3c432013-05-22 14:50:31 +09002679static inline void cond_resched_rcu(void)
2680{
2681#if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2682 rcu_read_unlock();
2683 cond_resched();
2684 rcu_read_lock();
2685#endif
2686}
2687
Linus Torvalds1da177e2005-04-16 15:20:36 -07002688/*
2689 * Does a critical section need to be broken due to another
Nick Piggin95c354f2008-01-30 13:31:20 +01002690 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2691 * but a general need for low latency)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002692 */
Nick Piggin95c354f2008-01-30 13:31:20 +01002693static inline int spin_needbreak(spinlock_t *lock)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002694{
Nick Piggin95c354f2008-01-30 13:31:20 +01002695#ifdef CONFIG_PREEMPT
2696 return spin_is_contended(lock);
2697#else
Linus Torvalds1da177e2005-04-16 15:20:36 -07002698 return 0;
Nick Piggin95c354f2008-01-30 13:31:20 +01002699#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002700}
2701
Roland McGrath7bb44ad2007-05-23 13:57:44 -07002702/*
Thomas Gleixneree761f62013-03-21 22:49:32 +01002703 * Idle thread specific functions to determine the need_resched
2704 * polling state. We have two versions, one based on TS_POLLING in
2705 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2706 * thread_info.flags
2707 */
2708#ifdef TS_POLLING
2709static inline int tsk_is_polling(struct task_struct *p)
2710{
2711 return task_thread_info(p)->status & TS_POLLING;
2712}
Peter Zijlstraea811742013-09-11 12:43:13 +02002713static inline void __current_set_polling(void)
Thomas Gleixner3a98f872013-03-21 22:49:33 +01002714{
2715 current_thread_info()->status |= TS_POLLING;
2716}
2717
Peter Zijlstraea811742013-09-11 12:43:13 +02002718static inline bool __must_check current_set_polling_and_test(void)
2719{
2720 __current_set_polling();
2721
2722 /*
2723 * Polling state must be visible before we test NEED_RESCHED,
2724 * paired by resched_task()
2725 */
2726 smp_mb();
2727
2728 return unlikely(tif_need_resched());
2729}
2730
2731static inline void __current_clr_polling(void)
Thomas Gleixner3a98f872013-03-21 22:49:33 +01002732{
2733 current_thread_info()->status &= ~TS_POLLING;
Peter Zijlstraea811742013-09-11 12:43:13 +02002734}
2735
2736static inline bool __must_check current_clr_polling_and_test(void)
2737{
2738 __current_clr_polling();
2739
2740 /*
2741 * Polling state must be visible before we test NEED_RESCHED,
2742 * paired by resched_task()
2743 */
2744 smp_mb();
2745
2746 return unlikely(tif_need_resched());
Thomas Gleixner3a98f872013-03-21 22:49:33 +01002747}
Thomas Gleixneree761f62013-03-21 22:49:32 +01002748#elif defined(TIF_POLLING_NRFLAG)
2749static inline int tsk_is_polling(struct task_struct *p)
2750{
2751 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2752}
Peter Zijlstraea811742013-09-11 12:43:13 +02002753
2754static inline void __current_set_polling(void)
Thomas Gleixner3a98f872013-03-21 22:49:33 +01002755{
2756 set_thread_flag(TIF_POLLING_NRFLAG);
2757}
2758
Peter Zijlstraea811742013-09-11 12:43:13 +02002759static inline bool __must_check current_set_polling_and_test(void)
2760{
2761 __current_set_polling();
2762
2763 /*
2764 * Polling state must be visible before we test NEED_RESCHED,
2765 * paired by resched_task()
2766 *
2767 * XXX: assumes set/clear bit are identical barrier wise.
2768 */
2769 smp_mb__after_clear_bit();
2770
2771 return unlikely(tif_need_resched());
2772}
2773
2774static inline void __current_clr_polling(void)
Thomas Gleixner3a98f872013-03-21 22:49:33 +01002775{
2776 clear_thread_flag(TIF_POLLING_NRFLAG);
2777}
Peter Zijlstraea811742013-09-11 12:43:13 +02002778
2779static inline bool __must_check current_clr_polling_and_test(void)
2780{
2781 __current_clr_polling();
2782
2783 /*
2784 * Polling state must be visible before we test NEED_RESCHED,
2785 * paired by resched_task()
2786 */
2787 smp_mb__after_clear_bit();
2788
2789 return unlikely(tif_need_resched());
2790}
2791
Thomas Gleixneree761f62013-03-21 22:49:32 +01002792#else
2793static inline int tsk_is_polling(struct task_struct *p) { return 0; }
Peter Zijlstraea811742013-09-11 12:43:13 +02002794static inline void __current_set_polling(void) { }
2795static inline void __current_clr_polling(void) { }
2796
2797static inline bool __must_check current_set_polling_and_test(void)
2798{
2799 return unlikely(tif_need_resched());
2800}
2801static inline bool __must_check current_clr_polling_and_test(void)
2802{
2803 return unlikely(tif_need_resched());
2804}
Thomas Gleixneree761f62013-03-21 22:49:32 +01002805#endif
2806
Peter Zijlstra8cb75e02013-11-20 12:22:37 +01002807static inline void current_clr_polling(void)
2808{
2809 __current_clr_polling();
2810
2811 /*
2812 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2813 * Once the bit is cleared, we'll get IPIs with every new
2814 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2815 * fold.
2816 */
2817 smp_mb(); /* paired with resched_task() */
2818
2819 preempt_fold_need_resched();
2820}
2821
Peter Zijlstra75f93fe2013-09-27 17:30:03 +02002822static __always_inline bool need_resched(void)
2823{
2824 return unlikely(tif_need_resched());
2825}
2826
Thomas Gleixneree761f62013-03-21 22:49:32 +01002827/*
Frank Mayharf06febc2008-09-12 09:54:39 -07002828 * Thread group CPU time accounting.
2829 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +01002830void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
Peter Zijlstra4da94d492009-02-11 11:30:27 +01002831void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
Frank Mayharf06febc2008-09-12 09:54:39 -07002832
2833static inline void thread_group_cputime_init(struct signal_struct *sig)
2834{
Thomas Gleixneree30a7b2009-07-25 18:56:56 +02002835 raw_spin_lock_init(&sig->cputimer.lock);
Frank Mayharf06febc2008-09-12 09:54:39 -07002836}
2837
Frank Mayharf06febc2008-09-12 09:54:39 -07002838/*
Roland McGrath7bb44ad2007-05-23 13:57:44 -07002839 * Reevaluate whether the task has signals pending delivery.
2840 * Wake the task if so.
2841 * This is required every time the blocked sigset_t changes.
2842 * callers must hold sighand->siglock.
2843 */
2844extern void recalc_sigpending_and_wake(struct task_struct *t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002845extern void recalc_sigpending(void);
2846
Oleg Nesterov910ffdb2013-01-21 20:47:41 +01002847extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2848
2849static inline void signal_wake_up(struct task_struct *t, bool resume)
2850{
2851 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2852}
2853static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2854{
2855 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2856}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002857
2858/*
2859 * Wrappers for p->thread_info->cpu access. No-op on UP.
2860 */
2861#ifdef CONFIG_SMP
2862
2863static inline unsigned int task_cpu(const struct task_struct *p)
2864{
Al Viroa1261f542005-11-13 16:06:55 -08002865 return task_thread_info(p)->cpu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002866}
2867
Ingo Molnarb32e86b2013-10-07 11:29:30 +01002868static inline int task_node(const struct task_struct *p)
2869{
2870 return cpu_to_node(task_cpu(p));
2871}
2872
Ingo Molnarc65cc872007-07-09 18:51:58 +02002873extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002874
2875#else
2876
2877static inline unsigned int task_cpu(const struct task_struct *p)
2878{
2879 return 0;
2880}
2881
2882static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2883{
2884}
2885
2886#endif /* CONFIG_SMP */
2887
Rusty Russell96f874e22008-11-25 02:35:14 +10302888extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2889extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
Siddha, Suresh B5c45bf22006-06-27 02:54:42 -07002890
Dhaval Giani7c941432010-01-20 13:26:18 +01002891#ifdef CONFIG_CGROUP_SCHED
Yong Zhang07e06b02011-01-07 15:17:36 +08002892extern struct task_group root_task_group;
Peter Zijlstra8323f262012-06-22 13:36:05 +02002893#endif /* CONFIG_CGROUP_SCHED */
Srivatsa Vaddagiri9b5b7752007-10-15 17:00:09 +02002894
Dhaval Giani54e99122009-02-27 15:13:54 +05302895extern int task_can_switch_user(struct user_struct *up,
2896 struct task_struct *tsk);
2897
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002898#ifdef CONFIG_TASK_XACCT
2899static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2900{
Andrea Righi940389b2008-07-28 00:48:12 +02002901 tsk->ioac.rchar += amt;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002902}
2903
2904static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2905{
Andrea Righi940389b2008-07-28 00:48:12 +02002906 tsk->ioac.wchar += amt;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002907}
2908
2909static inline void inc_syscr(struct task_struct *tsk)
2910{
Andrea Righi940389b2008-07-28 00:48:12 +02002911 tsk->ioac.syscr++;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002912}
2913
2914static inline void inc_syscw(struct task_struct *tsk)
2915{
Andrea Righi940389b2008-07-28 00:48:12 +02002916 tsk->ioac.syscw++;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002917}
2918#else
2919static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2920{
2921}
2922
2923static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2924{
2925}
2926
2927static inline void inc_syscr(struct task_struct *tsk)
2928{
2929}
2930
2931static inline void inc_syscw(struct task_struct *tsk)
2932{
2933}
2934#endif
2935
Dave Hansen82455252008-02-04 22:28:59 -08002936#ifndef TASK_SIZE_OF
2937#define TASK_SIZE_OF(tsk) TASK_SIZE
2938#endif
2939
Balbir Singhcf475ad2008-04-29 01:00:16 -07002940#ifdef CONFIG_MM_OWNER
2941extern void mm_update_next_owner(struct mm_struct *mm);
2942extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2943#else
2944static inline void mm_update_next_owner(struct mm_struct *mm)
2945{
2946}
2947
2948static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2949{
2950}
2951#endif /* CONFIG_MM_OWNER */
2952
Jiri Slaby3e10e712009-11-19 17:16:37 +01002953static inline unsigned long task_rlimit(const struct task_struct *tsk,
2954 unsigned int limit)
2955{
2956 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2957}
2958
2959static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2960 unsigned int limit)
2961{
2962 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2963}
2964
2965static inline unsigned long rlimit(unsigned int limit)
2966{
2967 return task_rlimit(current, limit);
2968}
2969
2970static inline unsigned long rlimit_max(unsigned int limit)
2971{
2972 return task_rlimit_max(current, limit);
2973}
2974
Linus Torvalds1da177e2005-04-16 15:20:36 -07002975#endif