blob: ad86e1d7dbc27f446d0c5112f08eb4720b5aa6a2 [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 Morton7c3ab732006-12-10 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 */
Oleg Nesterovad866222014-04-07 15:38:46 -0700213#define EXIT_DEAD 16
214#define EXIT_ZOMBIE 32
Oleg Nesterovabd50b32014-04-07 15:38:42 -0700215#define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700216/* in tsk->state again */
Mike Galbraithaf927232007-10-15 17:00:13 +0200217#define TASK_DEAD 64
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500218#define TASK_WAKEKILL 128
Peter Zijlstrae9c84312009-09-15 14:43:03 +0200219#define TASK_WAKING 256
Thomas Gleixnerf2530dc2013-04-09 09:33:34 +0200220#define TASK_PARKED 512
221#define TASK_STATE_MAX 1024
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500222
Masatake YAMATOad0f6142014-05-22 11:54:20 -0700223#define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWP"
Peter Zijlstra73342152009-12-17 13:16:27 +0100224
Peter Zijlstrae1781532009-12-17 13:16:30 +0100225extern char ___assert_task_state[1 - 2*!!(
226 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500227
228/* Convenience macros for the sake of set_task_state */
229#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
230#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
231#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700232
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500233/* Convenience macros for the sake of wake_up */
234#define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500235#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500236
237/* get_task_state() */
238#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500239 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
Oleg Nesterov74e37202014-01-23 15:55:35 -0800240 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500241
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500242#define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
243#define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500244#define task_is_stopped_or_traced(task) \
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500245 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500246#define task_contributes_to_load(task) \
Nathan Lynche3c8ca82009-04-08 19:45:12 -0500247 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
Tejun Heo376fede2011-11-21 12:32:24 -0800248 (task->flags & PF_FROZEN) == 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700249
250#define __set_task_state(tsk, state_value) \
251 do { (tsk)->state = (state_value); } while (0)
252#define set_task_state(tsk, state_value) \
253 set_mb((tsk)->state, (state_value))
254
Andrew Morton498d0c52005-09-13 01:25:14 -0700255/*
256 * set_current_state() includes a barrier so that the write of current->state
257 * is correctly serialised wrt the caller's subsequent test of whether to
258 * actually sleep:
259 *
260 * set_current_state(TASK_UNINTERRUPTIBLE);
261 * if (do_i_need_to_sleep())
262 * schedule();
263 *
264 * If the caller does not need such serialisation then use __set_current_state()
265 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700266#define __set_current_state(state_value) \
267 do { current->state = (state_value); } while (0)
268#define set_current_state(state_value) \
269 set_mb(current->state, (state_value))
270
271/* Task command name length */
272#define TASK_COMM_LEN 16
273
Linus Torvalds1da177e2005-04-16 15:20:36 -0700274#include <linux/spinlock.h>
275
276/*
277 * This serializes "schedule()" and also protects
278 * the run-queue from deletions/modifications (but
279 * _adding_ to the beginning of the run-queue has
280 * a separate lock).
281 */
282extern rwlock_t tasklist_lock;
283extern spinlock_t mmlist_lock;
284
Ingo Molnar36c8b582006-07-03 00:25:41 -0700285struct task_struct;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700286
Paul E. McKenneydb1466b2010-03-03 07:46:56 -0800287#ifdef CONFIG_PROVE_RCU
288extern int lockdep_tasklist_lock_is_held(void);
289#endif /* #ifdef CONFIG_PROVE_RCU */
290
Linus Torvalds1da177e2005-04-16 15:20:36 -0700291extern void sched_init(void);
292extern void sched_init_smp(void);
Harvey Harrison2d07b252008-02-15 09:56:34 -0800293extern asmlinkage void schedule_tail(struct task_struct *prev);
Ingo Molnar36c8b582006-07-03 00:25:41 -0700294extern void init_idle(struct task_struct *idle, int cpu);
Ingo Molnar1df21052007-07-09 18:51:58 +0200295extern void init_idle_bootup_task(struct task_struct *idle);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700296
Andrew Morton89f19f02009-09-19 11:55:44 -0700297extern int runqueue_is_locked(int cpu);
Ingo Molnar017730c2008-05-12 21:20:52 +0200298
Frederic Weisbecker3451d022011-08-10 23:21:01 +0200299#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
Alex Shic1cc0172012-09-10 15:10:58 +0800300extern void nohz_balance_enter_idle(int cpu);
Suresh Siddha69e1e812011-12-01 17:07:33 -0800301extern void set_cpu_sd_state_idle(void);
Viresh Kumar6201b4d2014-03-18 16:26:07 +0530302extern int get_nohz_timer_target(int pinned);
Siddha, Suresh B46cb4b72007-05-08 00:32:51 -0700303#else
Alex Shic1cc0172012-09-10 15:10:58 +0800304static inline void nohz_balance_enter_idle(int cpu) { }
Peter Zijlstrafdaabd82011-12-06 12:47:55 +0100305static inline void set_cpu_sd_state_idle(void) { }
Viresh Kumar6201b4d2014-03-18 16:26:07 +0530306static inline int get_nohz_timer_target(int pinned)
307{
308 return smp_processor_id();
309}
Siddha, Suresh B46cb4b72007-05-08 00:32:51 -0700310#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700311
Ingo Molnare59e2ae2006-12-06 20:35:59 -0800312/*
Ingo Molnar39bc89f2007-04-25 20:50:03 -0700313 * Only dump TASK_* tasks. (0 for all tasks)
Ingo Molnare59e2ae2006-12-06 20:35:59 -0800314 */
315extern void show_state_filter(unsigned long state_filter);
316
317static inline void show_state(void)
318{
Ingo Molnar39bc89f2007-04-25 20:50:03 -0700319 show_state_filter(0);
Ingo Molnare59e2ae2006-12-06 20:35:59 -0800320}
321
Linus Torvalds1da177e2005-04-16 15:20:36 -0700322extern void show_regs(struct pt_regs *);
323
324/*
325 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
326 * task), SP is the stack pointer of the first frame that should be shown in the back
327 * trace (or NULL if the entire call-chain of the task should be shown).
328 */
329extern void show_stack(struct task_struct *task, unsigned long *sp);
330
331void io_schedule(void);
332long io_schedule_timeout(long timeout);
333
334extern void cpu_init (void);
335extern void trap_init(void);
336extern void update_process_times(int user);
337extern void scheduler_tick(void);
338
Ingo Molnar82a1fcb2008-01-25 21:08:02 +0100339extern void sched_show_task(struct task_struct *p);
340
Frederic Weisbecker19cc36c2010-05-13 02:30:49 +0200341#ifdef CONFIG_LOCKUP_DETECTOR
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700342extern void touch_softlockup_watchdog(void);
Jason Wesseld6ad3e22010-01-27 16:25:22 -0600343extern void touch_softlockup_watchdog_sync(void);
Jeremy Fitzhardinge04c91672007-05-08 00:28:05 -0700344extern void touch_all_softlockup_watchdogs(void);
Don Zickus332fbdb2010-05-07 17:11:45 -0400345extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
346 void __user *buffer,
347 size_t *lenp, loff_t *ppos);
Ingo Molnar9c44bc02008-05-12 21:21:04 +0200348extern unsigned int softlockup_panic;
Peter Zijlstra004417a2010-11-25 18:38:29 +0100349void lockup_detector_init(void);
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700350#else
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700351static inline void touch_softlockup_watchdog(void)
352{
353}
Jason Wesseld6ad3e22010-01-27 16:25:22 -0600354static inline void touch_softlockup_watchdog_sync(void)
355{
356}
Jeremy Fitzhardinge04c91672007-05-08 00:28:05 -0700357static inline void touch_all_softlockup_watchdogs(void)
358{
359}
Peter Zijlstra004417a2010-11-25 18:38:29 +0100360static inline void lockup_detector_init(void)
361{
362}
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700363#endif
364
Marcelo Tosatti8b414522013-10-11 21:39:26 -0300365#ifdef CONFIG_DETECT_HUNG_TASK
366void reset_hung_task_detector(void);
367#else
368static inline void reset_hung_task_detector(void)
369{
370}
371#endif
372
Linus Torvalds1da177e2005-04-16 15:20:36 -0700373/* Attach to any functions which should be ignored in wchan output. */
374#define __sched __attribute__((__section__(".sched.text")))
Ingo Molnardeaf2222007-11-28 15:52:56 +0100375
376/* Linker adds these: start and end of __sched functions */
377extern char __sched_text_start[], __sched_text_end[];
378
Linus Torvalds1da177e2005-04-16 15:20:36 -0700379/* Is this address in the __sched functions? */
380extern int in_sched_functions(unsigned long addr);
381
382#define MAX_SCHEDULE_TIMEOUT LONG_MAX
Harvey Harrisonb3c97522008-02-13 15:03:15 -0800383extern signed long schedule_timeout(signed long timeout);
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700384extern signed long schedule_timeout_interruptible(signed long timeout);
Matthew Wilcox294d5cc2007-12-06 11:59:46 -0500385extern signed long schedule_timeout_killable(signed long timeout);
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700386extern signed long schedule_timeout_uninterruptible(signed long timeout);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700387asmlinkage void schedule(void);
Thomas Gleixnerc5491ea2011-03-21 12:09:35 +0100388extern void schedule_preempt_disabled(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700389
Serge E. Hallynab516012006-10-02 02:18:06 -0700390struct nsproxy;
Cedric Le Goateracce2922007-07-15 23:40:59 -0700391struct user_namespace;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700392
David Howellsefc1a3b2010-01-15 17:01:35 -0800393#ifdef CONFIG_MMU
394extern void arch_pick_mmap_layout(struct mm_struct *mm);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700395extern unsigned long
396arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
397 unsigned long, unsigned long);
398extern unsigned long
399arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
400 unsigned long len, unsigned long pgoff,
401 unsigned long flags);
David Howellsefc1a3b2010-01-15 17:01:35 -0800402#else
403static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
404#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700405
Kees Cookd049f742013-11-12 15:11:17 -0800406#define SUID_DUMP_DISABLE 0 /* No setuid dumping */
407#define SUID_DUMP_USER 1 /* Dump as user of process */
408#define SUID_DUMP_ROOT 2 /* Dump as root */
409
Kawai, Hidehiro6c5d5232007-07-19 01:48:27 -0700410/* mm flags */
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700411
Oleg Nesterov7288e112014-01-23 15:55:32 -0800412/* for SUID_DUMP_* above */
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700413#define MMF_DUMPABLE_BITS 2
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700414#define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700415
Oleg Nesterov942be382014-01-23 15:55:34 -0800416extern void set_dumpable(struct mm_struct *mm, int value);
417/*
418 * This returns the actual value of the suid_dumpable flag. For things
419 * that are using this for checking for privilege transitions, it must
420 * test against SUID_DUMP_USER rather than treating it as a boolean
421 * value.
422 */
423static inline int __get_dumpable(unsigned long mm_flags)
424{
425 return mm_flags & MMF_DUMPABLE_MASK;
426}
427
428static inline int get_dumpable(struct mm_struct *mm)
429{
430 return __get_dumpable(mm->flags);
431}
432
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700433/* coredump filter bits */
434#define MMF_DUMP_ANON_PRIVATE 2
435#define MMF_DUMP_ANON_SHARED 3
436#define MMF_DUMP_MAPPED_PRIVATE 4
437#define MMF_DUMP_MAPPED_SHARED 5
Roland McGrath82df3972007-10-16 23:27:02 -0700438#define MMF_DUMP_ELF_HEADERS 6
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700439#define MMF_DUMP_HUGETLB_PRIVATE 7
440#define MMF_DUMP_HUGETLB_SHARED 8
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700441
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700442#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700443#define MMF_DUMP_FILTER_BITS 7
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700444#define MMF_DUMP_FILTER_MASK \
445 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
446#define MMF_DUMP_FILTER_DEFAULT \
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700447 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
Roland McGrath656eb2c2008-10-18 20:28:23 -0700448 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
449
450#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
451# define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
452#else
453# define MMF_DUMP_MASK_DEFAULT_ELF 0
454#endif
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700455 /* leave room for more dump flags */
456#define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
Andrea Arcangeliba761492011-01-13 15:46:58 -0800457#define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
Konstantin Khlebnikovbafb2822012-06-07 14:21:11 -0700458#define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700459
Oleg Nesterov9f68f6722012-08-19 16:15:09 +0200460#define MMF_HAS_UPROBES 19 /* has uprobes */
461#define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
Oleg Nesterovf8ac4ec2012-08-08 17:11:42 +0200462
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700463#define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
Kawai, Hidehiro6c5d5232007-07-19 01:48:27 -0700464
Linus Torvalds1da177e2005-04-16 15:20:36 -0700465struct sighand_struct {
466 atomic_t count;
467 struct k_sigaction action[_NSIG];
468 spinlock_t siglock;
Davide Libenzib8fceee2007-09-20 12:40:16 -0700469 wait_queue_head_t signalfd_wqh;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700470};
471
KaiGai Kohei0e464812006-06-25 05:49:24 -0700472struct pacct_struct {
KaiGai Koheif6ec29a2006-06-25 05:49:25 -0700473 int ac_flag;
474 long ac_exitcode;
KaiGai Kohei0e464812006-06-25 05:49:24 -0700475 unsigned long ac_mem;
KaiGai Kohei77787bf2006-06-25 05:49:26 -0700476 cputime_t ac_utime, ac_stime;
477 unsigned long ac_minflt, ac_majflt;
KaiGai Kohei0e464812006-06-25 05:49:24 -0700478};
479
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200480struct cpu_itimer {
481 cputime_t expires;
482 cputime_t incr;
Stanislaw Gruszka8356b5f2009-07-29 12:15:27 +0200483 u32 error;
484 u32 incr_error;
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200485};
486
Frank Mayharf06febc2008-09-12 09:54:39 -0700487/**
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +0100488 * struct cputime - snaphsot of system and user cputime
489 * @utime: time spent in user mode
490 * @stime: time spent in system mode
491 *
492 * Gathers a generic snapshot of user and system time.
493 */
494struct cputime {
495 cputime_t utime;
496 cputime_t stime;
497};
498
499/**
Frank Mayharf06febc2008-09-12 09:54:39 -0700500 * struct task_cputime - collected CPU time counts
501 * @utime: time spent in user mode, in &cputime_t units
502 * @stime: time spent in kernel mode, in &cputime_t units
503 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
Ingo Molnar5ce73a42008-09-14 17:11:46 +0200504 *
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +0100505 * This is an extension of struct cputime that includes the total runtime
506 * spent by the task from the scheduler point of view.
507 *
508 * As a result, this structure groups together three kinds of CPU time
509 * that are tracked for threads and thread groups. Most things considering
Frank Mayharf06febc2008-09-12 09:54:39 -0700510 * CPU time want to group these counts together and treat all three
511 * of them in parallel.
512 */
513struct task_cputime {
514 cputime_t utime;
515 cputime_t stime;
516 unsigned long long sum_exec_runtime;
517};
518/* Alternate field names when used to cache expirations. */
519#define prof_exp stime
520#define virt_exp utime
521#define sched_exp sum_exec_runtime
522
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100523#define INIT_CPUTIME \
524 (struct task_cputime) { \
Martin Schwidefsky64861632011-12-15 14:56:09 +0100525 .utime = 0, \
526 .stime = 0, \
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100527 .sum_exec_runtime = 0, \
528 }
529
Peter Zijlstraa233f112013-09-23 19:04:26 +0200530#ifdef CONFIG_PREEMPT_COUNT
531#define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
532#else
533#define PREEMPT_DISABLED PREEMPT_ENABLED
534#endif
535
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200536/*
537 * Disable preemption until the scheduler is running.
538 * Reset by start_kernel()->sched_init()->init_idle().
Peter Zijlstrad86ee482009-07-10 14:57:57 +0200539 *
540 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
541 * before the scheduler is active -- see should_resched().
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200542 */
Peter Zijlstraa233f112013-09-23 19:04:26 +0200543#define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200544
Frank Mayharf06febc2008-09-12 09:54:39 -0700545/**
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100546 * struct thread_group_cputimer - thread group interval timer counts
547 * @cputime: thread group interval timers.
548 * @running: non-zero when there are timers running and
549 * @cputime receives updates.
550 * @lock: lock for fields in this struct.
Frank Mayharf06febc2008-09-12 09:54:39 -0700551 *
552 * This structure contains the version of task_cputime, above, that is
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100553 * used for thread group CPU timer calculations.
Frank Mayharf06febc2008-09-12 09:54:39 -0700554 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100555struct thread_group_cputimer {
556 struct task_cputime cputime;
557 int running;
Thomas Gleixneree30a7b2009-07-25 18:56:56 +0200558 raw_spinlock_t lock;
Frank Mayharf06febc2008-09-12 09:54:39 -0700559};
Frank Mayharf06febc2008-09-12 09:54:39 -0700560
Ben Blum4714d1d2011-05-26 16:25:18 -0700561#include <linux/rwsem.h>
Mike Galbraith5091faa2010-11-30 14:18:03 +0100562struct autogroup;
563
Linus Torvalds1da177e2005-04-16 15:20:36 -0700564/*
Jonathan Neuschäfere815f0a2011-03-21 20:24:47 +0100565 * NOTE! "signal_struct" does not have its own
Linus Torvalds1da177e2005-04-16 15:20:36 -0700566 * locking, because a shared signal_struct always
567 * implies a shared sighand_struct, so locking
568 * sighand_struct is always a proper superset of
569 * the locking of signal_struct.
570 */
571struct signal_struct {
Oleg Nesterovea6d2902010-05-26 14:43:16 -0700572 atomic_t sigcnt;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700573 atomic_t live;
Oleg Nesterovb3ac0222010-05-26 14:43:24 -0700574 int nr_threads;
Oleg Nesterov0c740d02014-01-21 15:49:56 -0800575 struct list_head thread_head;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700576
577 wait_queue_head_t wait_chldexit; /* for wait4() */
578
579 /* current thread group signal load-balancing target: */
Ingo Molnar36c8b582006-07-03 00:25:41 -0700580 struct task_struct *curr_target;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700581
582 /* shared signal handling: */
583 struct sigpending shared_pending;
584
585 /* thread group exit support */
586 int group_exit_code;
587 /* overloaded:
588 * - notify group_exit_task when ->count is equal to notify_count
589 * - everyone except group_exit_task is stopped during signal delivery
590 * of fatal signals, group_exit_task processes the signal.
591 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700592 int notify_count;
Richard Kennedy07dd20e2008-08-01 13:18:04 +0100593 struct task_struct *group_exit_task;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700594
595 /* thread group stop support, overloads group_exit_code too */
596 int group_stop_count;
597 unsigned int flags; /* see SIGNAL_* flags below */
598
Lennart Poetteringebec18a2012-03-23 15:01:54 -0700599 /*
600 * PR_SET_CHILD_SUBREAPER marks a process, like a service
601 * manager, to re-parent orphan (double-forking) child processes
602 * to this process instead of 'init'. The service manager is
603 * able to receive SIGCHLD signals and is able to investigate
604 * the process until it calls wait(). All children of this
605 * process will inherit a flag if they should look for a
606 * child_subreaper process at exit.
607 */
608 unsigned int is_child_subreaper:1;
609 unsigned int has_child_subreaper:1;
610
Linus Torvalds1da177e2005-04-16 15:20:36 -0700611 /* POSIX.1b Interval Timers */
Pavel Emelyanov5ed67f02013-03-11 13:12:21 +0400612 int posix_timer_id;
613 struct list_head posix_timers;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700614
615 /* ITIMER_REAL timer for the process */
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800616 struct hrtimer real_timer;
Oleg Nesterovfea9d172008-02-08 04:19:19 -0800617 struct pid *leader_pid;
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800618 ktime_t it_real_incr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700619
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200620 /*
621 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
622 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
623 * values are defined to 0 and 1 respectively
624 */
625 struct cpu_itimer it[2];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700626
Frank Mayharf06febc2008-09-12 09:54:39 -0700627 /*
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100628 * Thread group totals for process CPU timers.
629 * See thread_group_cputimer(), et al, for details.
Frank Mayharf06febc2008-09-12 09:54:39 -0700630 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100631 struct thread_group_cputimer cputimer;
Frank Mayharf06febc2008-09-12 09:54:39 -0700632
633 /* Earliest-expiration cache. */
634 struct task_cputime cputime_expires;
635
636 struct list_head cpu_timers[3];
637
Eric W. Biedermanab521dc2007-02-12 00:53:00 -0800638 struct pid *tty_old_pgrp;
Cedric Le Goater1ec320a2006-12-08 02:37:55 -0800639
Linus Torvalds1da177e2005-04-16 15:20:36 -0700640 /* boolean value for session group leader */
641 int leader;
642
643 struct tty_struct *tty; /* NULL if no tty */
644
Mike Galbraith5091faa2010-11-30 14:18:03 +0100645#ifdef CONFIG_SCHED_AUTOGROUP
646 struct autogroup *autogroup;
647#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700648 /*
649 * Cumulative resource counters for dead threads in the group,
650 * and for reaped dead child processes forked by this group.
651 * Live threads maintain their own counters and add to these
652 * in __exit_signal, except for the group leader.
653 */
Peter Zijlstra32bd6712009-02-05 12:24:15 +0100654 cputime_t utime, stime, cutime, cstime;
Laurent Vivier9ac52312007-10-15 17:00:19 +0200655 cputime_t gtime;
656 cputime_t cgtime;
Frederic Weisbecker9fbc42e2013-02-25 17:25:39 +0100657#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +0100658 struct cputime prev_cputime;
Hidetoshi Seto0cf55e12009-12-02 17:28:07 +0900659#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700660 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
661 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
Eric Dumazet6eaeeab2007-05-10 22:22:37 -0700662 unsigned long inblock, oublock, cinblock, coublock;
Jiri Pirko1f102062009-09-22 16:44:10 -0700663 unsigned long maxrss, cmaxrss;
Andrea Righi940389b2008-07-28 00:48:12 +0200664 struct task_io_accounting ioac;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700665
666 /*
Peter Zijlstra32bd6712009-02-05 12:24:15 +0100667 * Cumulative ns of schedule CPU time fo dead threads in the
668 * group, not including a zombie group leader, (This only differs
669 * from jiffies_to_ns(utime + stime) if sched_clock uses something
670 * other than jiffies.)
671 */
672 unsigned long long sum_sched_runtime;
673
674 /*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700675 * We don't bother to synchronize most readers of this at all,
676 * because there is no reader checking a limit that actually needs
677 * to get both rlim_cur and rlim_max atomically, and either one
678 * alone is a single word that can safely be read normally.
679 * getrlimit/setrlimit use task_lock(current->group_leader) to
680 * protect this instead of the siglock, because they really
681 * have no need to disable irqs.
682 */
683 struct rlimit rlim[RLIM_NLIMITS];
684
KaiGai Kohei0e464812006-06-25 05:49:24 -0700685#ifdef CONFIG_BSD_PROCESS_ACCT
686 struct pacct_struct pacct; /* per-process accounting information */
687#endif
Shailabh Nagarad4ecbc2006-07-14 00:24:44 -0700688#ifdef CONFIG_TASKSTATS
Shailabh Nagarad4ecbc2006-07-14 00:24:44 -0700689 struct taskstats *stats;
690#endif
Miloslav Trmac522ed772007-07-15 23:40:56 -0700691#ifdef CONFIG_AUDIT
692 unsigned audit_tty;
Richard Guy Briggs46e959e2013-05-03 14:03:50 -0400693 unsigned audit_tty_log_passwd;
Miloslav Trmac522ed772007-07-15 23:40:56 -0700694 struct tty_audit_buf *tty_audit_buf;
695#endif
Ben Blum4714d1d2011-05-26 16:25:18 -0700696#ifdef CONFIG_CGROUPS
697 /*
Tejun Heo77e4ef92011-12-12 18:12:21 -0800698 * group_rwsem prevents new tasks from entering the threadgroup and
699 * member tasks from exiting,a more specifically, setting of
700 * PF_EXITING. fork and exit paths are protected with this rwsem
701 * using threadgroup_change_begin/end(). Users which require
702 * threadgroup to remain stable should use threadgroup_[un]lock()
703 * which also takes care of exec path. Currently, cgroup is the
704 * only user.
Ben Blum4714d1d2011-05-26 16:25:18 -0700705 */
Tejun Heo257058a2011-12-12 18:12:21 -0800706 struct rw_semaphore group_rwsem;
Ben Blum4714d1d2011-05-26 16:25:18 -0700707#endif
KOSAKI Motohiro28b83c52009-09-21 17:03:13 -0700708
David Rientjese1e12d22012-12-11 16:02:56 -0800709 oom_flags_t oom_flags;
David Rientjesa9c58b902012-12-11 16:02:54 -0800710 short oom_score_adj; /* OOM kill score adjustment */
711 short oom_score_adj_min; /* OOM kill score adjustment min value.
712 * Only settable by CAP_SYS_RESOURCE. */
KOSAKI Motohiro9b1bf122010-10-27 15:34:08 -0700713
714 struct mutex cred_guard_mutex; /* guard against foreign influences on
715 * credential calculations
716 * (notably. ptrace) */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700717};
718
719/*
720 * Bits in flags field of signal_struct.
721 */
722#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
Oleg Nesterovee77f072011-04-01 20:12:38 +0200723#define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
724#define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
Oleg Nesterov403bad72013-04-30 15:28:10 -0700725#define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
Oleg Nesterove4420552008-04-30 00:52:44 -0700726/*
727 * Pending notifications to parent.
728 */
729#define SIGNAL_CLD_STOPPED 0x00000010
730#define SIGNAL_CLD_CONTINUED 0x00000020
731#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700732
Oleg Nesterovfae5fa42008-04-30 00:53:03 -0700733#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
734
Oleg Nesteroved5d2ca2008-02-04 22:27:24 -0800735/* If true, all threads except ->group_exit_task have pending SIGKILL */
736static inline int signal_group_exit(const struct signal_struct *sig)
737{
738 return (sig->flags & SIGNAL_GROUP_EXIT) ||
739 (sig->group_exit_task != NULL);
740}
741
Linus Torvalds1da177e2005-04-16 15:20:36 -0700742/*
743 * Some day this will be a full-fledged user tracking system..
744 */
745struct user_struct {
746 atomic_t __count; /* reference count */
747 atomic_t processes; /* How many processes does this user have? */
748 atomic_t files; /* How many open files does this user have? */
749 atomic_t sigpending; /* How many pending signals does this user have? */
Amy Griffis2d9048e2006-06-01 13:10:59 -0700750#ifdef CONFIG_INOTIFY_USER
Robert Love0eeca282005-07-12 17:06:03 -0400751 atomic_t inotify_watches; /* How many inotify watches does this user have? */
752 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
753#endif
Eric Paris4afeff82010-10-28 17:21:58 -0400754#ifdef CONFIG_FANOTIFY
755 atomic_t fanotify_listeners;
756#endif
Davide Libenzi7ef99642008-12-01 13:13:55 -0800757#ifdef CONFIG_EPOLL
Robin Holt52bd19f72011-01-12 17:00:01 -0800758 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
Davide Libenzi7ef99642008-12-01 13:13:55 -0800759#endif
Alexey Dobriyan970a8642007-10-16 23:30:09 -0700760#ifdef CONFIG_POSIX_MQUEUE
Linus Torvalds1da177e2005-04-16 15:20:36 -0700761 /* protected by mq_lock */
762 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
Alexey Dobriyan970a8642007-10-16 23:30:09 -0700763#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700764 unsigned long locked_shm; /* How many pages of mlocked shm ? */
765
766#ifdef CONFIG_KEYS
767 struct key *uid_keyring; /* UID specific keyring */
768 struct key *session_keyring; /* UID's default session keyring */
769#endif
770
771 /* Hash table maintenance information */
Pavel Emelyanov735de222007-09-18 22:46:44 -0700772 struct hlist_node uidhash_node;
Eric W. Biederman7b44ab92011-11-16 23:20:58 -0800773 kuid_t uid;
Srivatsa Vaddagiri24e377a2007-10-15 17:00:09 +0200774
Ingo Molnarcdd6c482009-09-21 12:02:48 +0200775#ifdef CONFIG_PERF_EVENTS
Peter Zijlstra789f90f2009-05-15 15:19:27 +0200776 atomic_long_t locked_vm;
777#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700778};
779
Kay Sieverseb41d942007-11-02 13:47:53 +0100780extern int uids_sysfs_init(void);
Dhaval Giani5cb350b2007-10-15 17:00:14 +0200781
Eric W. Biederman7b44ab92011-11-16 23:20:58 -0800782extern struct user_struct *find_user(kuid_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700783
784extern struct user_struct root_user;
785#define INIT_USER (&root_user)
786
David Howellsb6dff3e2008-11-14 10:39:16 +1100787
Linus Torvalds1da177e2005-04-16 15:20:36 -0700788struct backing_dev_info;
789struct reclaim_state;
790
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700791#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700792struct sched_info {
793 /* cumulative counters */
Ingo Molnar2d723762007-10-15 17:00:12 +0200794 unsigned long pcount; /* # of times run on this cpu */
Ken Chen9c2c4802008-12-16 23:41:22 -0800795 unsigned long long run_delay; /* time spent waiting on a runqueue */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700796
797 /* timestamps */
Balbir Singh172ba842007-07-09 18:52:00 +0200798 unsigned long long last_arrival,/* when we last ran on a cpu */
799 last_queued; /* when we were last queued to run */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700800};
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700801#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700802
Shailabh Nagarca74e922006-07-14 00:24:36 -0700803#ifdef CONFIG_TASK_DELAY_ACCT
804struct task_delay_info {
805 spinlock_t lock;
806 unsigned int flags; /* Private per-task flags */
807
808 /* For each stat XXX, add following, aligned appropriately
809 *
810 * struct timespec XXX_start, XXX_end;
811 * u64 XXX_delay;
812 * u32 XXX_count;
813 *
814 * Atomicity of updates to XXX_delay, XXX_count protected by
815 * single lock above (split into XXX_lock if contention is an issue).
816 */
Shailabh Nagar0ff92242006-07-14 00:24:37 -0700817
818 /*
819 * XXX_count is incremented on every XXX operation, the delay
820 * associated with the operation is added to XXX_delay.
821 * XXX_delay contains the accumulated delay time in nanoseconds.
822 */
823 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
824 u64 blkio_delay; /* wait for sync block io completion */
825 u64 swapin_delay; /* wait for swapin block io completion */
826 u32 blkio_count; /* total count of the number of sync block */
827 /* io operations performed */
828 u32 swapin_count; /* total count of the number of swapin block */
829 /* io operations performed */
Keika Kobayashi873b4772008-07-25 01:48:52 -0700830
831 struct timespec freepages_start, freepages_end;
832 u64 freepages_delay; /* wait for memory reclaim */
833 u32 freepages_count; /* total count of memory reclaim */
Shailabh Nagarca74e922006-07-14 00:24:36 -0700834};
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700835#endif /* CONFIG_TASK_DELAY_ACCT */
836
837static inline int sched_info_on(void)
838{
839#ifdef CONFIG_SCHEDSTATS
840 return 1;
841#elif defined(CONFIG_TASK_DELAY_ACCT)
842 extern int delayacct_on;
843 return delayacct_on;
844#else
845 return 0;
Shailabh Nagarca74e922006-07-14 00:24:36 -0700846#endif
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700847}
Shailabh Nagarca74e922006-07-14 00:24:36 -0700848
Ingo Molnard15bcfd2007-07-09 18:51:57 +0200849enum cpu_idle_type {
850 CPU_IDLE,
851 CPU_NOT_IDLE,
852 CPU_NEWLY_IDLE,
853 CPU_MAX_IDLE_TYPES
Linus Torvalds1da177e2005-04-16 15:20:36 -0700854};
855
856/*
Nikhil Rao1399fa72011-05-18 10:09:39 -0700857 * Increase resolution of cpu_power calculations
858 */
859#define SCHED_POWER_SHIFT 10
860#define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700861
Nikhil Rao1399fa72011-05-18 10:09:39 -0700862/*
863 * sched-domains (multiprocessor balancing) declarations:
864 */
Peter Williams2dd73a42006-06-27 02:54:34 -0700865#ifdef CONFIG_SMP
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200866#define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
867#define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
868#define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
869#define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
Peter Zijlstrac88d5912009-09-10 13:50:02 +0200870#define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200871#define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200872#define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200873#define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
874#define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
Michael Neuling532cb4c2010-06-08 14:57:02 +1000875#define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200876#define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
Peter Zijlstrae3589f62011-07-15 10:35:52 +0200877#define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
Mel Gorman3a7053b2013-10-07 11:29:00 +0100878#define SD_NUMA 0x4000 /* cross-node balancing */
Siddha, Suresh B5c45bf22006-06-27 02:54:42 -0700879
Michael Neuling532cb4c2010-06-08 14:57:02 +1000880extern int __weak arch_sd_sibiling_asym_packing(void);
881
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +0900882struct sched_domain_attr {
883 int relax_domain_level;
884};
885
886#define SD_ATTR_INIT (struct sched_domain_attr) { \
887 .relax_domain_level = -1, \
888}
889
Peter Zijlstra60495e72011-04-07 14:10:04 +0200890extern int sched_domain_level_max;
891
Li Zefan5e6521e2013-03-05 16:06:23 +0800892struct sched_group;
893
Linus Torvalds1da177e2005-04-16 15:20:36 -0700894struct sched_domain {
895 /* These fields must be setup */
896 struct sched_domain *parent; /* top domain must be null terminated */
Siddha, Suresh B1a848872006-10-03 01:14:08 -0700897 struct sched_domain *child; /* bottom domain must be null terminated */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700898 struct sched_group *groups; /* the balancing groups of the domain */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700899 unsigned long min_interval; /* Minimum balance interval ms */
900 unsigned long max_interval; /* Maximum balance interval ms */
901 unsigned int busy_factor; /* less balancing by factor if busy */
902 unsigned int imbalance_pct; /* No balance until over watermark */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700903 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
Nick Piggin78979862005-06-25 14:57:13 -0700904 unsigned int busy_idx;
905 unsigned int idle_idx;
906 unsigned int newidle_idx;
907 unsigned int wake_idx;
Nick Piggin147cbb42005-06-25 14:57:19 -0700908 unsigned int forkexec_idx;
Peter Zijlstraa52bfd732009-09-01 10:34:35 +0200909 unsigned int smt_gain;
Vincent Guittot25f55d92013-04-23 16:59:02 +0200910
911 int nohz_idle; /* NOHZ IDLE status */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700912 int flags; /* See SD_* */
Peter Zijlstra60495e72011-04-07 14:10:04 +0200913 int level;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700914
915 /* Runtime fields. */
916 unsigned long last_balance; /* init to jiffies. units in jiffies */
917 unsigned int balance_interval; /* initialise to 1. units in ms. */
918 unsigned int nr_balance_failed; /* initialise to 0 */
919
Jason Lowf48627e2013-09-13 11:26:53 -0700920 /* idle_balance() stats */
Jason Low9bd721c2013-09-13 11:26:52 -0700921 u64 max_newidle_lb_cost;
Jason Lowf48627e2013-09-13 11:26:53 -0700922 unsigned long next_decay_max_lb_cost;
Peter Zijlstra2398f2c2008-06-27 13:41:35 +0200923
Linus Torvalds1da177e2005-04-16 15:20:36 -0700924#ifdef CONFIG_SCHEDSTATS
925 /* load_balance() stats */
Ken Chen480b9432007-10-18 21:32:56 +0200926 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
927 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
928 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
929 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
930 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
931 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
932 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
933 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700934
935 /* Active load balancing */
Ken Chen480b9432007-10-18 21:32:56 +0200936 unsigned int alb_count;
937 unsigned int alb_failed;
938 unsigned int alb_pushed;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700939
Nick Piggin68767a02005-06-25 14:57:20 -0700940 /* SD_BALANCE_EXEC stats */
Ken Chen480b9432007-10-18 21:32:56 +0200941 unsigned int sbe_count;
942 unsigned int sbe_balanced;
943 unsigned int sbe_pushed;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700944
Nick Piggin68767a02005-06-25 14:57:20 -0700945 /* SD_BALANCE_FORK stats */
Ken Chen480b9432007-10-18 21:32:56 +0200946 unsigned int sbf_count;
947 unsigned int sbf_balanced;
948 unsigned int sbf_pushed;
Nick Piggin68767a02005-06-25 14:57:20 -0700949
Linus Torvalds1da177e2005-04-16 15:20:36 -0700950 /* try_to_wake_up() stats */
Ken Chen480b9432007-10-18 21:32:56 +0200951 unsigned int ttwu_wake_remote;
952 unsigned int ttwu_move_affine;
953 unsigned int ttwu_move_balance;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700954#endif
Ingo Molnara5d8c342008-10-09 11:35:51 +0200955#ifdef CONFIG_SCHED_DEBUG
956 char *name;
957#endif
Peter Zijlstradce840a2011-04-07 14:09:50 +0200958 union {
959 void *private; /* used during construction */
960 struct rcu_head rcu; /* used during destruction */
961 };
Rusty Russell6c99e9a2008-11-25 02:35:04 +1030962
Peter Zijlstra669c55e2010-04-16 14:59:29 +0200963 unsigned int span_weight;
Ingo Molnar4200efd2009-05-19 09:22:19 +0200964 /*
965 * Span of all CPUs in this domain.
966 *
967 * NOTE: this field is variable length. (Allocated dynamically
968 * by attaching extra space to the end of the structure,
969 * depending on how many CPUs the kernel has booted up with)
Ingo Molnar4200efd2009-05-19 09:22:19 +0200970 */
971 unsigned long span[0];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700972};
973
Rusty Russell758b2cd2008-11-25 02:35:04 +1030974static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
975{
Rusty Russell6c99e9a2008-11-25 02:35:04 +1030976 return to_cpumask(sd->span);
Rusty Russell758b2cd2008-11-25 02:35:04 +1030977}
978
Rusty Russellacc3f5d2009-11-03 14:53:40 +1030979extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +0900980 struct sched_domain_attr *dattr_new);
Paul Jackson029190c2007-10-18 23:40:20 -0700981
Rusty Russellacc3f5d2009-11-03 14:53:40 +1030982/* Allocate an array of sched domains, for partition_sched_domains(). */
983cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
984void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
985
Peter Zijlstra39be3502012-01-26 12:44:34 +0100986bool cpus_share_cache(int this_cpu, int that_cpu);
987
Ingo Molnar1b427c12008-07-18 14:01:39 +0200988#else /* CONFIG_SMP */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700989
Ingo Molnar1b427c12008-07-18 14:01:39 +0200990struct sched_domain_attr;
991
992static inline void
Rusty Russellacc3f5d2009-11-03 14:53:40 +1030993partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
Ingo Molnar1b427c12008-07-18 14:01:39 +0200994 struct sched_domain_attr *dattr_new)
Con Kolivasd02c7a82007-07-26 13:40:43 +0200995{
Con Kolivasd02c7a82007-07-26 13:40:43 +0200996}
Peter Zijlstra39be3502012-01-26 12:44:34 +0100997
998static inline bool cpus_share_cache(int this_cpu, int that_cpu)
999{
1000 return true;
1001}
1002
Ingo Molnar1b427c12008-07-18 14:01:39 +02001003#endif /* !CONFIG_SMP */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001004
Peter Zijlstra47fe38f2009-09-02 13:49:18 +02001005
Linus Torvalds1da177e2005-04-16 15:20:36 -07001006struct io_context; /* See blkdev.h */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001007
Linus Torvalds1da177e2005-04-16 15:20:36 -07001008
Chen, Kenneth W383f2832005-09-09 13:02:02 -07001009#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
Ingo Molnar36c8b582006-07-03 00:25:41 -07001010extern void prefetch_stack(struct task_struct *t);
Chen, Kenneth W383f2832005-09-09 13:02:02 -07001011#else
1012static inline void prefetch_stack(struct task_struct *t) { }
1013#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001014
1015struct audit_context; /* See audit.c */
1016struct mempolicy;
Jens Axboeb92ce552006-04-11 13:52:07 +02001017struct pipe_inode_info;
Serge E. Hallyn4865ecf2006-10-02 02:18:14 -07001018struct uts_namespace;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001019
Ingo Molnar20b8a592007-07-09 18:51:58 +02001020struct load_weight {
Peter Zijlstra9dbdb152013-11-18 18:27:06 +01001021 unsigned long weight;
1022 u32 inv_weight;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001023};
1024
Paul Turner9d85f212012-10-04 13:18:29 +02001025struct sched_avg {
1026 /*
1027 * These sums represent an infinite geometric series and so are bound
Kamalesh Babulal239003e2013-06-27 11:34:09 +05301028 * above by 1024/(1-y). Thus we only need a u32 to store them for all
Paul Turner9d85f212012-10-04 13:18:29 +02001029 * choices of y < 1-2^(-32)*1024.
1030 */
1031 u32 runnable_avg_sum, runnable_avg_period;
1032 u64 last_runnable_update;
Paul Turner9ee474f2012-10-04 13:18:30 +02001033 s64 decay_count;
Paul Turner2dac7542012-10-04 13:18:30 +02001034 unsigned long load_avg_contrib;
Paul Turner9d85f212012-10-04 13:18:29 +02001035};
1036
Ingo Molnar94c18222007-08-02 17:41:40 +02001037#ifdef CONFIG_SCHEDSTATS
Lucas De Marchi41acab82010-03-10 23:37:45 -03001038struct sched_statistics {
Ingo Molnar94c18222007-08-02 17:41:40 +02001039 u64 wait_start;
1040 u64 wait_max;
Arjan van de Ven6d082592008-01-25 21:08:35 +01001041 u64 wait_count;
1042 u64 wait_sum;
Arjan van de Ven8f0dfc32009-07-20 11:26:58 -07001043 u64 iowait_count;
1044 u64 iowait_sum;
Ingo Molnar94c18222007-08-02 17:41:40 +02001045
1046 u64 sleep_start;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001047 u64 sleep_max;
Ingo Molnar94c18222007-08-02 17:41:40 +02001048 s64 sum_sleep_runtime;
1049
1050 u64 block_start;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001051 u64 block_max;
1052 u64 exec_max;
Ingo Molnareba1ed42007-10-15 17:00:02 +02001053 u64 slice_max;
Ingo Molnarcc367732007-10-15 17:00:18 +02001054
Ingo Molnarcc367732007-10-15 17:00:18 +02001055 u64 nr_migrations_cold;
1056 u64 nr_failed_migrations_affine;
1057 u64 nr_failed_migrations_running;
1058 u64 nr_failed_migrations_hot;
1059 u64 nr_forced_migrations;
Ingo Molnarcc367732007-10-15 17:00:18 +02001060
1061 u64 nr_wakeups;
1062 u64 nr_wakeups_sync;
1063 u64 nr_wakeups_migrate;
1064 u64 nr_wakeups_local;
1065 u64 nr_wakeups_remote;
1066 u64 nr_wakeups_affine;
1067 u64 nr_wakeups_affine_attempts;
1068 u64 nr_wakeups_passive;
1069 u64 nr_wakeups_idle;
Lucas De Marchi41acab82010-03-10 23:37:45 -03001070};
1071#endif
1072
1073struct sched_entity {
1074 struct load_weight load; /* for load-balancing */
1075 struct rb_node run_node;
1076 struct list_head group_node;
1077 unsigned int on_rq;
1078
1079 u64 exec_start;
1080 u64 sum_exec_runtime;
1081 u64 vruntime;
1082 u64 prev_sum_exec_runtime;
1083
Lucas De Marchi41acab82010-03-10 23:37:45 -03001084 u64 nr_migrations;
1085
Lucas De Marchi41acab82010-03-10 23:37:45 -03001086#ifdef CONFIG_SCHEDSTATS
1087 struct sched_statistics statistics;
Ingo Molnar94c18222007-08-02 17:41:40 +02001088#endif
1089
Ingo Molnar20b8a592007-07-09 18:51:58 +02001090#ifdef CONFIG_FAIR_GROUP_SCHED
Peter Zijlstrafed14d42012-02-11 06:05:00 +01001091 int depth;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001092 struct sched_entity *parent;
1093 /* rq on which this entity is (to be) queued: */
1094 struct cfs_rq *cfs_rq;
1095 /* rq "owned" by this entity/group: */
1096 struct cfs_rq *my_q;
1097#endif
Clark Williams8bd75c72013-02-07 09:47:07 -06001098
Alex Shi141965c2013-06-26 13:05:39 +08001099#ifdef CONFIG_SMP
Paul Turnerf4e26b12012-10-04 13:18:32 +02001100 /* Per-entity load-tracking */
Paul Turner9d85f212012-10-04 13:18:29 +02001101 struct sched_avg avg;
1102#endif
Ingo Molnar20b8a592007-07-09 18:51:58 +02001103};
Ingo Molnar70b97a72006-07-03 00:25:42 -07001104
Peter Zijlstrafa717062008-01-25 21:08:27 +01001105struct sched_rt_entity {
1106 struct list_head run_list;
Peter Zijlstra78f2c7d2008-01-25 21:08:27 +01001107 unsigned long timeout;
Ying Xue57d2aa02012-07-17 15:03:43 +08001108 unsigned long watchdog_stamp;
Richard Kennedybee367e2008-08-01 13:24:08 +01001109 unsigned int time_slice;
Peter Zijlstra6f505b12008-01-25 21:08:30 +01001110
Peter Zijlstra58d6c2d2008-04-19 19:45:00 +02001111 struct sched_rt_entity *back;
Peter Zijlstra052f1dc2008-02-13 15:45:40 +01001112#ifdef CONFIG_RT_GROUP_SCHED
Peter Zijlstra6f505b12008-01-25 21:08:30 +01001113 struct sched_rt_entity *parent;
1114 /* rq on which this entity is (to be) queued: */
1115 struct rt_rq *rt_rq;
1116 /* rq "owned" by this entity/group: */
1117 struct rt_rq *my_q;
1118#endif
Peter Zijlstrafa717062008-01-25 21:08:27 +01001119};
1120
Dario Faggioliaab03e02013-11-28 11:14:43 +01001121struct sched_dl_entity {
1122 struct rb_node rb_node;
1123
1124 /*
1125 * Original scheduling parameters. Copied here from sched_attr
1126 * during sched_setscheduler2(), they will remain the same until
1127 * the next sched_setscheduler2().
1128 */
1129 u64 dl_runtime; /* maximum runtime for each instance */
1130 u64 dl_deadline; /* relative deadline of each instance */
Harald Gustafsson755378a2013-11-07 14:43:40 +01001131 u64 dl_period; /* separation of two instances (period) */
Dario Faggioli332ac172013-11-07 14:43:45 +01001132 u64 dl_bw; /* dl_runtime / dl_deadline */
Dario Faggioliaab03e02013-11-28 11:14:43 +01001133
1134 /*
1135 * Actual scheduling parameters. Initialized with the values above,
1136 * they are continously updated during task execution. Note that
1137 * the remaining runtime could be < 0 in case we are in overrun.
1138 */
1139 s64 runtime; /* remaining runtime for this instance */
1140 u64 deadline; /* absolute deadline for this instance */
1141 unsigned int flags; /* specifying the scheduler behaviour */
1142
1143 /*
1144 * Some bool flags:
1145 *
1146 * @dl_throttled tells if we exhausted the runtime. If so, the
1147 * task has to wait for a replenishment to be performed at the
1148 * next firing of dl_timer.
1149 *
1150 * @dl_new tells if a new instance arrived. If so we must
1151 * start executing it with full runtime and reset its absolute
1152 * deadline;
Dario Faggioli2d3d8912013-11-07 14:43:44 +01001153 *
1154 * @dl_boosted tells if we are boosted due to DI. If so we are
1155 * outside bandwidth enforcement mechanism (but only until we
Juri Lelli5bfd1262014-04-15 13:49:04 +02001156 * exit the critical section);
1157 *
1158 * @dl_yielded tells if task gave up the cpu before consuming
1159 * all its available runtime during the last job.
Dario Faggioliaab03e02013-11-28 11:14:43 +01001160 */
Juri Lelli5bfd1262014-04-15 13:49:04 +02001161 int dl_throttled, dl_new, dl_boosted, dl_yielded;
Dario Faggioliaab03e02013-11-28 11:14:43 +01001162
1163 /*
1164 * Bandwidth enforcement timer. Each -deadline task has its
1165 * own bandwidth to be enforced, thus we need one timer per task.
1166 */
1167 struct hrtimer dl_timer;
1168};
Clark Williams8bd75c72013-02-07 09:47:07 -06001169
Paul E. McKenney86848962009-08-27 15:00:12 -07001170struct rcu_node;
1171
Peter Zijlstra8dc85d52010-09-02 16:50:03 +02001172enum perf_event_task_context {
1173 perf_invalid_context = -1,
1174 perf_hw_context = 0,
Peter Zijlstra89a1e182010-09-07 17:34:50 +02001175 perf_sw_context,
Peter Zijlstra8dc85d52010-09-02 16:50:03 +02001176 perf_nr_task_contexts,
1177};
1178
Linus Torvalds1da177e2005-04-16 15:20:36 -07001179struct task_struct {
1180 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
Roman Zippelf7e42172007-05-09 02:35:17 -07001181 void *stack;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001182 atomic_t usage;
William Cohen97dc32c2007-05-08 00:23:41 -07001183 unsigned int flags; /* per process flags, defined below */
1184 unsigned int ptrace;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001185
Peter Williams2dd73a42006-06-27 02:54:34 -07001186#ifdef CONFIG_SMP
Peter Zijlstrafa14ff42011-09-12 13:06:17 +02001187 struct llist_node wake_entry;
Peter Zijlstra3ca7a442011-04-05 17:23:40 +02001188 int on_cpu;
Michael Wang62470412013-07-04 12:55:51 +08001189 struct task_struct *last_wakee;
1190 unsigned long wakee_flips;
1191 unsigned long wakee_flip_decay_ts;
Peter Zijlstraac66f542013-10-07 11:29:16 +01001192
1193 int wake_cpu;
Nick Piggin4866cde2005-06-25 14:57:23 -07001194#endif
Peter Zijlstrafd2f4412011-04-05 17:23:44 +02001195 int on_rq;
Ingo Molnar50e645a2007-07-09 18:52:00 +02001196
Ingo Molnarb29739f2006-06-27 02:54:51 -07001197 int prio, static_prio, normal_prio;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001198 unsigned int rt_priority;
Ingo Molnar5522d5d2007-10-15 17:00:12 +02001199 const struct sched_class *sched_class;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001200 struct sched_entity se;
Peter Zijlstrafa717062008-01-25 21:08:27 +01001201 struct sched_rt_entity rt;
Peter Zijlstra8323f262012-06-22 13:36:05 +02001202#ifdef CONFIG_CGROUP_SCHED
1203 struct task_group *sched_task_group;
1204#endif
Dario Faggioliaab03e02013-11-28 11:14:43 +01001205 struct sched_dl_entity dl;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001206
Avi Kivitye107be32007-07-26 13:40:43 +02001207#ifdef CONFIG_PREEMPT_NOTIFIERS
1208 /* list of struct preempt_notifier: */
1209 struct hlist_head preempt_notifiers;
1210#endif
1211
Alexey Dobriyan6c5c9342006-09-29 01:59:40 -07001212#ifdef CONFIG_BLK_DEV_IO_TRACE
Jens Axboe2056a782006-03-23 20:00:26 +01001213 unsigned int btrace_seq;
Alexey Dobriyan6c5c9342006-09-29 01:59:40 -07001214#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001215
William Cohen97dc32c2007-05-08 00:23:41 -07001216 unsigned int policy;
Peter Zijlstra29baa742012-04-23 12:11:21 +02001217 int nr_cpus_allowed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001218 cpumask_t cpus_allowed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001219
Paul E. McKenneya57eb942010-06-29 16:49:16 -07001220#ifdef CONFIG_PREEMPT_RCU
Paul E. McKenneye260be62008-01-25 21:08:24 +01001221 int rcu_read_lock_nesting;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001222 char rcu_read_unlock_special;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001223 struct list_head rcu_node_entry;
Paul E. McKenneya57eb942010-06-29 16:49:16 -07001224#endif /* #ifdef CONFIG_PREEMPT_RCU */
1225#ifdef CONFIG_TREE_PREEMPT_RCU
1226 struct rcu_node *rcu_blocked_node;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001227#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
Paul E. McKenney24278d12010-09-27 17:25:23 -07001228#ifdef CONFIG_RCU_BOOST
1229 struct rt_mutex *rcu_boost_mutex;
1230#endif /* #ifdef CONFIG_RCU_BOOST */
Paul E. McKenneye260be62008-01-25 21:08:24 +01001231
Chandra Seetharaman52f17b62006-07-14 00:24:38 -07001232#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001233 struct sched_info sched_info;
1234#endif
1235
1236 struct list_head tasks;
Dario Faggioli806c09a2010-11-30 19:51:33 +01001237#ifdef CONFIG_SMP
Gregory Haskins917b6272008-12-29 09:39:53 -05001238 struct plist_node pushable_tasks;
Juri Lelli1baca4c2013-11-07 14:43:38 +01001239 struct rb_node pushable_dl_tasks;
Dario Faggioli806c09a2010-11-30 19:51:33 +01001240#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001241
1242 struct mm_struct *mm, *active_mm;
Jiri Kosina4471a672011-04-14 15:22:09 -07001243#ifdef CONFIG_COMPAT_BRK
1244 unsigned brk_randomized:1;
1245#endif
Davidlohr Bueso615d6e82014-04-07 15:37:25 -07001246 /* per-thread vma caching */
1247 u32 vmacache_seqnum;
1248 struct vm_area_struct *vmacache[VMACACHE_SIZE];
KAMEZAWA Hiroyuki34e55232010-03-05 13:41:40 -08001249#if defined(SPLIT_RSS_COUNTING)
1250 struct task_rss_stat rss_stat;
1251#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001252/* task state */
William Cohen97dc32c2007-05-08 00:23:41 -07001253 int exit_state;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001254 int exit_code, exit_signal;
1255 int pdeath_signal; /* The signal sent when the parent dies */
Tejun Heoa8f072c2011-06-02 11:13:59 +02001256 unsigned int jobctl; /* JOBCTL_*, siglock protected */
Andrei Epure9b89f6b2013-04-11 20:30:29 +03001257
1258 /* Used for emulating ABI behavior of previous Linux versions */
William Cohen97dc32c2007-05-08 00:23:41 -07001259 unsigned int personality;
Andrei Epure9b89f6b2013-04-11 20:30:29 +03001260
Kentaro Takedaf9ce1f12009-02-05 17:18:11 +09001261 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1262 * execve */
Arjan van de Ven8f0dfc32009-07-20 11:26:58 -07001263 unsigned in_iowait:1;
1264
Andy Lutomirski259e5e62012-04-12 16:47:50 -05001265 /* task may not gain privileges */
1266 unsigned no_new_privs:1;
Lennart Poetteringca94c442009-06-15 17:17:47 +02001267
1268 /* Revert to default priority/policy when forking */
1269 unsigned sched_reset_on_fork:1;
Peter Zijlstraa8e4f2e2011-04-05 17:23:49 +02001270 unsigned sched_contributes_to_load:1;
Lennart Poetteringca94c442009-06-15 17:17:47 +02001271
Linus Torvalds1da177e2005-04-16 15:20:36 -07001272 pid_t pid;
1273 pid_t tgid;
Arjan van de Ven0a4254052006-09-26 10:52:38 +02001274
Hiroshi Shimamoto13145622009-08-18 15:06:02 +09001275#ifdef CONFIG_CC_STACKPROTECTOR
Arjan van de Ven0a4254052006-09-26 10:52:38 +02001276 /* Canary value for the -fstack-protector gcc feature */
1277 unsigned long stack_canary;
Hiroshi Shimamoto13145622009-08-18 15:06:02 +09001278#endif
Oleg Nesterov4d1d61a2012-05-11 10:59:08 +10001279 /*
Linus Torvalds1da177e2005-04-16 15:20:36 -07001280 * pointers to (original) parent process, youngest child, younger sibling,
Oleg Nesterov4d1d61a2012-05-11 10:59:08 +10001281 * older sibling, respectively. (p->father can be replaced with
Roland McGrathf4700212008-03-24 18:36:23 -07001282 * p->real_parent->pid)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001283 */
Kees Cookabd63bc2011-12-14 14:39:26 -08001284 struct task_struct __rcu *real_parent; /* real parent process */
1285 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001286 /*
Roland McGrathf4700212008-03-24 18:36:23 -07001287 * children/sibling forms the list of my natural children
Linus Torvalds1da177e2005-04-16 15:20:36 -07001288 */
1289 struct list_head children; /* list of my children */
1290 struct list_head sibling; /* linkage in my parent's children list */
1291 struct task_struct *group_leader; /* threadgroup leader */
1292
Roland McGrathf4700212008-03-24 18:36:23 -07001293 /*
1294 * ptraced is the list of tasks this task is using ptrace on.
1295 * This includes both natural children and PTRACE_ATTACH targets.
1296 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1297 */
1298 struct list_head ptraced;
1299 struct list_head ptrace_entry;
1300
Linus Torvalds1da177e2005-04-16 15:20:36 -07001301 /* PID/PID hash table linkage. */
Eric W. Biederman92476d72006-03-31 02:31:42 -08001302 struct pid_link pids[PIDTYPE_MAX];
Oleg Nesterov47e65322006-03-28 16:11:25 -08001303 struct list_head thread_group;
Oleg Nesterov0c740d02014-01-21 15:49:56 -08001304 struct list_head thread_node;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001305
1306 struct completion *vfork_done; /* for vfork() */
1307 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1308 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1309
Michael Neulingc66f08b2007-10-18 03:06:34 -07001310 cputime_t utime, stime, utimescaled, stimescaled;
Laurent Vivier9ac52312007-10-15 17:00:19 +02001311 cputime_t gtime;
Frederic Weisbecker9fbc42e2013-02-25 17:25:39 +01001312#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +01001313 struct cputime prev_cputime;
Hidetoshi Setod99ca3b2009-12-02 17:26:47 +09001314#endif
Frederic Weisbecker6a616712012-12-16 20:00:34 +01001315#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1316 seqlock_t vtime_seqlock;
1317 unsigned long long vtime_snap;
1318 enum {
1319 VTIME_SLEEPING = 0,
1320 VTIME_USER,
1321 VTIME_SYS,
1322 } vtime_snap_whence;
1323#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001324 unsigned long nvcsw, nivcsw; /* context switch counts */
Tomas Janousek924b42d2007-07-15 23:39:42 -07001325 struct timespec start_time; /* monotonic time */
1326 struct timespec real_start_time; /* boot based time */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001327/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1328 unsigned long min_flt, maj_flt;
1329
Frank Mayharf06febc2008-09-12 09:54:39 -07001330 struct task_cputime cputime_expires;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001331 struct list_head cpu_timers[3];
1332
1333/* process credentials */
Arnd Bergmann1b0ba1c2010-02-24 19:45:09 +01001334 const struct cred __rcu *real_cred; /* objective and real subjective task
David Howells3b11a1d2008-11-14 10:39:26 +11001335 * credentials (COW) */
Arnd Bergmann1b0ba1c2010-02-24 19:45:09 +01001336 const struct cred __rcu *cred; /* effective (overridable) subjective task
David Howells3b11a1d2008-11-14 10:39:26 +11001337 * credentials (COW) */
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -07001338 char comm[TASK_COMM_LEN]; /* executable name excluding path
1339 - access with [gs]et_task_comm (which lock
1340 it with task_lock())
Linus Torvalds221af7f2010-01-28 22:14:42 -08001341 - initialized normally by setup_new_exec */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001342/* file system info */
1343 int link_count, total_link_count;
Alexey Dobriyan3d5b6fc2006-09-29 01:59:40 -07001344#ifdef CONFIG_SYSVIPC
Linus Torvalds1da177e2005-04-16 15:20:36 -07001345/* ipc stuff */
1346 struct sysv_sem sysvsem;
Alexey Dobriyan3d5b6fc2006-09-29 01:59:40 -07001347#endif
Mandeep Singh Bainese162b392009-01-15 11:08:40 -08001348#ifdef CONFIG_DETECT_HUNG_TASK
Ingo Molnar82a1fcb2008-01-25 21:08:02 +01001349/* hung task detection */
Ingo Molnar82a1fcb2008-01-25 21:08:02 +01001350 unsigned long last_switch_count;
1351#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001352/* CPU-specific state of this task */
1353 struct thread_struct thread;
1354/* filesystem information */
1355 struct fs_struct *fs;
1356/* open file information */
1357 struct files_struct *files;
Serge E. Hallyn1651e142006-10-02 02:18:08 -07001358/* namespaces */
Serge E. Hallynab516012006-10-02 02:18:06 -07001359 struct nsproxy *nsproxy;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001360/* signal handlers */
1361 struct signal_struct *signal;
1362 struct sighand_struct *sighand;
1363
1364 sigset_t blocked, real_blocked;
Roland McGrathf3de2722008-04-30 00:53:09 -07001365 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001366 struct sigpending pending;
1367
1368 unsigned long sas_ss_sp;
1369 size_t sas_ss_size;
1370 int (*notifier)(void *priv);
1371 void *notifier_data;
1372 sigset_t *notifier_mask;
Al Viro67d12142012-06-27 11:07:19 +04001373 struct callback_head *task_works;
Oleg Nesterove73f8952012-05-11 10:59:07 +10001374
Linus Torvalds1da177e2005-04-16 15:20:36 -07001375 struct audit_context *audit_context;
Al Virobfef93a2008-01-10 04:53:18 -05001376#ifdef CONFIG_AUDITSYSCALL
Eric W. Biedermane1760bd2012-09-10 22:39:43 -07001377 kuid_t loginuid;
Eric Paris4746ec52008-01-08 10:06:53 -05001378 unsigned int sessionid;
Al Virobfef93a2008-01-10 04:53:18 -05001379#endif
Will Drewry932eceb2012-04-12 16:47:54 -05001380 struct seccomp seccomp;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001381
1382/* Thread group tracking */
1383 u32 parent_exec_id;
1384 u32 self_exec_id;
Miao Xie58568d22009-06-16 15:31:49 -07001385/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1386 * mempolicy */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001387 spinlock_t alloc_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001388
Ingo Molnarb29739f2006-06-27 02:54:51 -07001389 /* Protection of the PI data structures: */
Thomas Gleixner1d615482009-11-17 14:54:03 +01001390 raw_spinlock_t pi_lock;
Ingo Molnarb29739f2006-06-27 02:54:51 -07001391
Ingo Molnar23f78d4a2006-06-27 02:54:53 -07001392#ifdef CONFIG_RT_MUTEXES
1393 /* PI waiters blocked on a rt_mutex held by this task */
Peter Zijlstrafb00aca2013-11-07 14:43:43 +01001394 struct rb_root pi_waiters;
1395 struct rb_node *pi_waiters_leftmost;
Ingo Molnar23f78d4a2006-06-27 02:54:53 -07001396 /* Deadlock detection and priority inheritance handling */
1397 struct rt_mutex_waiter *pi_blocked_on;
Dario Faggioli2d3d8912013-11-07 14:43:44 +01001398 /* Top pi_waiters task */
1399 struct task_struct *pi_top_task;
Ingo Molnar23f78d4a2006-06-27 02:54:53 -07001400#endif
1401
Ingo Molnar408894e2006-01-09 15:59:20 -08001402#ifdef CONFIG_DEBUG_MUTEXES
1403 /* mutex deadlock detection */
1404 struct mutex_waiter *blocked_on;
1405#endif
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001406#ifdef CONFIG_TRACE_IRQFLAGS
1407 unsigned int irq_events;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001408 unsigned long hardirq_enable_ip;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001409 unsigned long hardirq_disable_ip;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001410 unsigned int hardirq_enable_event;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001411 unsigned int hardirq_disable_event;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001412 int hardirqs_enabled;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001413 int hardirq_context;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001414 unsigned long softirq_disable_ip;
1415 unsigned long softirq_enable_ip;
1416 unsigned int softirq_disable_event;
1417 unsigned int softirq_enable_event;
1418 int softirqs_enabled;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001419 int softirq_context;
1420#endif
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001421#ifdef CONFIG_LOCKDEP
Peter Zijlstrabdb94412008-02-25 23:02:48 +01001422# define MAX_LOCK_DEPTH 48UL
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001423 u64 curr_chain_key;
1424 int lockdep_depth;
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001425 unsigned int lockdep_recursion;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001426 struct held_lock held_locks[MAX_LOCK_DEPTH];
Nick Piggincf40bd12009-01-21 08:12:39 +01001427 gfp_t lockdep_reclaim_gfp;
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001428#endif
Ingo Molnar408894e2006-01-09 15:59:20 -08001429
Linus Torvalds1da177e2005-04-16 15:20:36 -07001430/* journalling filesystem info */
1431 void *journal_info;
1432
Neil Brownd89d8792007-05-01 09:53:42 +02001433/* stacked block device info */
Akinobu Mitabddd87c2010-02-23 08:55:42 +01001434 struct bio_list *bio_list;
Neil Brownd89d8792007-05-01 09:53:42 +02001435
Jens Axboe73c10102011-03-08 13:19:51 +01001436#ifdef CONFIG_BLOCK
1437/* stack plugging */
1438 struct blk_plug *plug;
1439#endif
1440
Linus Torvalds1da177e2005-04-16 15:20:36 -07001441/* VM state */
1442 struct reclaim_state *reclaim_state;
1443
Linus Torvalds1da177e2005-04-16 15:20:36 -07001444 struct backing_dev_info *backing_dev_info;
1445
1446 struct io_context *io_context;
1447
1448 unsigned long ptrace_message;
1449 siginfo_t *last_siginfo; /* For ptrace use. */
Andrew Morton7c3ab732006-12-10 02:19:19 -08001450 struct task_io_accounting ioac;
Jay Lan8f0ab512006-09-30 23:28:59 -07001451#if defined(CONFIG_TASK_XACCT)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001452 u64 acct_rss_mem1; /* accumulated rss usage */
1453 u64 acct_vm_mem1; /* accumulated virtual memory usage */
Jonathan Lim49b5cf32008-07-25 01:48:40 -07001454 cputime_t acct_timexpd; /* stime + utime since last update */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001455#endif
1456#ifdef CONFIG_CPUSETS
Miao Xie58568d22009-06-16 15:31:49 -07001457 nodemask_t mems_allowed; /* Protected by alloc_lock */
Mel Gormancc9a6c82012-03-21 16:34:11 -07001458 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
Paul Jackson825a46a2006-03-24 03:16:03 -08001459 int cpuset_mem_spread_rotor;
Jack Steiner6adef3e2010-05-26 14:42:49 -07001460 int cpuset_slab_spread_rotor;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001461#endif
Paul Menageddbcc7e2007-10-18 23:39:30 -07001462#ifdef CONFIG_CGROUPS
Paul Menage817929e2007-10-18 23:39:36 -07001463 /* Control Group info protected by css_set_lock */
Arnd Bergmann2c392b82010-02-24 19:41:39 +01001464 struct css_set __rcu *cgroups;
Paul Menage817929e2007-10-18 23:39:36 -07001465 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1466 struct list_head cg_list;
Paul Menageddbcc7e2007-10-18 23:39:30 -07001467#endif
Alexey Dobriyan42b2dd02007-10-16 23:27:30 -07001468#ifdef CONFIG_FUTEX
Ingo Molnar0771dfe2006-03-27 01:16:22 -08001469 struct robust_list_head __user *robust_list;
Ingo Molnar34f192c2006-03-27 01:16:24 -08001470#ifdef CONFIG_COMPAT
1471 struct compat_robust_list_head __user *compat_robust_list;
1472#endif
Ingo Molnarc87e2832006-06-27 02:54:58 -07001473 struct list_head pi_state_list;
1474 struct futex_pi_state *pi_state_cache;
Alexey Dobriyan42b2dd02007-10-16 23:27:30 -07001475#endif
Ingo Molnarcdd6c482009-09-21 12:02:48 +02001476#ifdef CONFIG_PERF_EVENTS
Peter Zijlstra8dc85d52010-09-02 16:50:03 +02001477 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
Ingo Molnarcdd6c482009-09-21 12:02:48 +02001478 struct mutex perf_event_mutex;
1479 struct list_head perf_event_list;
Paul Mackerrasa63eaf32009-05-22 14:17:31 +10001480#endif
Thomas Gleixner8f47b182014-02-07 20:58:39 +01001481#ifdef CONFIG_DEBUG_PREEMPT
1482 unsigned long preempt_disable_ip;
1483#endif
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001484#ifdef CONFIG_NUMA
Miao Xie58568d22009-06-16 15:31:49 -07001485 struct mempolicy *mempolicy; /* Protected by alloc_lock */
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001486 short il_next;
Eric Dumazet207205a2011-03-22 16:30:44 -07001487 short pref_node_fork;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001488#endif
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001489#ifdef CONFIG_NUMA_BALANCING
1490 int numa_scan_seq;
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001491 unsigned int numa_scan_period;
Mel Gorman598f0ec2013-10-07 11:28:55 +01001492 unsigned int numa_scan_period_max;
Rik van Rielde1c9ce62013-10-07 11:29:39 +01001493 int numa_preferred_nid;
Mel Gorman6b9a7462013-10-07 11:29:11 +01001494 unsigned long numa_migrate_retry;
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001495 u64 node_stamp; /* migration stamp */
Rik van Riel7e2703e2014-01-27 17:03:45 -05001496 u64 last_task_numa_placement;
1497 u64 last_sum_exec_runtime;
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001498 struct callback_head numa_work;
Mel Gormanf809ca92013-10-07 11:28:57 +01001499
Peter Zijlstra8c8a7432013-10-07 11:29:21 +01001500 struct list_head numa_entry;
1501 struct numa_group *numa_group;
1502
Mel Gorman745d6142013-10-07 11:28:59 +01001503 /*
1504 * Exponential decaying average of faults on a per-node basis.
1505 * Scheduling placement decisions are made based on the these counts.
1506 * The values remain static for the duration of a PTE scan
1507 */
Rik van Rielff1df892014-01-27 17:03:41 -05001508 unsigned long *numa_faults_memory;
Mel Gorman83e1d2c2013-10-07 11:29:27 +01001509 unsigned long total_numa_faults;
Mel Gorman745d6142013-10-07 11:28:59 +01001510
1511 /*
1512 * numa_faults_buffer records faults per node during the current
Rik van Rielff1df892014-01-27 17:03:41 -05001513 * scan window. When the scan completes, the counts in
1514 * numa_faults_memory decay and these values are copied.
Mel Gorman745d6142013-10-07 11:28:59 +01001515 */
Rik van Rielff1df892014-01-27 17:03:41 -05001516 unsigned long *numa_faults_buffer_memory;
Mel Gorman745d6142013-10-07 11:28:59 +01001517
Rik van Riel04bb2f92013-10-07 11:29:36 +01001518 /*
Rik van Riel50ec8a42014-01-27 17:03:42 -05001519 * Track the nodes the process was running on when a NUMA hinting
1520 * fault was incurred.
1521 */
1522 unsigned long *numa_faults_cpu;
1523 unsigned long *numa_faults_buffer_cpu;
1524
1525 /*
Rik van Riel04bb2f92013-10-07 11:29:36 +01001526 * numa_faults_locality tracks if faults recorded during the last
1527 * scan window were remote/local. The task scan period is adapted
1528 * based on the locality of the faults with different weights
1529 * depending on whether they were shared or private faults
1530 */
1531 unsigned long numa_faults_locality[2];
1532
Ingo Molnarb32e86b2013-10-07 11:29:30 +01001533 unsigned long numa_pages_migrated;
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001534#endif /* CONFIG_NUMA_BALANCING */
1535
Ingo Molnare56d0902006-01-08 01:01:37 -08001536 struct rcu_head rcu;
Jens Axboeb92ce552006-04-11 13:52:07 +02001537
1538 /*
1539 * cache last used pipe for splice
1540 */
1541 struct pipe_inode_info *splice_pipe;
Eric Dumazet5640f762012-09-23 23:04:42 +00001542
1543 struct page_frag task_frag;
1544
Shailabh Nagarca74e922006-07-14 00:24:36 -07001545#ifdef CONFIG_TASK_DELAY_ACCT
1546 struct task_delay_info *delays;
1547#endif
Akinobu Mitaf4f154f2006-12-08 02:39:47 -08001548#ifdef CONFIG_FAULT_INJECTION
1549 int make_it_fail;
1550#endif
Wu Fengguang9d823e82011-06-11 18:10:12 -06001551 /*
1552 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1553 * balance_dirty_pages() for some dirty throttling pause
1554 */
1555 int nr_dirtied;
1556 int nr_dirtied_pause;
Wu Fengguang83712352011-06-11 19:25:42 -06001557 unsigned long dirty_paused_when; /* start of a write-and-pause period */
Wu Fengguang9d823e82011-06-11 18:10:12 -06001558
Arjan van de Ven97455122008-01-25 21:08:34 +01001559#ifdef CONFIG_LATENCYTOP
1560 int latency_record_count;
1561 struct latency_record latency_record[LT_SAVECOUNT];
1562#endif
Arjan van de Ven69766752008-09-01 15:52:40 -07001563 /*
1564 * time slack values; these are used to round up poll() and
1565 * select() etc timeout values. These are in nanoseconds.
1566 */
1567 unsigned long timer_slack_ns;
1568 unsigned long default_timer_slack_ns;
David Millerf8d570a2008-11-06 00:37:40 -08001569
Frederic Weisbeckerfb526072008-11-25 21:07:04 +01001570#ifdef CONFIG_FUNCTION_GRAPH_TRACER
Daniel Mack3ad2f3f2010-02-03 08:01:28 +08001571 /* Index of current stored address in ret_stack */
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001572 int curr_ret_stack;
1573 /* Stack of return addresses for return function tracing */
1574 struct ftrace_ret_stack *ret_stack;
Steven Rostedt8aef2d22009-03-24 01:10:15 -04001575 /* time stamp for last schedule */
1576 unsigned long long ftrace_timestamp;
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001577 /*
1578 * Number of functions that haven't been traced
1579 * because of depth overrun.
1580 */
1581 atomic_t trace_overrun;
Frederic Weisbecker380c4b12008-12-06 03:43:41 +01001582 /* Pause for the tracing */
1583 atomic_t tracing_graph_pause;
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001584#endif
Steven Rostedtea4e2bc2008-12-03 15:36:57 -05001585#ifdef CONFIG_TRACING
1586 /* state flags for use by tracers */
1587 unsigned long trace;
Steven Rostedtb1cff0a2011-05-25 14:27:43 -04001588 /* bitmask and counter of trace recursion */
Steven Rostedt261842b2009-04-16 21:41:52 -04001589 unsigned long trace_recursion;
1590#endif /* CONFIG_TRACING */
Andrew Mortonc255a452012-07-31 16:43:02 -07001591#ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
KAMEZAWA Hiroyuki569b8462009-12-15 16:47:03 -08001592 struct memcg_batch_info {
1593 int do_batch; /* incremented when batch uncharge started */
1594 struct mem_cgroup *memcg; /* target memcg of uncharge */
Johannes Weiner7ffd4ca2011-03-23 16:42:35 -07001595 unsigned long nr_pages; /* uncharged usage */
1596 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
KAMEZAWA Hiroyuki569b8462009-12-15 16:47:03 -08001597 } memcg_batch;
Glauber Costa0e9d92f2012-12-18 14:22:42 -08001598 unsigned int memcg_kmem_skip_account;
Johannes Weiner519e5242013-09-12 15:13:42 -07001599 struct memcg_oom_info {
Johannes Weiner49426422013-10-16 13:46:59 -07001600 struct mem_cgroup *memcg;
1601 gfp_t gfp_mask;
1602 int order;
Johannes Weiner519e5242013-09-12 15:13:42 -07001603 unsigned int may_oom:1;
1604 } memcg_oom;
KAMEZAWA Hiroyuki569b8462009-12-15 16:47:03 -08001605#endif
Srikar Dronamraju0326f5a2012-03-13 23:30:11 +05301606#ifdef CONFIG_UPROBES
1607 struct uprobe_task *utask;
Srikar Dronamraju0326f5a2012-03-13 23:30:11 +05301608#endif
Kent Overstreetcafe5632013-03-23 16:11:31 -07001609#if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1610 unsigned int sequential_io;
1611 unsigned int sequential_io_avg;
1612#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001613};
1614
Rusty Russell76e6eee2009-03-12 14:35:43 -06001615/* Future-safe accessor for struct task_struct's cpus_allowed. */
Rusty Russella4636812009-12-17 11:43:29 -06001616#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
Rusty Russell76e6eee2009-03-12 14:35:43 -06001617
Peter Zijlstra6688cc02013-10-07 11:29:24 +01001618#define TNF_MIGRATED 0x01
1619#define TNF_NO_GROUP 0x02
Rik van Rieldabe1d92013-10-07 11:29:34 +01001620#define TNF_SHARED 0x04
Rik van Riel04bb2f92013-10-07 11:29:36 +01001621#define TNF_FAULT_LOCAL 0x08
Peter Zijlstra6688cc02013-10-07 11:29:24 +01001622
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001623#ifdef CONFIG_NUMA_BALANCING
Peter Zijlstra6688cc02013-10-07 11:29:24 +01001624extern void task_numa_fault(int last_node, int node, int pages, int flags);
Mel Gormane29cf082013-10-07 11:29:22 +01001625extern pid_t task_numa_group_id(struct task_struct *p);
Mel Gorman1a687c22012-11-22 11:16:36 +00001626extern void set_numabalancing_state(bool enabled);
Rik van Riel82727012013-10-07 11:29:28 +01001627extern void task_numa_free(struct task_struct *p);
Rik van Riel10f39042014-01-27 17:03:44 -05001628extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1629 int src_nid, int dst_cpu);
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001630#else
Mel Gormanac8e8952013-10-07 11:29:03 +01001631static inline void task_numa_fault(int last_node, int node, int pages,
Peter Zijlstra6688cc02013-10-07 11:29:24 +01001632 int flags)
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001633{
1634}
Mel Gormane29cf082013-10-07 11:29:22 +01001635static inline pid_t task_numa_group_id(struct task_struct *p)
1636{
1637 return 0;
1638}
Mel Gorman1a687c22012-11-22 11:16:36 +00001639static inline void set_numabalancing_state(bool enabled)
1640{
1641}
Rik van Riel82727012013-10-07 11:29:28 +01001642static inline void task_numa_free(struct task_struct *p)
1643{
1644}
Rik van Riel10f39042014-01-27 17:03:44 -05001645static inline bool should_numa_migrate_memory(struct task_struct *p,
1646 struct page *page, int src_nid, int dst_cpu)
1647{
1648 return true;
1649}
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001650#endif
1651
Alexey Dobriyane8681712007-10-26 12:17:22 +04001652static inline struct pid *task_pid(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001653{
1654 return task->pids[PIDTYPE_PID].pid;
1655}
1656
Alexey Dobriyane8681712007-10-26 12:17:22 +04001657static inline struct pid *task_tgid(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001658{
1659 return task->group_leader->pids[PIDTYPE_PID].pid;
1660}
1661
Oleg Nesterov6dda81f2009-04-02 16:58:35 -07001662/*
1663 * Without tasklist or rcu lock it is not safe to dereference
1664 * the result of task_pgrp/task_session even if task == current,
1665 * we can race with another thread doing sys_setsid/sys_setpgid.
1666 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04001667static inline struct pid *task_pgrp(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001668{
1669 return task->group_leader->pids[PIDTYPE_PGID].pid;
1670}
1671
Alexey Dobriyane8681712007-10-26 12:17:22 +04001672static inline struct pid *task_session(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001673{
1674 return task->group_leader->pids[PIDTYPE_SID].pid;
1675}
1676
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001677struct pid_namespace;
1678
1679/*
1680 * the helpers to get the task's different pids as they are seen
1681 * from various namespaces
1682 *
1683 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
Eric W. Biederman44c4e1b2008-02-08 04:19:15 -08001684 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1685 * current.
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001686 * task_xid_nr_ns() : id seen from the ns specified;
1687 *
1688 * set_task_vxid() : assigns a virtual id to a task;
1689 *
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001690 * see also pid_nr() etc in include/linux/pid.h
1691 */
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001692pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1693 struct pid_namespace *ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001694
Alexey Dobriyane8681712007-10-26 12:17:22 +04001695static inline pid_t task_pid_nr(struct task_struct *tsk)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001696{
1697 return tsk->pid;
1698}
1699
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001700static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1701 struct pid_namespace *ns)
1702{
1703 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1704}
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001705
1706static inline pid_t task_pid_vnr(struct task_struct *tsk)
1707{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001708 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001709}
1710
1711
Alexey Dobriyane8681712007-10-26 12:17:22 +04001712static inline pid_t task_tgid_nr(struct task_struct *tsk)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001713{
1714 return tsk->tgid;
1715}
1716
Pavel Emelyanov2f2a3a42007-10-18 23:40:19 -07001717pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001718
1719static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1720{
1721 return pid_vnr(task_tgid(tsk));
1722}
1723
1724
Richard Guy Briggs80e0b6e2014-03-16 14:00:19 -04001725static inline int pid_alive(const struct task_struct *p);
Richard Guy Briggsad36d282013-08-15 18:05:12 -04001726static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1727{
1728 pid_t pid = 0;
1729
1730 rcu_read_lock();
1731 if (pid_alive(tsk))
1732 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1733 rcu_read_unlock();
1734
1735 return pid;
1736}
1737
1738static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1739{
1740 return task_ppid_nr_ns(tsk, &init_pid_ns);
1741}
1742
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001743static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1744 struct pid_namespace *ns)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001745{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001746 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001747}
1748
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001749static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1750{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001751 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001752}
1753
1754
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001755static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1756 struct pid_namespace *ns)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001757{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001758 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001759}
1760
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001761static inline pid_t task_session_vnr(struct task_struct *tsk)
1762{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001763 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001764}
1765
Oleg Nesterov1b0f7ff2009-04-02 16:58:39 -07001766/* obsolete, do not use */
1767static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1768{
1769 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1770}
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001771
Linus Torvalds1da177e2005-04-16 15:20:36 -07001772/**
1773 * pid_alive - check that a task structure is not stale
1774 * @p: Task structure to be checked.
1775 *
1776 * Test if a process is not yet dead (at most zombie state)
1777 * If pid_alive fails, then pointers within the task structure
1778 * can be stale and must not be dereferenced.
Yacine Belkadie69f6182013-07-12 20:45:47 +02001779 *
1780 * Return: 1 if the process is alive. 0 otherwise.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001781 */
Richard Guy Briggsad36d282013-08-15 18:05:12 -04001782static inline int pid_alive(const struct task_struct *p)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001783{
Eric W. Biederman92476d72006-03-31 02:31:42 -08001784 return p->pids[PIDTYPE_PID].pid != NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001785}
1786
Sukadev Bhattiproluf400e192006-09-29 02:00:07 -07001787/**
Serge E. Hallynb460cbc2007-10-18 23:39:52 -07001788 * is_global_init - check if a task structure is init
Henne32602592006-10-06 00:44:01 -07001789 * @tsk: Task structure to be checked.
1790 *
1791 * Check if a task structure is the first user space task the kernel created.
Yacine Belkadie69f6182013-07-12 20:45:47 +02001792 *
1793 * Return: 1 if the task structure is init. 0 otherwise.
Sukadev Bhattiproluf400e192006-09-29 02:00:07 -07001794 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04001795static inline int is_global_init(struct task_struct *tsk)
Pavel Emelyanovb461cc02007-10-18 23:40:09 -07001796{
1797 return tsk->pid == 1;
1798}
Serge E. Hallynb460cbc2007-10-18 23:39:52 -07001799
Cedric Le Goater9ec52092006-10-02 02:19:00 -07001800extern struct pid *cad_pid;
1801
Linus Torvalds1da177e2005-04-16 15:20:36 -07001802extern void free_task(struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001803#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
Ingo Molnare56d0902006-01-08 01:01:37 -08001804
Andrew Morton158d9eb2006-03-31 02:31:34 -08001805extern void __put_task_struct(struct task_struct *t);
Ingo Molnare56d0902006-01-08 01:01:37 -08001806
1807static inline void put_task_struct(struct task_struct *t)
1808{
1809 if (atomic_dec_and_test(&t->usage))
Eric W. Biederman8c7904a2006-03-31 02:31:37 -08001810 __put_task_struct(t);
Ingo Molnare56d0902006-01-08 01:01:37 -08001811}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001812
Frederic Weisbecker6a616712012-12-16 20:00:34 +01001813#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1814extern void task_cputime(struct task_struct *t,
1815 cputime_t *utime, cputime_t *stime);
1816extern void task_cputime_scaled(struct task_struct *t,
1817 cputime_t *utimescaled, cputime_t *stimescaled);
1818extern cputime_t task_gtime(struct task_struct *t);
1819#else
Frederic Weisbecker6fac4822012-11-13 14:20:55 +01001820static inline void task_cputime(struct task_struct *t,
1821 cputime_t *utime, cputime_t *stime)
1822{
1823 if (utime)
1824 *utime = t->utime;
1825 if (stime)
1826 *stime = t->stime;
1827}
1828
1829static inline void task_cputime_scaled(struct task_struct *t,
1830 cputime_t *utimescaled,
1831 cputime_t *stimescaled)
1832{
1833 if (utimescaled)
1834 *utimescaled = t->utimescaled;
1835 if (stimescaled)
1836 *stimescaled = t->stimescaled;
1837}
Frederic Weisbecker6a616712012-12-16 20:00:34 +01001838
1839static inline cputime_t task_gtime(struct task_struct *t)
1840{
1841 return t->gtime;
1842}
1843#endif
Frederic Weisbeckere80d0a1a2012-11-21 16:26:44 +01001844extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1845extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
Balbir Singh49048622008-09-05 18:12:23 +02001846
Linus Torvalds1da177e2005-04-16 15:20:36 -07001847/*
1848 * Per process flags
1849 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001850#define PF_EXITING 0x00000004 /* getting shut down */
Alexey Kuznetsov778e9a92007-06-08 13:47:00 -07001851#define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
Laurent Vivier94886b82007-10-15 17:00:19 +02001852#define PF_VCPU 0x00000010 /* I'm a virtual CPU */
Tejun Heo21aa9af2010-06-08 21:40:37 +02001853#define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001854#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
Andi Kleen4db96cf2009-09-16 11:50:14 +02001855#define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001856#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1857#define PF_DUMPCORE 0x00000200 /* dumped core */
1858#define PF_SIGNALED 0x00000400 /* killed by a signal */
1859#define PF_MEMALLOC 0x00000800 /* Allocating memory */
Vasiliy Kulikov72fa5992011-08-08 19:02:04 +04001860#define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001861#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
Tejun Heo774a1222013-01-15 18:52:51 -08001862#define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001863#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1864#define PF_FROZEN 0x00010000 /* frozen for system suspend */
1865#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1866#define PF_KSWAPD 0x00040000 /* I am kswapd */
Ming Lei21caf2f2013-02-22 16:34:08 -08001867#define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001868#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
Oleg Nesterov246bb0b2008-07-25 01:47:38 -07001869#define PF_KTHREAD 0x00200000 /* I am a kernel thread */
Jens Axboeb31dc662006-06-13 08:26:10 +02001870#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1871#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1872#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1873#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
Tejun Heo14a40ff2013-03-19 13:45:20 -07001874#define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
Andi Kleen4db96cf2009-09-16 11:50:14 +02001875#define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
Thomas Gleixner61a87122006-06-27 02:54:56 -07001876#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
Tejun Heo58a69cb2011-02-16 09:25:31 +01001877#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
Colin Cross2b44c4d2013-07-24 17:41:33 -07001878#define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001879
1880/*
1881 * Only the _current_ task can read/write to tsk->flags, but other
1882 * tasks can access tsk->flags in readonly mode for example
1883 * with tsk_used_math (like during threaded core dumping).
1884 * There is however an exception to this rule during ptrace
1885 * or during fork: the ptracer task is allowed to write to the
1886 * child->flags of its traced child (same goes for fork, the parent
1887 * can write to the child->flags), because we're guaranteed the
1888 * child is not running and in turn not changing child->flags
1889 * at the same time the parent does it.
1890 */
1891#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1892#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1893#define clear_used_math() clear_stopped_child_used_math(current)
1894#define set_used_math() set_stopped_child_used_math(current)
1895#define conditional_stopped_child_used_math(condition, child) \
1896 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1897#define conditional_used_math(condition) \
1898 conditional_stopped_child_used_math(condition, current)
1899#define copy_to_stopped_child_used_math(child) \
1900 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1901/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1902#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1903#define used_math() tsk_used_math(current)
1904
Ming Lei21caf2f2013-02-22 16:34:08 -08001905/* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1906static inline gfp_t memalloc_noio_flags(gfp_t flags)
1907{
1908 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1909 flags &= ~__GFP_IO;
1910 return flags;
1911}
1912
1913static inline unsigned int memalloc_noio_save(void)
1914{
1915 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1916 current->flags |= PF_MEMALLOC_NOIO;
1917 return flags;
1918}
1919
1920static inline void memalloc_noio_restore(unsigned int flags)
1921{
1922 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1923}
1924
Tejun Heoe5c19022011-03-23 10:37:00 +01001925/*
Tejun Heoa8f072c2011-06-02 11:13:59 +02001926 * task->jobctl flags
Tejun Heoe5c19022011-03-23 10:37:00 +01001927 */
Tejun Heoa8f072c2011-06-02 11:13:59 +02001928#define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
Tejun Heoe5c19022011-03-23 10:37:00 +01001929
Tejun Heoa8f072c2011-06-02 11:13:59 +02001930#define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1931#define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1932#define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
Tejun Heo73ddff22011-06-14 11:20:14 +02001933#define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
Tejun Heofb1d9102011-06-14 11:20:17 +02001934#define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
Tejun Heoa8f072c2011-06-02 11:13:59 +02001935#define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
Tejun Heo544b2c92011-06-14 11:20:18 +02001936#define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
Tejun Heoa8f072c2011-06-02 11:13:59 +02001937
1938#define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1939#define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1940#define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
Tejun Heo73ddff22011-06-14 11:20:14 +02001941#define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
Tejun Heofb1d9102011-06-14 11:20:17 +02001942#define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
Tejun Heoa8f072c2011-06-02 11:13:59 +02001943#define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
Tejun Heo544b2c92011-06-14 11:20:18 +02001944#define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
Tejun Heoa8f072c2011-06-02 11:13:59 +02001945
Tejun Heofb1d9102011-06-14 11:20:17 +02001946#define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
Tejun Heo73ddff22011-06-14 11:20:14 +02001947#define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
Tejun Heo3759a0d2011-06-02 11:14:00 +02001948
Tejun Heo7dd3db52011-06-02 11:14:00 +02001949extern bool task_set_jobctl_pending(struct task_struct *task,
1950 unsigned int mask);
Tejun Heo73ddff22011-06-14 11:20:14 +02001951extern void task_clear_jobctl_trapping(struct task_struct *task);
Tejun Heo3759a0d2011-06-02 11:14:00 +02001952extern void task_clear_jobctl_pending(struct task_struct *task,
1953 unsigned int mask);
Tejun Heo39efa3e2011-03-23 10:37:00 +01001954
Paul E. McKenneya57eb942010-06-29 16:49:16 -07001955#ifdef CONFIG_PREEMPT_RCU
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001956
1957#define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
Paul E. McKenney1aa03f12012-01-11 17:25:17 -08001958#define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001959
1960static inline void rcu_copy_process(struct task_struct *p)
1961{
1962 p->rcu_read_lock_nesting = 0;
1963 p->rcu_read_unlock_special = 0;
Paul E. McKenneya57eb942010-06-29 16:49:16 -07001964#ifdef CONFIG_TREE_PREEMPT_RCU
Paul E. McKenneydd5d19b2009-08-27 14:58:16 -07001965 p->rcu_blocked_node = NULL;
Paul E. McKenney24278d12010-09-27 17:25:23 -07001966#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1967#ifdef CONFIG_RCU_BOOST
1968 p->rcu_boost_mutex = NULL;
1969#endif /* #ifdef CONFIG_RCU_BOOST */
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001970 INIT_LIST_HEAD(&p->rcu_node_entry);
1971}
1972
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001973#else
1974
1975static inline void rcu_copy_process(struct task_struct *p)
1976{
1977}
1978
1979#endif
1980
Mel Gorman907aed42012-07-31 16:44:07 -07001981static inline void tsk_restore_flags(struct task_struct *task,
1982 unsigned long orig_flags, unsigned long flags)
1983{
1984 task->flags &= ~flags;
1985 task->flags |= orig_flags & flags;
1986}
1987
Linus Torvalds1da177e2005-04-16 15:20:36 -07001988#ifdef CONFIG_SMP
KOSAKI Motohiro1e1b6c52011-05-19 15:08:58 +09001989extern void do_set_cpus_allowed(struct task_struct *p,
1990 const struct cpumask *new_mask);
1991
Mike Traviscd8ba7c2008-03-26 14:23:49 -07001992extern int set_cpus_allowed_ptr(struct task_struct *p,
Rusty Russell96f874e2008-11-25 02:35:14 +10301993 const struct cpumask *new_mask);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001994#else
KOSAKI Motohiro1e1b6c52011-05-19 15:08:58 +09001995static inline void do_set_cpus_allowed(struct task_struct *p,
1996 const struct cpumask *new_mask)
1997{
1998}
Mike Traviscd8ba7c2008-03-26 14:23:49 -07001999static inline int set_cpus_allowed_ptr(struct task_struct *p,
Rusty Russell96f874e2008-11-25 02:35:14 +10302000 const struct cpumask *new_mask)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002001{
Rusty Russell96f874e2008-11-25 02:35:14 +10302002 if (!cpumask_test_cpu(0, new_mask))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002003 return -EINVAL;
2004 return 0;
2005}
2006#endif
Rusty Russelle0ad9552009-09-24 09:34:38 -06002007
Frederic Weisbecker3451d022011-08-10 23:21:01 +02002008#ifdef CONFIG_NO_HZ_COMMON
Peter Zijlstra5167e8d2012-06-22 15:52:09 +02002009void calc_load_enter_idle(void);
2010void calc_load_exit_idle(void);
2011#else
2012static inline void calc_load_enter_idle(void) { }
2013static inline void calc_load_exit_idle(void) { }
Frederic Weisbecker3451d022011-08-10 23:21:01 +02002014#endif /* CONFIG_NO_HZ_COMMON */
Peter Zijlstra5167e8d2012-06-22 15:52:09 +02002015
Rusty Russelle0ad9552009-09-24 09:34:38 -06002016#ifndef CONFIG_CPUMASK_OFFSTACK
Mike Traviscd8ba7c2008-03-26 14:23:49 -07002017static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2018{
2019 return set_cpus_allowed_ptr(p, &new_mask);
2020}
Rusty Russelle0ad9552009-09-24 09:34:38 -06002021#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002022
Ingo Molnarb3425012009-02-26 20:20:29 +01002023/*
Peter Zijlstrac6763292010-05-25 10:48:51 +02002024 * Do not use outside of architecture code which knows its limitations.
2025 *
2026 * sched_clock() has no promise of monotonicity or bounded drift between
2027 * CPUs, use (which you should not) requires disabling IRQs.
2028 *
2029 * Please use one of the three interfaces below.
Ingo Molnarb3425012009-02-26 20:20:29 +01002030 */
Mike Frysinger1bbfa6f2009-12-09 20:07:03 -05002031extern unsigned long long notrace sched_clock(void);
Peter Zijlstrac6763292010-05-25 10:48:51 +02002032/*
Hiroshi Shimamoto489a71b2012-04-02 17:00:44 +09002033 * See the comment in kernel/sched/clock.c
Peter Zijlstrac6763292010-05-25 10:48:51 +02002034 */
2035extern u64 cpu_clock(int cpu);
2036extern u64 local_clock(void);
2037extern u64 sched_clock_cpu(int cpu);
2038
Ingo Molnare436d802007-07-19 21:28:35 +02002039
Peter Zijlstrac1955a32008-08-11 08:59:03 +02002040extern void sched_clock_init(void);
Peter Zijlstrac1955a32008-08-11 08:59:03 +02002041
Peter Zijlstra3e51f332008-05-03 18:29:28 +02002042#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
Peter Zijlstra3e51f332008-05-03 18:29:28 +02002043static inline void sched_clock_tick(void)
2044{
2045}
2046
2047static inline void sched_clock_idle_sleep_event(void)
2048{
2049}
2050
2051static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2052{
2053}
2054#else
Peter Zijlstrac6763292010-05-25 10:48:51 +02002055/*
2056 * Architectures can set this to 1 if they have specified
2057 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2058 * but then during bootup it turns out that sched_clock()
2059 * is reliable after all:
2060 */
Peter Zijlstra35af99e2013-11-28 19:38:42 +01002061extern int sched_clock_stable(void);
2062extern void set_sched_clock_stable(void);
2063extern void clear_sched_clock_stable(void);
Peter Zijlstrac6763292010-05-25 10:48:51 +02002064
Peter Zijlstra3e51f332008-05-03 18:29:28 +02002065extern void sched_clock_tick(void);
2066extern void sched_clock_idle_sleep_event(void);
2067extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2068#endif
2069
Venkatesh Pallipadib52bfee2010-10-04 17:03:19 -07002070#ifdef CONFIG_IRQ_TIME_ACCOUNTING
2071/*
2072 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2073 * The reason for this explicit opt-in is not to have perf penalty with
2074 * slow sched_clocks.
2075 */
2076extern void enable_sched_clock_irqtime(void);
2077extern void disable_sched_clock_irqtime(void);
2078#else
2079static inline void enable_sched_clock_irqtime(void) {}
2080static inline void disable_sched_clock_irqtime(void) {}
2081#endif
2082
Ingo Molnar36c8b582006-07-03 00:25:41 -07002083extern unsigned long long
Ingo Molnar41b86e92007-07-09 18:51:58 +02002084task_sched_runtime(struct task_struct *task);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002085
2086/* sched_exec is called by processes performing an exec */
2087#ifdef CONFIG_SMP
2088extern void sched_exec(void);
2089#else
2090#define sched_exec() {}
2091#endif
2092
Ingo Molnar2aa44d02007-08-23 15:18:02 +02002093extern void sched_clock_idle_sleep_event(void);
2094extern void sched_clock_idle_wakeup_event(u64 delta_ns);
Ingo Molnarbb29ab22007-07-09 18:51:59 +02002095
Linus Torvalds1da177e2005-04-16 15:20:36 -07002096#ifdef CONFIG_HOTPLUG_CPU
2097extern void idle_task_exit(void);
2098#else
2099static inline void idle_task_exit(void) {}
2100#endif
2101
Frederic Weisbecker3451d022011-08-10 23:21:01 +02002102#if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
Frederic Weisbecker1c200912011-08-10 23:21:01 +02002103extern void wake_up_nohz_cpu(int cpu);
Thomas Gleixner06d83082008-03-22 09:20:24 +01002104#else
Frederic Weisbecker1c200912011-08-10 23:21:01 +02002105static inline void wake_up_nohz_cpu(int cpu) { }
Thomas Gleixner06d83082008-03-22 09:20:24 +01002106#endif
2107
Frederic Weisbeckerce831b32013-04-20 15:15:35 +02002108#ifdef CONFIG_NO_HZ_FULL
2109extern bool sched_can_stop_tick(void);
Frederic Weisbecker265f22a2013-05-03 03:39:05 +02002110extern u64 scheduler_tick_max_deferment(void);
Frederic Weisbeckerce831b32013-04-20 15:15:35 +02002111#else
2112static inline bool sched_can_stop_tick(void) { return false; }
Ingo Molnarbf0f6f22007-07-09 18:51:58 +02002113#endif
Ingo Molnarbf0f6f22007-07-09 18:51:58 +02002114
Mike Galbraith5091faa2010-11-30 14:18:03 +01002115#ifdef CONFIG_SCHED_AUTOGROUP
Mike Galbraith5091faa2010-11-30 14:18:03 +01002116extern void sched_autogroup_create_attach(struct task_struct *p);
2117extern void sched_autogroup_detach(struct task_struct *p);
2118extern void sched_autogroup_fork(struct signal_struct *sig);
2119extern void sched_autogroup_exit(struct signal_struct *sig);
2120#ifdef CONFIG_PROC_FS
2121extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
Hiroshi Shimamoto2e5b5b32012-02-23 17:41:27 +09002122extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
Mike Galbraith5091faa2010-11-30 14:18:03 +01002123#endif
2124#else
2125static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2126static inline void sched_autogroup_detach(struct task_struct *p) { }
2127static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2128static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2129#endif
2130
Mike Galbraithd95f4122011-02-01 09:50:51 -05002131extern bool yield_to(struct task_struct *p, bool preempt);
Ingo Molnar36c8b582006-07-03 00:25:41 -07002132extern void set_user_nice(struct task_struct *p, long nice);
2133extern int task_prio(const struct task_struct *p);
Dongsheng Yangd0ea0262014-01-27 22:00:45 -05002134/**
2135 * task_nice - return the nice value of a given task.
2136 * @p: the task in question.
2137 *
2138 * Return: The nice value [ -20 ... 0 ... 19 ].
2139 */
2140static inline int task_nice(const struct task_struct *p)
2141{
2142 return PRIO_TO_NICE((p)->static_prio);
2143}
Ingo Molnar36c8b582006-07-03 00:25:41 -07002144extern int can_nice(const struct task_struct *p, const int nice);
2145extern int task_curr(const struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002146extern int idle_cpu(int cpu);
KOSAKI Motohirofe7de492010-10-20 16:01:12 -07002147extern int sched_setscheduler(struct task_struct *, int,
2148 const struct sched_param *);
Rusty Russell961ccdd2008-06-23 13:55:38 +10002149extern int sched_setscheduler_nocheck(struct task_struct *, int,
KOSAKI Motohirofe7de492010-10-20 16:01:12 -07002150 const struct sched_param *);
Dario Faggiolid50dde52013-11-07 14:43:36 +01002151extern int sched_setattr(struct task_struct *,
2152 const struct sched_attr *);
Ingo Molnar36c8b582006-07-03 00:25:41 -07002153extern struct task_struct *idle_task(int cpu);
Paul E. McKenneyc4f30602011-11-10 12:41:56 -08002154/**
2155 * is_idle_task - is the specified task an idle task?
Randy Dunlapfa757282012-01-21 11:03:13 -08002156 * @p: the task in question.
Yacine Belkadie69f6182013-07-12 20:45:47 +02002157 *
2158 * Return: 1 if @p is an idle task. 0 otherwise.
Paul E. McKenneyc4f30602011-11-10 12:41:56 -08002159 */
Paul E. McKenney7061ca32011-12-20 08:20:46 -08002160static inline bool is_idle_task(const struct task_struct *p)
Paul E. McKenneyc4f30602011-11-10 12:41:56 -08002161{
2162 return p->pid == 0;
2163}
Ingo Molnar36c8b582006-07-03 00:25:41 -07002164extern struct task_struct *curr_task(int cpu);
2165extern void set_curr_task(int cpu, struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002166
2167void yield(void);
2168
2169/*
2170 * The default (Linux) execution domain.
2171 */
2172extern struct exec_domain default_exec_domain;
2173
2174union thread_union {
2175 struct thread_info thread_info;
2176 unsigned long stack[THREAD_SIZE/sizeof(long)];
2177};
2178
2179#ifndef __HAVE_ARCH_KSTACK_END
2180static inline int kstack_end(void *addr)
2181{
2182 /* Reliable end of stack detection:
2183 * Some APM bios versions misalign the stack
2184 */
2185 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2186}
2187#endif
2188
2189extern union thread_union init_thread_union;
2190extern struct task_struct init_task;
2191
2192extern struct mm_struct init_mm;
2193
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002194extern struct pid_namespace init_pid_ns;
2195
2196/*
2197 * find a task by one of its numerical ids
2198 *
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002199 * find_task_by_pid_ns():
2200 * finds a task by its pid in the specified namespace
Pavel Emelyanov228ebcb2007-10-18 23:40:16 -07002201 * find_task_by_vpid():
2202 * finds a task by its virtual pid
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002203 *
Pavel Emelyanove49859e2008-07-25 01:48:36 -07002204 * see also find_vpid() etc in include/linux/pid.h
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002205 */
2206
Pavel Emelyanov228ebcb2007-10-18 23:40:16 -07002207extern struct task_struct *find_task_by_vpid(pid_t nr);
2208extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2209 struct pid_namespace *ns);
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002210
Linus Torvalds1da177e2005-04-16 15:20:36 -07002211/* per-UID process charging. */
Eric W. Biederman7b44ab92011-11-16 23:20:58 -08002212extern struct user_struct * alloc_uid(kuid_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002213static inline struct user_struct *get_uid(struct user_struct *u)
2214{
2215 atomic_inc(&u->__count);
2216 return u;
2217}
2218extern void free_uid(struct user_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002219
2220#include <asm/current.h>
2221
Torben Hohnf0af911a92011-01-27 15:59:10 +01002222extern void xtime_update(unsigned long ticks);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002223
Harvey Harrisonb3c97522008-02-13 15:03:15 -08002224extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2225extern int wake_up_process(struct task_struct *tsk);
Samir Bellabes3e51e3e2011-05-11 18:18:05 +02002226extern void wake_up_new_task(struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002227#ifdef CONFIG_SMP
2228 extern void kick_process(struct task_struct *tsk);
2229#else
2230 static inline void kick_process(struct task_struct *tsk) { }
2231#endif
Dario Faggioliaab03e02013-11-28 11:14:43 +01002232extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
Ingo Molnarad46c2c2007-07-09 18:52:00 +02002233extern void sched_dead(struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002234
Linus Torvalds1da177e2005-04-16 15:20:36 -07002235extern void proc_caches_init(void);
2236extern void flush_signals(struct task_struct *);
David Howells3bcac022009-04-29 13:45:05 +01002237extern void __flush_signals(struct task_struct *);
Oleg Nesterov10ab8252007-05-09 02:34:37 -07002238extern void ignore_signals(struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002239extern void flush_signal_handlers(struct task_struct *, int force_default);
2240extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2241
2242static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2243{
2244 unsigned long flags;
2245 int ret;
2246
2247 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2248 ret = dequeue_signal(tsk, mask, info);
2249 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2250
2251 return ret;
Oleg Nesterov53c8f9f2011-06-22 23:08:18 +02002252}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002253
2254extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2255 sigset_t *mask);
2256extern void unblock_all_signals(void);
2257extern void release_task(struct task_struct * p);
2258extern int send_sig_info(int, struct siginfo *, struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002259extern int force_sigsegv(int, struct task_struct *);
2260extern int force_sig_info(int, struct siginfo *, struct task_struct *);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002261extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002262extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
Serge Hallynd178bc32011-09-26 10:45:18 -05002263extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2264 const struct cred *, u32);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002265extern int kill_pgrp(struct pid *pid, int sig, int priv);
2266extern int kill_pid(struct pid *pid, int sig, int priv);
Matthew Wilcoxc3de4b32007-02-09 08:11:47 -07002267extern int kill_proc_info(int, struct siginfo *, pid_t);
Oleg Nesterov86773472011-06-22 23:09:09 +02002268extern __must_check bool do_notify_parent(struct task_struct *, int);
Oleg Nesterova7f07652009-09-23 15:56:44 -07002269extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002270extern void force_sig(int, struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002271extern int send_sig(int, struct task_struct *, int);
Oleg Nesterov09faef12010-05-26 14:43:11 -07002272extern int zap_other_threads(struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002273extern struct sigqueue *sigqueue_alloc(void);
2274extern void sigqueue_free(struct sigqueue *);
Oleg Nesterovac5c2152008-04-30 00:52:57 -07002275extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
Oleg Nesterov9ac95f22006-02-09 22:41:50 +03002276extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002277
Al Viro51a7b442012-05-21 23:33:55 -04002278static inline void restore_saved_sigmask(void)
2279{
2280 if (test_and_clear_restore_sigmask())
Al Viro77097ae2012-04-27 13:58:59 -04002281 __set_current_blocked(&current->saved_sigmask);
Al Viro51a7b442012-05-21 23:33:55 -04002282}
2283
Al Virob7f9a112012-05-02 09:59:21 -04002284static inline sigset_t *sigmask_to_save(void)
2285{
2286 sigset_t *res = &current->blocked;
2287 if (unlikely(test_restore_sigmask()))
2288 res = &current->saved_sigmask;
2289 return res;
2290}
2291
Cedric Le Goater9ec52092006-10-02 02:19:00 -07002292static inline int kill_cad_pid(int sig, int priv)
2293{
2294 return kill_pid(cad_pid, sig, priv);
2295}
2296
Linus Torvalds1da177e2005-04-16 15:20:36 -07002297/* These can be the second arg to send_sig_info/send_group_sig_info. */
2298#define SEND_SIG_NOINFO ((struct siginfo *) 0)
2299#define SEND_SIG_PRIV ((struct siginfo *) 1)
2300#define SEND_SIG_FORCED ((struct siginfo *) 2)
2301
Sebastian Andrzej Siewior2a855dd2009-10-25 15:37:58 +01002302/*
2303 * True if we are on the alternate signal stack.
2304 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002305static inline int on_sig_stack(unsigned long sp)
2306{
Sebastian Andrzej Siewior2a855dd2009-10-25 15:37:58 +01002307#ifdef CONFIG_STACK_GROWSUP
2308 return sp >= current->sas_ss_sp &&
2309 sp - current->sas_ss_sp < current->sas_ss_size;
2310#else
2311 return sp > current->sas_ss_sp &&
2312 sp - current->sas_ss_sp <= current->sas_ss_size;
2313#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002314}
2315
2316static inline int sas_ss_flags(unsigned long sp)
2317{
2318 return (current->sas_ss_size == 0 ? SS_DISABLE
2319 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2320}
2321
Al Viro5a1b98d2012-11-06 13:28:21 -05002322static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2323{
2324 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2325#ifdef CONFIG_STACK_GROWSUP
2326 return current->sas_ss_sp;
2327#else
2328 return current->sas_ss_sp + current->sas_ss_size;
2329#endif
2330 return sp;
2331}
2332
Linus Torvalds1da177e2005-04-16 15:20:36 -07002333/*
2334 * Routines for handling mm_structs
2335 */
2336extern struct mm_struct * mm_alloc(void);
2337
2338/* mmdrop drops the mm and the page tables */
Harvey Harrisonb3c97522008-02-13 15:03:15 -08002339extern void __mmdrop(struct mm_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002340static inline void mmdrop(struct mm_struct * mm)
2341{
Ingo Molnar6fb43d72007-07-09 18:52:01 +02002342 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002343 __mmdrop(mm);
2344}
2345
2346/* mmput gets rid of the mappings and all user-space */
2347extern void mmput(struct mm_struct *);
2348/* Grab a reference to a task's mm, if it is not already going away */
2349extern struct mm_struct *get_task_mm(struct task_struct *task);
Christopher Yeoh8cdb8782012-02-02 11:34:09 +10302350/*
2351 * Grab a reference to a task's mm, if it is not already going away
2352 * and ptrace_may_access with the mode parameter passed to it
2353 * succeeds.
2354 */
2355extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002356/* Remove the current tasks stale references to the old mm_struct */
2357extern void mm_release(struct task_struct *, struct mm_struct *);
2358
Alexey Dobriyan6f2c55b2009-04-02 16:56:59 -07002359extern int copy_thread(unsigned long, unsigned long, unsigned long,
Al Viroafa86fc2012-10-22 22:51:14 -04002360 struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002361extern void flush_thread(void);
2362extern void exit_thread(void);
2363
Linus Torvalds1da177e2005-04-16 15:20:36 -07002364extern void exit_files(struct task_struct *);
Oleg Nesterova7e53282006-03-28 16:11:27 -08002365extern void __cleanup_sighand(struct sighand_struct *);
Oleg Nesterovcbaffba2008-05-26 20:55:42 +04002366
Linus Torvalds1da177e2005-04-16 15:20:36 -07002367extern void exit_itimers(struct signal_struct *);
Oleg Nesterovcbaffba2008-05-26 20:55:42 +04002368extern void flush_itimer_signals(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002369
Joe Perches9402c952012-01-12 17:17:17 -08002370extern void do_group_exit(int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002371
Linus Torvalds1da177e2005-04-16 15:20:36 -07002372extern int allow_signal(int);
2373extern int disallow_signal(int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002374
Linus Torvaldsc4ad8f92014-02-05 12:54:53 -08002375extern int do_execve(struct filename *,
David Howellsd7627462010-08-17 23:52:56 +01002376 const char __user * const __user *,
Al Viroda3d4c52012-10-20 21:49:33 -04002377 const char __user * const __user *);
Al Viroe80d6662012-10-22 23:10:08 -04002378extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
Ingo Molnar36c8b582006-07-03 00:25:41 -07002379struct task_struct *fork_idle(int);
Al Viro2aa3a7f2012-09-21 19:55:31 -04002380extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002381
Adrian Hunter82b89772014-05-28 11:45:04 +03002382extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2383static inline void set_task_comm(struct task_struct *tsk, const char *from)
2384{
2385 __set_task_comm(tsk, from, false);
2386}
Andrew Morton59714d62008-02-04 22:27:21 -08002387extern char *get_task_comm(char *to, struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002388
2389#ifdef CONFIG_SMP
Peter Zijlstra317f3942011-04-05 17:23:58 +02002390void scheduler_ipi(void);
Roland McGrath85ba2d82008-07-25 19:45:58 -07002391extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002392#else
Peter Zijlstra184748c2011-04-05 17:23:39 +02002393static inline void scheduler_ipi(void) { }
Roland McGrath85ba2d82008-07-25 19:45:58 -07002394static inline unsigned long wait_task_inactive(struct task_struct *p,
2395 long match_state)
2396{
2397 return 1;
2398}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002399#endif
2400
Jiri Pirko05725f72009-04-14 20:17:16 +02002401#define next_task(p) \
2402 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002403
2404#define for_each_process(p) \
2405 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2406
Oleg Nesterov5bb459b2009-07-10 03:48:23 +02002407extern bool current_is_single_threaded(void);
David Howellsd84f4f92008-11-14 10:39:23 +11002408
Linus Torvalds1da177e2005-04-16 15:20:36 -07002409/*
2410 * Careful: do_each_thread/while_each_thread is a double loop so
2411 * 'break' will not work as expected - use goto instead.
2412 */
2413#define do_each_thread(g, t) \
2414 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2415
2416#define while_each_thread(g, t) \
2417 while ((t = next_thread(t)) != g)
2418
Oleg Nesterov0c740d02014-01-21 15:49:56 -08002419#define __for_each_thread(signal, t) \
2420 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2421
2422#define for_each_thread(p, t) \
2423 __for_each_thread((p)->signal, t)
2424
2425/* Careful: this is a double loop, 'break' won't work as expected. */
2426#define for_each_process_thread(p, t) \
2427 for_each_process(p) for_each_thread(p, t)
2428
Oleg Nesterov7e498272010-05-26 14:43:22 -07002429static inline int get_nr_threads(struct task_struct *tsk)
2430{
Oleg Nesterovb3ac0222010-05-26 14:43:24 -07002431 return tsk->signal->nr_threads;
Oleg Nesterov7e498272010-05-26 14:43:22 -07002432}
2433
Oleg Nesterov087806b2011-06-22 23:10:26 +02002434static inline bool thread_group_leader(struct task_struct *p)
2435{
2436 return p->exit_signal >= 0;
2437}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002438
Eric W. Biederman0804ef42006-10-02 02:17:04 -07002439/* Do to the insanities of de_thread it is possible for a process
2440 * to have the pid of the thread group leader without actually being
2441 * the thread group leader. For iteration through the pids in proc
2442 * all we care about is that we have a task with the appropriate
2443 * pid, we don't actually care if we have the right task.
2444 */
Oleg Nesterove1403b82013-09-11 14:20:06 -07002445static inline bool has_group_leader_pid(struct task_struct *p)
Eric W. Biederman0804ef42006-10-02 02:17:04 -07002446{
Oleg Nesterove1403b82013-09-11 14:20:06 -07002447 return task_pid(p) == p->signal->leader_pid;
Eric W. Biederman0804ef42006-10-02 02:17:04 -07002448}
2449
Pavel Emelyanovbac0abd2007-10-18 23:40:18 -07002450static inline
Oleg Nesterove1403b82013-09-11 14:20:06 -07002451bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
Pavel Emelyanovbac0abd2007-10-18 23:40:18 -07002452{
Oleg Nesterove1403b82013-09-11 14:20:06 -07002453 return p1->signal == p2->signal;
Pavel Emelyanovbac0abd2007-10-18 23:40:18 -07002454}
2455
Ingo Molnar36c8b582006-07-03 00:25:41 -07002456static inline struct task_struct *next_thread(const struct task_struct *p)
Oleg Nesterov47e65322006-03-28 16:11:25 -08002457{
Jiri Pirko05725f72009-04-14 20:17:16 +02002458 return list_entry_rcu(p->thread_group.next,
2459 struct task_struct, thread_group);
Oleg Nesterov47e65322006-03-28 16:11:25 -08002460}
2461
Alexey Dobriyane8681712007-10-26 12:17:22 +04002462static inline int thread_group_empty(struct task_struct *p)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002463{
Oleg Nesterov47e65322006-03-28 16:11:25 -08002464 return list_empty(&p->thread_group);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002465}
2466
2467#define delay_group_leader(p) \
2468 (thread_group_leader(p) && !thread_group_empty(p))
2469
Linus Torvalds1da177e2005-04-16 15:20:36 -07002470/*
Eric W. Biederman260ea102006-06-23 02:05:18 -07002471 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
Jens Axboe22e2c502005-06-27 10:55:12 +02002472 * subscriptions and synchronises with wait4(). Also used in procfs. Also
Paul Menageddbcc7e2007-10-18 23:39:30 -07002473 * pins the final release of task.io_context. Also protects ->cpuset and
Oleg Nesterovd68b46f2012-03-05 14:59:13 -08002474 * ->cgroup.subsys[]. And ->vfork_done.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002475 *
2476 * Nests both inside and outside of read_lock(&tasklist_lock).
2477 * It must not be nested with write_lock_irq(&tasklist_lock),
2478 * neither inside nor outside.
2479 */
2480static inline void task_lock(struct task_struct *p)
2481{
2482 spin_lock(&p->alloc_lock);
2483}
2484
2485static inline void task_unlock(struct task_struct *p)
2486{
2487 spin_unlock(&p->alloc_lock);
2488}
2489
Namhyung Kimb8ed3742010-10-27 15:34:06 -07002490extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
Oleg Nesterovf63ee722006-03-28 16:11:13 -08002491 unsigned long *flags);
2492
Anton Vorontsov9388dc32012-02-09 20:45:19 +04002493static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2494 unsigned long *flags)
2495{
2496 struct sighand_struct *ret;
2497
2498 ret = __lock_task_sighand(tsk, flags);
2499 (void)__cond_lock(&tsk->sighand->siglock, ret);
2500 return ret;
2501}
Namhyung Kimb8ed3742010-10-27 15:34:06 -07002502
Oleg Nesterovf63ee722006-03-28 16:11:13 -08002503static inline void unlock_task_sighand(struct task_struct *tsk,
2504 unsigned long *flags)
2505{
2506 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2507}
2508
Ben Blum4714d1d2011-05-26 16:25:18 -07002509#ifdef CONFIG_CGROUPS
Tejun Heo257058a2011-12-12 18:12:21 -08002510static inline void threadgroup_change_begin(struct task_struct *tsk)
Ben Blum4714d1d2011-05-26 16:25:18 -07002511{
Tejun Heo257058a2011-12-12 18:12:21 -08002512 down_read(&tsk->signal->group_rwsem);
Ben Blum4714d1d2011-05-26 16:25:18 -07002513}
Tejun Heo257058a2011-12-12 18:12:21 -08002514static inline void threadgroup_change_end(struct task_struct *tsk)
Ben Blum4714d1d2011-05-26 16:25:18 -07002515{
Tejun Heo257058a2011-12-12 18:12:21 -08002516 up_read(&tsk->signal->group_rwsem);
Ben Blum4714d1d2011-05-26 16:25:18 -07002517}
Tejun Heo77e4ef92011-12-12 18:12:21 -08002518
2519/**
2520 * threadgroup_lock - lock threadgroup
2521 * @tsk: member task of the threadgroup to lock
2522 *
2523 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2524 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
Oleg Nesterove56fb282013-04-30 15:28:20 -07002525 * change ->group_leader/pid. This is useful for cases where the threadgroup
2526 * needs to stay stable across blockable operations.
Tejun Heo77e4ef92011-12-12 18:12:21 -08002527 *
2528 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2529 * synchronization. While held, no new task will be added to threadgroup
2530 * and no existing live task will have its PF_EXITING set.
2531 *
Oleg Nesterove56fb282013-04-30 15:28:20 -07002532 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2533 * sub-thread becomes a new leader.
Tejun Heo77e4ef92011-12-12 18:12:21 -08002534 */
Tejun Heo257058a2011-12-12 18:12:21 -08002535static inline void threadgroup_lock(struct task_struct *tsk)
Ben Blum4714d1d2011-05-26 16:25:18 -07002536{
Tejun Heo257058a2011-12-12 18:12:21 -08002537 down_write(&tsk->signal->group_rwsem);
Ben Blum4714d1d2011-05-26 16:25:18 -07002538}
Tejun Heo77e4ef92011-12-12 18:12:21 -08002539
2540/**
2541 * threadgroup_unlock - unlock threadgroup
2542 * @tsk: member task of the threadgroup to unlock
2543 *
2544 * Reverse threadgroup_lock().
2545 */
Tejun Heo257058a2011-12-12 18:12:21 -08002546static inline void threadgroup_unlock(struct task_struct *tsk)
Ben Blum4714d1d2011-05-26 16:25:18 -07002547{
Tejun Heo257058a2011-12-12 18:12:21 -08002548 up_write(&tsk->signal->group_rwsem);
Ben Blum4714d1d2011-05-26 16:25:18 -07002549}
2550#else
Tejun Heo257058a2011-12-12 18:12:21 -08002551static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2552static inline void threadgroup_change_end(struct task_struct *tsk) {}
2553static inline void threadgroup_lock(struct task_struct *tsk) {}
2554static inline void threadgroup_unlock(struct task_struct *tsk) {}
Ben Blum4714d1d2011-05-26 16:25:18 -07002555#endif
2556
Al Virof0373602005-11-13 16:06:57 -08002557#ifndef __HAVE_THREAD_FUNCTIONS
2558
Roman Zippelf7e42172007-05-09 02:35:17 -07002559#define task_thread_info(task) ((struct thread_info *)(task)->stack)
2560#define task_stack_page(task) ((task)->stack)
Al Viroa1261f52005-11-13 16:06:55 -08002561
Al Viro10ebffd2005-11-13 16:06:56 -08002562static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2563{
2564 *task_thread_info(p) = *task_thread_info(org);
2565 task_thread_info(p)->task = p;
2566}
2567
2568static inline unsigned long *end_of_stack(struct task_struct *p)
2569{
Roman Zippelf7e42172007-05-09 02:35:17 -07002570 return (unsigned long *)(task_thread_info(p) + 1);
Al Viro10ebffd2005-11-13 16:06:56 -08002571}
2572
Al Virof0373602005-11-13 16:06:57 -08002573#endif
2574
FUJITA Tomonori8b05c7e2008-07-23 21:26:53 -07002575static inline int object_is_on_stack(void *obj)
2576{
2577 void *stack = task_stack_page(current);
2578
2579 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2580}
2581
Benjamin Herrenschmidt8c9843e2008-04-18 16:56:15 +10002582extern void thread_info_cache_init(void);
2583
Eric Sandeen7c9f8862008-04-22 16:38:23 -05002584#ifdef CONFIG_DEBUG_STACK_USAGE
2585static inline unsigned long stack_not_used(struct task_struct *p)
2586{
2587 unsigned long *n = end_of_stack(p);
2588
2589 do { /* Skip over canary */
2590 n++;
2591 } while (!*n);
2592
2593 return (unsigned long)n - (unsigned long)end_of_stack(p);
2594}
2595#endif
2596
Linus Torvalds1da177e2005-04-16 15:20:36 -07002597/* set thread flags in other task's structures
2598 * - see asm/thread_info.h for TIF_xxxx flags available
2599 */
2600static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2601{
Al Viroa1261f52005-11-13 16:06:55 -08002602 set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002603}
2604
2605static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2606{
Al Viroa1261f52005-11-13 16:06:55 -08002607 clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002608}
2609
2610static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2611{
Al Viroa1261f52005-11-13 16:06:55 -08002612 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002613}
2614
2615static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2616{
Al Viroa1261f52005-11-13 16:06:55 -08002617 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002618}
2619
2620static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2621{
Al Viroa1261f52005-11-13 16:06:55 -08002622 return test_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002623}
2624
2625static inline void set_tsk_need_resched(struct task_struct *tsk)
2626{
2627 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2628}
2629
2630static inline void clear_tsk_need_resched(struct task_struct *tsk)
2631{
2632 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2633}
2634
Gregory Haskins8ae121a2008-04-23 07:13:29 -04002635static inline int test_tsk_need_resched(struct task_struct *tsk)
2636{
2637 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2638}
2639
Eric W. Biederman690cc3f2009-05-13 16:55:10 +00002640static inline int restart_syscall(void)
2641{
2642 set_tsk_thread_flag(current, TIF_SIGPENDING);
2643 return -ERESTARTNOINTR;
2644}
2645
Linus Torvalds1da177e2005-04-16 15:20:36 -07002646static inline int signal_pending(struct task_struct *p)
2647{
2648 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2649}
Matthew Wilcoxf776d122007-12-06 11:15:50 -05002650
Roland McGrathd9588722009-09-23 15:57:04 -07002651static inline int __fatal_signal_pending(struct task_struct *p)
2652{
2653 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2654}
Matthew Wilcoxf776d122007-12-06 11:15:50 -05002655
2656static inline int fatal_signal_pending(struct task_struct *p)
2657{
2658 return signal_pending(p) && __fatal_signal_pending(p);
2659}
2660
Oleg Nesterov16882c12008-06-08 21:20:41 +04002661static inline int signal_pending_state(long state, struct task_struct *p)
2662{
2663 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2664 return 0;
2665 if (!signal_pending(p))
2666 return 0;
2667
Oleg Nesterov16882c12008-06-08 21:20:41 +04002668 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2669}
2670
Linus Torvalds1da177e2005-04-16 15:20:36 -07002671/*
2672 * cond_resched() and cond_resched_lock(): latency reduction via
2673 * explicit rescheduling in places that are safe. The return
2674 * value indicates whether a reschedule was done in fact.
2675 * cond_resched_lock() will drop the spinlock before scheduling,
2676 * cond_resched_softirq() will enable bhs before scheduling.
2677 */
Linus Torvaldsc3921ab2008-05-11 16:04:48 -07002678extern int _cond_resched(void);
Frederic Weisbecker6f80bd92009-07-16 15:44:29 +02002679
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002680#define cond_resched() ({ \
2681 __might_sleep(__FILE__, __LINE__, 0); \
2682 _cond_resched(); \
2683})
Frederic Weisbecker6f80bd92009-07-16 15:44:29 +02002684
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002685extern int __cond_resched_lock(spinlock_t *lock);
2686
Frederic Weisbeckerbdd4e852011-06-08 01:13:27 +02002687#ifdef CONFIG_PREEMPT_COUNT
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002688#define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
Herbert Xu02b67cc32008-01-25 21:08:28 +01002689#else
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002690#define PREEMPT_LOCK_OFFSET 0
Herbert Xu02b67cc32008-01-25 21:08:28 +01002691#endif
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002692
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002693#define cond_resched_lock(lock) ({ \
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002694 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002695 __cond_resched_lock(lock); \
2696})
2697
2698extern int __cond_resched_softirq(void);
2699
Venkatesh Pallipadi75e10562010-10-04 17:03:16 -07002700#define cond_resched_softirq() ({ \
2701 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2702 __cond_resched_softirq(); \
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002703})
Linus Torvalds1da177e2005-04-16 15:20:36 -07002704
Simon Hormanf6f3c432013-05-22 14:50:31 +09002705static inline void cond_resched_rcu(void)
2706{
2707#if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2708 rcu_read_unlock();
2709 cond_resched();
2710 rcu_read_lock();
2711#endif
2712}
2713
Linus Torvalds1da177e2005-04-16 15:20:36 -07002714/*
2715 * Does a critical section need to be broken due to another
Nick Piggin95c354f2008-01-30 13:31:20 +01002716 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2717 * but a general need for low latency)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002718 */
Nick Piggin95c354f2008-01-30 13:31:20 +01002719static inline int spin_needbreak(spinlock_t *lock)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002720{
Nick Piggin95c354f2008-01-30 13:31:20 +01002721#ifdef CONFIG_PREEMPT
2722 return spin_is_contended(lock);
2723#else
Linus Torvalds1da177e2005-04-16 15:20:36 -07002724 return 0;
Nick Piggin95c354f2008-01-30 13:31:20 +01002725#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002726}
2727
Roland McGrath7bb44ad2007-05-23 13:57:44 -07002728/*
Thomas Gleixneree761f62013-03-21 22:49:32 +01002729 * Idle thread specific functions to determine the need_resched
2730 * polling state. We have two versions, one based on TS_POLLING in
2731 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2732 * thread_info.flags
2733 */
2734#ifdef TS_POLLING
2735static inline int tsk_is_polling(struct task_struct *p)
2736{
2737 return task_thread_info(p)->status & TS_POLLING;
2738}
Peter Zijlstraea811742013-09-11 12:43:13 +02002739static inline void __current_set_polling(void)
Thomas Gleixner3a98f872013-03-21 22:49:33 +01002740{
2741 current_thread_info()->status |= TS_POLLING;
2742}
2743
Peter Zijlstraea811742013-09-11 12:43:13 +02002744static inline bool __must_check current_set_polling_and_test(void)
2745{
2746 __current_set_polling();
2747
2748 /*
2749 * Polling state must be visible before we test NEED_RESCHED,
2750 * paired by resched_task()
2751 */
2752 smp_mb();
2753
2754 return unlikely(tif_need_resched());
2755}
2756
2757static inline void __current_clr_polling(void)
Thomas Gleixner3a98f872013-03-21 22:49:33 +01002758{
2759 current_thread_info()->status &= ~TS_POLLING;
Peter Zijlstraea811742013-09-11 12:43:13 +02002760}
2761
2762static inline bool __must_check current_clr_polling_and_test(void)
2763{
2764 __current_clr_polling();
2765
2766 /*
2767 * Polling state must be visible before we test NEED_RESCHED,
2768 * paired by resched_task()
2769 */
2770 smp_mb();
2771
2772 return unlikely(tif_need_resched());
Thomas Gleixner3a98f872013-03-21 22:49:33 +01002773}
Thomas Gleixneree761f62013-03-21 22:49:32 +01002774#elif defined(TIF_POLLING_NRFLAG)
2775static inline int tsk_is_polling(struct task_struct *p)
2776{
2777 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2778}
Peter Zijlstraea811742013-09-11 12:43:13 +02002779
2780static inline void __current_set_polling(void)
Thomas Gleixner3a98f872013-03-21 22:49:33 +01002781{
2782 set_thread_flag(TIF_POLLING_NRFLAG);
2783}
2784
Peter Zijlstraea811742013-09-11 12:43:13 +02002785static inline bool __must_check current_set_polling_and_test(void)
2786{
2787 __current_set_polling();
2788
2789 /*
2790 * Polling state must be visible before we test NEED_RESCHED,
2791 * paired by resched_task()
2792 *
2793 * XXX: assumes set/clear bit are identical barrier wise.
2794 */
2795 smp_mb__after_clear_bit();
2796
2797 return unlikely(tif_need_resched());
2798}
2799
2800static inline void __current_clr_polling(void)
Thomas Gleixner3a98f872013-03-21 22:49:33 +01002801{
2802 clear_thread_flag(TIF_POLLING_NRFLAG);
2803}
Peter Zijlstraea811742013-09-11 12:43:13 +02002804
2805static inline bool __must_check current_clr_polling_and_test(void)
2806{
2807 __current_clr_polling();
2808
2809 /*
2810 * Polling state must be visible before we test NEED_RESCHED,
2811 * paired by resched_task()
2812 */
2813 smp_mb__after_clear_bit();
2814
2815 return unlikely(tif_need_resched());
2816}
2817
Thomas Gleixneree761f62013-03-21 22:49:32 +01002818#else
2819static inline int tsk_is_polling(struct task_struct *p) { return 0; }
Peter Zijlstraea811742013-09-11 12:43:13 +02002820static inline void __current_set_polling(void) { }
2821static inline void __current_clr_polling(void) { }
2822
2823static inline bool __must_check current_set_polling_and_test(void)
2824{
2825 return unlikely(tif_need_resched());
2826}
2827static inline bool __must_check current_clr_polling_and_test(void)
2828{
2829 return unlikely(tif_need_resched());
2830}
Thomas Gleixneree761f62013-03-21 22:49:32 +01002831#endif
2832
Peter Zijlstra8cb75e02013-11-20 12:22:37 +01002833static inline void current_clr_polling(void)
2834{
2835 __current_clr_polling();
2836
2837 /*
2838 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2839 * Once the bit is cleared, we'll get IPIs with every new
2840 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2841 * fold.
2842 */
2843 smp_mb(); /* paired with resched_task() */
2844
2845 preempt_fold_need_resched();
2846}
2847
Peter Zijlstra75f93fe2013-09-27 17:30:03 +02002848static __always_inline bool need_resched(void)
2849{
2850 return unlikely(tif_need_resched());
2851}
2852
Thomas Gleixneree761f62013-03-21 22:49:32 +01002853/*
Frank Mayharf06febc2008-09-12 09:54:39 -07002854 * Thread group CPU time accounting.
2855 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +01002856void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
Peter Zijlstra4da94d492009-02-11 11:30:27 +01002857void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
Frank Mayharf06febc2008-09-12 09:54:39 -07002858
2859static inline void thread_group_cputime_init(struct signal_struct *sig)
2860{
Thomas Gleixneree30a7b2009-07-25 18:56:56 +02002861 raw_spin_lock_init(&sig->cputimer.lock);
Frank Mayharf06febc2008-09-12 09:54:39 -07002862}
2863
Frank Mayharf06febc2008-09-12 09:54:39 -07002864/*
Roland McGrath7bb44ad2007-05-23 13:57:44 -07002865 * Reevaluate whether the task has signals pending delivery.
2866 * Wake the task if so.
2867 * This is required every time the blocked sigset_t changes.
2868 * callers must hold sighand->siglock.
2869 */
2870extern void recalc_sigpending_and_wake(struct task_struct *t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002871extern void recalc_sigpending(void);
2872
Oleg Nesterov910ffdb2013-01-21 20:47:41 +01002873extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2874
2875static inline void signal_wake_up(struct task_struct *t, bool resume)
2876{
2877 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2878}
2879static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2880{
2881 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2882}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002883
2884/*
2885 * Wrappers for p->thread_info->cpu access. No-op on UP.
2886 */
2887#ifdef CONFIG_SMP
2888
2889static inline unsigned int task_cpu(const struct task_struct *p)
2890{
Al Viroa1261f52005-11-13 16:06:55 -08002891 return task_thread_info(p)->cpu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002892}
2893
Ingo Molnarb32e86b2013-10-07 11:29:30 +01002894static inline int task_node(const struct task_struct *p)
2895{
2896 return cpu_to_node(task_cpu(p));
2897}
2898
Ingo Molnarc65cc872007-07-09 18:51:58 +02002899extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002900
2901#else
2902
2903static inline unsigned int task_cpu(const struct task_struct *p)
2904{
2905 return 0;
2906}
2907
2908static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2909{
2910}
2911
2912#endif /* CONFIG_SMP */
2913
Rusty Russell96f874e2008-11-25 02:35:14 +10302914extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2915extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
Siddha, Suresh B5c45bf22006-06-27 02:54:42 -07002916
Dhaval Giani7c941432010-01-20 13:26:18 +01002917#ifdef CONFIG_CGROUP_SCHED
Yong Zhang07e06b02011-01-07 15:17:36 +08002918extern struct task_group root_task_group;
Peter Zijlstra8323f262012-06-22 13:36:05 +02002919#endif /* CONFIG_CGROUP_SCHED */
Srivatsa Vaddagiri9b5b7752007-10-15 17:00:09 +02002920
Dhaval Giani54e99122009-02-27 15:13:54 +05302921extern int task_can_switch_user(struct user_struct *up,
2922 struct task_struct *tsk);
2923
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002924#ifdef CONFIG_TASK_XACCT
2925static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2926{
Andrea Righi940389b2008-07-28 00:48:12 +02002927 tsk->ioac.rchar += amt;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002928}
2929
2930static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2931{
Andrea Righi940389b2008-07-28 00:48:12 +02002932 tsk->ioac.wchar += amt;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002933}
2934
2935static inline void inc_syscr(struct task_struct *tsk)
2936{
Andrea Righi940389b2008-07-28 00:48:12 +02002937 tsk->ioac.syscr++;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002938}
2939
2940static inline void inc_syscw(struct task_struct *tsk)
2941{
Andrea Righi940389b2008-07-28 00:48:12 +02002942 tsk->ioac.syscw++;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002943}
2944#else
2945static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2946{
2947}
2948
2949static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2950{
2951}
2952
2953static inline void inc_syscr(struct task_struct *tsk)
2954{
2955}
2956
2957static inline void inc_syscw(struct task_struct *tsk)
2958{
2959}
2960#endif
2961
Dave Hansen82455252008-02-04 22:28:59 -08002962#ifndef TASK_SIZE_OF
2963#define TASK_SIZE_OF(tsk) TASK_SIZE
2964#endif
2965
Balbir Singhcf475ad2008-04-29 01:00:16 -07002966#ifdef CONFIG_MM_OWNER
2967extern void mm_update_next_owner(struct mm_struct *mm);
2968extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2969#else
2970static inline void mm_update_next_owner(struct mm_struct *mm)
2971{
2972}
2973
2974static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2975{
2976}
2977#endif /* CONFIG_MM_OWNER */
2978
Jiri Slaby3e10e712009-11-19 17:16:37 +01002979static inline unsigned long task_rlimit(const struct task_struct *tsk,
2980 unsigned int limit)
2981{
2982 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2983}
2984
2985static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2986 unsigned int limit)
2987{
2988 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2989}
2990
2991static inline unsigned long rlimit(unsigned int limit)
2992{
2993 return task_rlimit(current, limit);
2994}
2995
2996static inline unsigned long rlimit_max(unsigned int limit)
2997{
2998 return task_rlimit_max(current, limit);
2999}
3000
Linus Torvalds1da177e2005-04-16 15:20:36 -07003001#endif