blob: eaacac4ae77d58eb3e6f5c237d0b3e1bd1e8f962 [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>
Jack Millerab602f72014-08-08 14:23:19 -070036#include <linux/shm.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070037#include <linux/signal.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070038#include <linux/compiler.h>
39#include <linux/completion.h>
40#include <linux/pid.h>
41#include <linux/percpu.h>
42#include <linux/topology.h>
Peter Zijlstra3e26c142007-10-16 23:25:50 -070043#include <linux/proportions.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070044#include <linux/seccomp.h>
Ingo Molnare56d0902006-01-08 01:01:37 -080045#include <linux/rcupdate.h>
Jiri Pirko05725f72009-04-14 20:17:16 +020046#include <linux/rculist.h>
Ingo Molnar23f78d4a2006-06-27 02:54:53 -070047#include <linux/rtmutex.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070048
David Woodhousea3b67142006-04-25 14:54:40 +010049#include <linux/time.h>
50#include <linux/param.h>
51#include <linux/resource.h>
52#include <linux/timer.h>
53#include <linux/hrtimer.h>
Andrew Morton7c3ab732006-12-10 02:19:19 -080054#include <linux/task_io_accounting.h>
Arjan van de Ven97455122008-01-25 21:08:34 +010055#include <linux/latencytop.h>
David Howells9e2b2dc2008-08-13 16:20:04 +010056#include <linux/cred.h>
Peter Zijlstrafa14ff42011-09-12 13:06:17 +020057#include <linux/llist.h>
Eric W. Biederman7b44ab92011-11-16 23:20:58 -080058#include <linux/uidgid.h>
Ming Lei21caf2f2013-02-22 16:34:08 -080059#include <linux/gfp.h>
David Woodhousea3b67142006-04-25 14:54:40 +010060
61#include <asm/processor.h>
H. J. Lu36d57ac2005-09-06 15:16:49 -070062
Dario Faggiolid50dde52013-11-07 14:43:36 +010063#define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
64
65/*
66 * Extended scheduling parameters data structure.
67 *
68 * This is needed because the original struct sched_param can not be
69 * altered without introducing ABI issues with legacy applications
70 * (e.g., in sched_getparam()).
71 *
72 * However, the possibility of specifying more than just a priority for
73 * the tasks may be useful for a wide variety of application fields, e.g.,
74 * multimedia, streaming, automation and control, and many others.
75 *
76 * This variant (sched_attr) is meant at describing a so-called
77 * sporadic time-constrained task. In such model a task is specified by:
78 * - the activation period or minimum instance inter-arrival time;
79 * - the maximum (or average, depending on the actual scheduling
80 * discipline) computation time of all instances, a.k.a. runtime;
81 * - the deadline (relative to the actual activation time) of each
82 * instance.
83 * Very briefly, a periodic (sporadic) task asks for the execution of
84 * some specific computation --which is typically called an instance--
85 * (at most) every period. Moreover, each instance typically lasts no more
86 * than the runtime and must be completed by time instant t equal to
87 * the instance activation time + the deadline.
88 *
89 * This is reflected by the actual fields of the sched_attr structure:
90 *
91 * @size size of the structure, for fwd/bwd compat.
92 *
93 * @sched_policy task's scheduling policy
94 * @sched_flags for customizing the scheduler behaviour
95 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
96 * @sched_priority task's static priority (SCHED_FIFO/RR)
97 * @sched_deadline representative of the task's deadline
98 * @sched_runtime representative of the task's runtime
99 * @sched_period representative of the task's period
100 *
101 * Given this task model, there are a multiplicity of scheduling algorithms
102 * and policies, that can be used to ensure all the tasks will make their
103 * timing constraints.
Dario Faggioliaab03e02013-11-28 11:14:43 +0100104 *
105 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
106 * only user of this new interface. More information about the algorithm
107 * available in the scheduling class file or in Documentation/.
Dario Faggiolid50dde52013-11-07 14:43:36 +0100108 */
109struct sched_attr {
110 u32 size;
111
112 u32 sched_policy;
113 u64 sched_flags;
114
115 /* SCHED_NORMAL, SCHED_BATCH */
116 s32 sched_nice;
117
118 /* SCHED_FIFO, SCHED_RR */
119 u32 sched_priority;
120
121 /* SCHED_DEADLINE */
122 u64 sched_runtime;
123 u64 sched_deadline;
124 u64 sched_period;
125};
126
Linus Torvalds1da177e2005-04-16 15:20:36 -0700127struct exec_domain;
Ingo Molnarc87e2832006-06-27 02:54:58 -0700128struct futex_pi_state;
Alexey Dobriyan286100a2008-01-25 21:08:34 +0100129struct robust_list_head;
Akinobu Mitabddd87c2010-02-23 08:55:42 +0100130struct bio_list;
Al Viro5ad4e532009-03-29 19:50:06 -0400131struct fs_struct;
Ingo Molnarcdd6c482009-09-21 12:02:48 +0200132struct perf_event_context;
Jens Axboe73c10102011-03-08 13:19:51 +0100133struct blk_plug;
Linus Torvaldsc4ad8f92014-02-05 12:54:53 -0800134struct filename;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700135
Davidlohr Bueso615d6e82014-04-07 15:37:25 -0700136#define VMACACHE_BITS 2
137#define VMACACHE_SIZE (1U << VMACACHE_BITS)
138#define VMACACHE_MASK (VMACACHE_SIZE - 1)
139
Linus Torvalds1da177e2005-04-16 15:20:36 -0700140/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700141 * These are the constant used to fake the fixed-point load-average
142 * counting. Some notes:
143 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
144 * a load-average precision of 10 bits integer + 11 bits fractional
145 * - if you want to count load-averages more often, you need more
146 * precision, or rounding will get you. With 2-second counting freq,
147 * the EXP_n values would be 1981, 2034 and 2043 if still using only
148 * 11 bit fractions.
149 */
150extern unsigned long avenrun[]; /* Load averages */
Thomas Gleixner2d024942009-05-02 20:08:52 +0200151extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700152
153#define FSHIFT 11 /* nr of bits of precision */
154#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
Linus Torvalds0c2043a2007-10-07 16:17:38 -0700155#define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700156#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
157#define EXP_5 2014 /* 1/exp(5sec/5min) */
158#define EXP_15 2037 /* 1/exp(5sec/15min) */
159
160#define CALC_LOAD(load,exp,n) \
161 load *= exp; \
162 load += n*(FIXED_1-exp); \
163 load >>= FSHIFT;
164
165extern unsigned long total_forks;
166extern int nr_threads;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700167DECLARE_PER_CPU(unsigned long, process_counts);
168extern int nr_processes(void);
169extern unsigned long nr_running(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700170extern unsigned long nr_iowait(void);
Peter Zijlstra8c215bd2010-07-01 09:07:17 +0200171extern unsigned long nr_iowait_cpu(int cpu);
Mel Gorman372ba8c2014-08-06 14:19:21 +0100172extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
Arjan van de Ven69d25872009-09-21 17:04:08 -0700173
Peter Zijlstra0f004f52010-11-30 19:48:45 +0100174extern void calc_global_load(unsigned long ticks);
Peter Zijlstra5aaa0b72012-05-17 17:15:29 +0200175extern void update_cpu_load_nohz(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700176
Steven Rostedt7e49fcc2009-01-22 19:01:40 -0500177extern unsigned long get_parent_ip(unsigned long addr);
178
Paul E. McKenneyb637a322012-09-19 16:58:38 -0700179extern void dump_cpu_task(int cpu);
180
Ingo Molnar43ae34c2007-07-09 18:52:00 +0200181struct seq_file;
182struct cfs_rq;
Ingo Molnar4cf86d72007-10-15 17:00:14 +0200183struct task_group;
Ingo Molnar43ae34c2007-07-09 18:52:00 +0200184#ifdef CONFIG_SCHED_DEBUG
185extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
186extern void proc_sched_set_task(struct task_struct *p);
187extern void
Ingo Molnar5cef9ec2007-08-09 11:16:47 +0200188print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
Ingo Molnar43ae34c2007-07-09 18:52:00 +0200189#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700190
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700191/*
192 * Task state bitmask. NOTE! These bits are also
193 * encoded in fs/proc/array.c: get_task_state().
194 *
195 * We have two separate sets of flags: task->state
196 * is about runnability, while task->exit_state are
197 * about the task exiting. Confusing, but this way
198 * modifying one set can't modify the other one by
199 * mistake.
200 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700201#define TASK_RUNNING 0
202#define TASK_INTERRUPTIBLE 1
203#define TASK_UNINTERRUPTIBLE 2
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500204#define __TASK_STOPPED 4
205#define __TASK_TRACED 8
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700206/* in tsk->exit_state */
Oleg Nesterovad866222014-04-07 15:38:46 -0700207#define EXIT_DEAD 16
208#define EXIT_ZOMBIE 32
Oleg Nesterovabd50b32014-04-07 15:38:42 -0700209#define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700210/* in tsk->state again */
Mike Galbraithaf927232007-10-15 17:00:13 +0200211#define TASK_DEAD 64
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500212#define TASK_WAKEKILL 128
Peter Zijlstrae9c84312009-09-15 14:43:03 +0200213#define TASK_WAKING 256
Thomas Gleixnerf2530dc2013-04-09 09:33:34 +0200214#define TASK_PARKED 512
215#define TASK_STATE_MAX 1024
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500216
Masatake YAMATOad0f6142014-05-22 11:54:20 -0700217#define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWP"
Peter Zijlstra73342152009-12-17 13:16:27 +0100218
Peter Zijlstrae1781532009-12-17 13:16:30 +0100219extern char ___assert_task_state[1 - 2*!!(
220 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500221
222/* Convenience macros for the sake of set_task_state */
223#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
224#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
225#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700226
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500227/* Convenience macros for the sake of wake_up */
228#define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500229#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500230
231/* get_task_state() */
232#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500233 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
Oleg Nesterov74e37202014-01-23 15:55:35 -0800234 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500235
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500236#define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
237#define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500238#define task_is_stopped_or_traced(task) \
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500239 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500240#define task_contributes_to_load(task) \
Nathan Lynche3c8ca82009-04-08 19:45:12 -0500241 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
Tejun Heo376fede2011-11-21 12:32:24 -0800242 (task->flags & PF_FROZEN) == 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700243
244#define __set_task_state(tsk, state_value) \
245 do { (tsk)->state = (state_value); } while (0)
246#define set_task_state(tsk, state_value) \
247 set_mb((tsk)->state, (state_value))
248
Andrew Morton498d0c52005-09-13 01:25:14 -0700249/*
250 * set_current_state() includes a barrier so that the write of current->state
251 * is correctly serialised wrt the caller's subsequent test of whether to
252 * actually sleep:
253 *
254 * set_current_state(TASK_UNINTERRUPTIBLE);
255 * if (do_i_need_to_sleep())
256 * schedule();
257 *
258 * If the caller does not need such serialisation then use __set_current_state()
259 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700260#define __set_current_state(state_value) \
261 do { current->state = (state_value); } while (0)
262#define set_current_state(state_value) \
263 set_mb(current->state, (state_value))
264
265/* Task command name length */
266#define TASK_COMM_LEN 16
267
Linus Torvalds1da177e2005-04-16 15:20:36 -0700268#include <linux/spinlock.h>
269
270/*
271 * This serializes "schedule()" and also protects
272 * the run-queue from deletions/modifications (but
273 * _adding_ to the beginning of the run-queue has
274 * a separate lock).
275 */
276extern rwlock_t tasklist_lock;
277extern spinlock_t mmlist_lock;
278
Ingo Molnar36c8b582006-07-03 00:25:41 -0700279struct task_struct;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700280
Paul E. McKenneydb1466b2010-03-03 07:46:56 -0800281#ifdef CONFIG_PROVE_RCU
282extern int lockdep_tasklist_lock_is_held(void);
283#endif /* #ifdef CONFIG_PROVE_RCU */
284
Linus Torvalds1da177e2005-04-16 15:20:36 -0700285extern void sched_init(void);
286extern void sched_init_smp(void);
Harvey Harrison2d07b252008-02-15 09:56:34 -0800287extern asmlinkage void schedule_tail(struct task_struct *prev);
Ingo Molnar36c8b582006-07-03 00:25:41 -0700288extern void init_idle(struct task_struct *idle, int cpu);
Ingo Molnar1df21052007-07-09 18:51:58 +0200289extern void init_idle_bootup_task(struct task_struct *idle);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700290
Andrew Morton89f19f02009-09-19 11:55:44 -0700291extern int runqueue_is_locked(int cpu);
Ingo Molnar017730c2008-05-12 21:20:52 +0200292
Frederic Weisbecker3451d022011-08-10 23:21:01 +0200293#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
Alex Shic1cc0172012-09-10 15:10:58 +0800294extern void nohz_balance_enter_idle(int cpu);
Suresh Siddha69e1e812011-12-01 17:07:33 -0800295extern void set_cpu_sd_state_idle(void);
Viresh Kumar6201b4d2014-03-18 16:26:07 +0530296extern int get_nohz_timer_target(int pinned);
Siddha, Suresh B46cb4b72007-05-08 00:32:51 -0700297#else
Alex Shic1cc0172012-09-10 15:10:58 +0800298static inline void nohz_balance_enter_idle(int cpu) { }
Peter Zijlstrafdaabd82011-12-06 12:47:55 +0100299static inline void set_cpu_sd_state_idle(void) { }
Viresh Kumar6201b4d2014-03-18 16:26:07 +0530300static inline int get_nohz_timer_target(int pinned)
301{
302 return smp_processor_id();
303}
Siddha, Suresh B46cb4b72007-05-08 00:32:51 -0700304#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700305
Ingo Molnare59e2ae2006-12-06 20:35:59 -0800306/*
Ingo Molnar39bc89f2007-04-25 20:50:03 -0700307 * Only dump TASK_* tasks. (0 for all tasks)
Ingo Molnare59e2ae2006-12-06 20:35:59 -0800308 */
309extern void show_state_filter(unsigned long state_filter);
310
311static inline void show_state(void)
312{
Ingo Molnar39bc89f2007-04-25 20:50:03 -0700313 show_state_filter(0);
Ingo Molnare59e2ae2006-12-06 20:35:59 -0800314}
315
Linus Torvalds1da177e2005-04-16 15:20:36 -0700316extern void show_regs(struct pt_regs *);
317
318/*
319 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
320 * task), SP is the stack pointer of the first frame that should be shown in the back
321 * trace (or NULL if the entire call-chain of the task should be shown).
322 */
323extern void show_stack(struct task_struct *task, unsigned long *sp);
324
325void io_schedule(void);
326long io_schedule_timeout(long timeout);
327
328extern void cpu_init (void);
329extern void trap_init(void);
330extern void update_process_times(int user);
331extern void scheduler_tick(void);
332
Ingo Molnar82a1fcb2008-01-25 21:08:02 +0100333extern void sched_show_task(struct task_struct *p);
334
Frederic Weisbecker19cc36c2010-05-13 02:30:49 +0200335#ifdef CONFIG_LOCKUP_DETECTOR
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700336extern void touch_softlockup_watchdog(void);
Jason Wesseld6ad3e22010-01-27 16:25:22 -0600337extern void touch_softlockup_watchdog_sync(void);
Jeremy Fitzhardinge04c91672007-05-08 00:28:05 -0700338extern void touch_all_softlockup_watchdogs(void);
Don Zickus332fbdb2010-05-07 17:11:45 -0400339extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
340 void __user *buffer,
341 size_t *lenp, loff_t *ppos);
Ingo Molnar9c44bc02008-05-12 21:21:04 +0200342extern unsigned int softlockup_panic;
Peter Zijlstra004417a2010-11-25 18:38:29 +0100343void lockup_detector_init(void);
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700344#else
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700345static inline void touch_softlockup_watchdog(void)
346{
347}
Jason Wesseld6ad3e22010-01-27 16:25:22 -0600348static inline void touch_softlockup_watchdog_sync(void)
349{
350}
Jeremy Fitzhardinge04c91672007-05-08 00:28:05 -0700351static inline void touch_all_softlockup_watchdogs(void)
352{
353}
Peter Zijlstra004417a2010-11-25 18:38:29 +0100354static inline void lockup_detector_init(void)
355{
356}
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700357#endif
358
Marcelo Tosatti8b414522013-10-11 21:39:26 -0300359#ifdef CONFIG_DETECT_HUNG_TASK
360void reset_hung_task_detector(void);
361#else
362static inline void reset_hung_task_detector(void)
363{
364}
365#endif
366
Linus Torvalds1da177e2005-04-16 15:20:36 -0700367/* Attach to any functions which should be ignored in wchan output. */
368#define __sched __attribute__((__section__(".sched.text")))
Ingo Molnardeaf2222007-11-28 15:52:56 +0100369
370/* Linker adds these: start and end of __sched functions */
371extern char __sched_text_start[], __sched_text_end[];
372
Linus Torvalds1da177e2005-04-16 15:20:36 -0700373/* Is this address in the __sched functions? */
374extern int in_sched_functions(unsigned long addr);
375
376#define MAX_SCHEDULE_TIMEOUT LONG_MAX
Harvey Harrisonb3c97522008-02-13 15:03:15 -0800377extern signed long schedule_timeout(signed long timeout);
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700378extern signed long schedule_timeout_interruptible(signed long timeout);
Matthew Wilcox294d5cc2007-12-06 11:59:46 -0500379extern signed long schedule_timeout_killable(signed long timeout);
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700380extern signed long schedule_timeout_uninterruptible(signed long timeout);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700381asmlinkage void schedule(void);
Thomas Gleixnerc5491ea2011-03-21 12:09:35 +0100382extern void schedule_preempt_disabled(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700383
Serge E. Hallynab516012006-10-02 02:18:06 -0700384struct nsproxy;
Cedric Le Goateracce2922007-07-15 23:40:59 -0700385struct user_namespace;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700386
David Howellsefc1a3b2010-01-15 17:01:35 -0800387#ifdef CONFIG_MMU
388extern void arch_pick_mmap_layout(struct mm_struct *mm);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700389extern unsigned long
390arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
391 unsigned long, unsigned long);
392extern unsigned long
393arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
394 unsigned long len, unsigned long pgoff,
395 unsigned long flags);
David Howellsefc1a3b2010-01-15 17:01:35 -0800396#else
397static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
398#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700399
Kees Cookd049f742013-11-12 15:11:17 -0800400#define SUID_DUMP_DISABLE 0 /* No setuid dumping */
401#define SUID_DUMP_USER 1 /* Dump as user of process */
402#define SUID_DUMP_ROOT 2 /* Dump as root */
403
Kawai, Hidehiro6c5d5232007-07-19 01:48:27 -0700404/* mm flags */
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700405
Oleg Nesterov7288e112014-01-23 15:55:32 -0800406/* for SUID_DUMP_* above */
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700407#define MMF_DUMPABLE_BITS 2
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700408#define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700409
Oleg Nesterov942be382014-01-23 15:55:34 -0800410extern void set_dumpable(struct mm_struct *mm, int value);
411/*
412 * This returns the actual value of the suid_dumpable flag. For things
413 * that are using this for checking for privilege transitions, it must
414 * test against SUID_DUMP_USER rather than treating it as a boolean
415 * value.
416 */
417static inline int __get_dumpable(unsigned long mm_flags)
418{
419 return mm_flags & MMF_DUMPABLE_MASK;
420}
421
422static inline int get_dumpable(struct mm_struct *mm)
423{
424 return __get_dumpable(mm->flags);
425}
426
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700427/* coredump filter bits */
428#define MMF_DUMP_ANON_PRIVATE 2
429#define MMF_DUMP_ANON_SHARED 3
430#define MMF_DUMP_MAPPED_PRIVATE 4
431#define MMF_DUMP_MAPPED_SHARED 5
Roland McGrath82df3972007-10-16 23:27:02 -0700432#define MMF_DUMP_ELF_HEADERS 6
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700433#define MMF_DUMP_HUGETLB_PRIVATE 7
434#define MMF_DUMP_HUGETLB_SHARED 8
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700435
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700436#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700437#define MMF_DUMP_FILTER_BITS 7
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700438#define MMF_DUMP_FILTER_MASK \
439 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
440#define MMF_DUMP_FILTER_DEFAULT \
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700441 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
Roland McGrath656eb2c2008-10-18 20:28:23 -0700442 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
443
444#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
445# define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
446#else
447# define MMF_DUMP_MASK_DEFAULT_ELF 0
448#endif
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700449 /* leave room for more dump flags */
450#define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
Andrea Arcangeliba761492011-01-13 15:46:58 -0800451#define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
Konstantin Khlebnikovbafb2822012-06-07 14:21:11 -0700452#define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700453
Oleg Nesterov9f68f672012-08-19 16:15:09 +0200454#define MMF_HAS_UPROBES 19 /* has uprobes */
455#define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
Oleg Nesterovf8ac4ec2012-08-08 17:11:42 +0200456
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700457#define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
Kawai, Hidehiro6c5d5232007-07-19 01:48:27 -0700458
Linus Torvalds1da177e2005-04-16 15:20:36 -0700459struct sighand_struct {
460 atomic_t count;
461 struct k_sigaction action[_NSIG];
462 spinlock_t siglock;
Davide Libenzib8fceee2007-09-20 12:40:16 -0700463 wait_queue_head_t signalfd_wqh;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700464};
465
KaiGai Kohei0e464812006-06-25 05:49:24 -0700466struct pacct_struct {
KaiGai Koheif6ec29a2006-06-25 05:49:25 -0700467 int ac_flag;
468 long ac_exitcode;
KaiGai Kohei0e464812006-06-25 05:49:24 -0700469 unsigned long ac_mem;
KaiGai Kohei77787bf2006-06-25 05:49:26 -0700470 cputime_t ac_utime, ac_stime;
471 unsigned long ac_minflt, ac_majflt;
KaiGai Kohei0e464812006-06-25 05:49:24 -0700472};
473
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200474struct cpu_itimer {
475 cputime_t expires;
476 cputime_t incr;
Stanislaw Gruszka8356b5f2009-07-29 12:15:27 +0200477 u32 error;
478 u32 incr_error;
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200479};
480
Frank Mayharf06febc2008-09-12 09:54:39 -0700481/**
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +0100482 * struct cputime - snaphsot of system and user cputime
483 * @utime: time spent in user mode
484 * @stime: time spent in system mode
485 *
486 * Gathers a generic snapshot of user and system time.
487 */
488struct cputime {
489 cputime_t utime;
490 cputime_t stime;
491};
492
493/**
Frank Mayharf06febc2008-09-12 09:54:39 -0700494 * struct task_cputime - collected CPU time counts
495 * @utime: time spent in user mode, in &cputime_t units
496 * @stime: time spent in kernel mode, in &cputime_t units
497 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
Ingo Molnar5ce73a42008-09-14 17:11:46 +0200498 *
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +0100499 * This is an extension of struct cputime that includes the total runtime
500 * spent by the task from the scheduler point of view.
501 *
502 * As a result, this structure groups together three kinds of CPU time
503 * that are tracked for threads and thread groups. Most things considering
Frank Mayharf06febc2008-09-12 09:54:39 -0700504 * CPU time want to group these counts together and treat all three
505 * of them in parallel.
506 */
507struct task_cputime {
508 cputime_t utime;
509 cputime_t stime;
510 unsigned long long sum_exec_runtime;
511};
512/* Alternate field names when used to cache expirations. */
513#define prof_exp stime
514#define virt_exp utime
515#define sched_exp sum_exec_runtime
516
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100517#define INIT_CPUTIME \
518 (struct task_cputime) { \
Martin Schwidefsky64861632011-12-15 14:56:09 +0100519 .utime = 0, \
520 .stime = 0, \
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100521 .sum_exec_runtime = 0, \
522 }
523
Peter Zijlstraa233f112013-09-23 19:04:26 +0200524#ifdef CONFIG_PREEMPT_COUNT
525#define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
526#else
527#define PREEMPT_DISABLED PREEMPT_ENABLED
528#endif
529
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200530/*
531 * Disable preemption until the scheduler is running.
532 * Reset by start_kernel()->sched_init()->init_idle().
Peter Zijlstrad86ee482009-07-10 14:57:57 +0200533 *
534 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
535 * before the scheduler is active -- see should_resched().
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200536 */
Peter Zijlstraa233f112013-09-23 19:04:26 +0200537#define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200538
Frank Mayharf06febc2008-09-12 09:54:39 -0700539/**
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100540 * struct thread_group_cputimer - thread group interval timer counts
541 * @cputime: thread group interval timers.
542 * @running: non-zero when there are timers running and
543 * @cputime receives updates.
544 * @lock: lock for fields in this struct.
Frank Mayharf06febc2008-09-12 09:54:39 -0700545 *
546 * This structure contains the version of task_cputime, above, that is
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100547 * used for thread group CPU timer calculations.
Frank Mayharf06febc2008-09-12 09:54:39 -0700548 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100549struct thread_group_cputimer {
550 struct task_cputime cputime;
551 int running;
Thomas Gleixneree30a7b2009-07-25 18:56:56 +0200552 raw_spinlock_t lock;
Frank Mayharf06febc2008-09-12 09:54:39 -0700553};
Frank Mayharf06febc2008-09-12 09:54:39 -0700554
Ben Blum4714d1d2011-05-26 16:25:18 -0700555#include <linux/rwsem.h>
Mike Galbraith5091faa2010-11-30 14:18:03 +0100556struct autogroup;
557
Linus Torvalds1da177e2005-04-16 15:20:36 -0700558/*
Jonathan Neuschäfere815f0a2011-03-21 20:24:47 +0100559 * NOTE! "signal_struct" does not have its own
Linus Torvalds1da177e2005-04-16 15:20:36 -0700560 * locking, because a shared signal_struct always
561 * implies a shared sighand_struct, so locking
562 * sighand_struct is always a proper superset of
563 * the locking of signal_struct.
564 */
565struct signal_struct {
Oleg Nesterovea6d2902010-05-26 14:43:16 -0700566 atomic_t sigcnt;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700567 atomic_t live;
Oleg Nesterovb3ac0222010-05-26 14:43:24 -0700568 int nr_threads;
Oleg Nesterov0c740d02014-01-21 15:49:56 -0800569 struct list_head thread_head;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700570
571 wait_queue_head_t wait_chldexit; /* for wait4() */
572
573 /* current thread group signal load-balancing target: */
Ingo Molnar36c8b582006-07-03 00:25:41 -0700574 struct task_struct *curr_target;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700575
576 /* shared signal handling: */
577 struct sigpending shared_pending;
578
579 /* thread group exit support */
580 int group_exit_code;
581 /* overloaded:
582 * - notify group_exit_task when ->count is equal to notify_count
583 * - everyone except group_exit_task is stopped during signal delivery
584 * of fatal signals, group_exit_task processes the signal.
585 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700586 int notify_count;
Richard Kennedy07dd20e2008-08-01 13:18:04 +0100587 struct task_struct *group_exit_task;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700588
589 /* thread group stop support, overloads group_exit_code too */
590 int group_stop_count;
591 unsigned int flags; /* see SIGNAL_* flags below */
592
Lennart Poetteringebec18a2012-03-23 15:01:54 -0700593 /*
594 * PR_SET_CHILD_SUBREAPER marks a process, like a service
595 * manager, to re-parent orphan (double-forking) child processes
596 * to this process instead of 'init'. The service manager is
597 * able to receive SIGCHLD signals and is able to investigate
598 * the process until it calls wait(). All children of this
599 * process will inherit a flag if they should look for a
600 * child_subreaper process at exit.
601 */
602 unsigned int is_child_subreaper:1;
603 unsigned int has_child_subreaper:1;
604
Linus Torvalds1da177e2005-04-16 15:20:36 -0700605 /* POSIX.1b Interval Timers */
Pavel Emelyanov5ed67f02013-03-11 13:12:21 +0400606 int posix_timer_id;
607 struct list_head posix_timers;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700608
609 /* ITIMER_REAL timer for the process */
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800610 struct hrtimer real_timer;
Oleg Nesterovfea9d172008-02-08 04:19:19 -0800611 struct pid *leader_pid;
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800612 ktime_t it_real_incr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700613
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200614 /*
615 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
616 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
617 * values are defined to 0 and 1 respectively
618 */
619 struct cpu_itimer it[2];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700620
Frank Mayharf06febc2008-09-12 09:54:39 -0700621 /*
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100622 * Thread group totals for process CPU timers.
623 * See thread_group_cputimer(), et al, for details.
Frank Mayharf06febc2008-09-12 09:54:39 -0700624 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100625 struct thread_group_cputimer cputimer;
Frank Mayharf06febc2008-09-12 09:54:39 -0700626
627 /* Earliest-expiration cache. */
628 struct task_cputime cputime_expires;
629
630 struct list_head cpu_timers[3];
631
Eric W. Biedermanab521dc2007-02-12 00:53:00 -0800632 struct pid *tty_old_pgrp;
Cedric Le Goater1ec320a2006-12-08 02:37:55 -0800633
Linus Torvalds1da177e2005-04-16 15:20:36 -0700634 /* boolean value for session group leader */
635 int leader;
636
637 struct tty_struct *tty; /* NULL if no tty */
638
Mike Galbraith5091faa2010-11-30 14:18:03 +0100639#ifdef CONFIG_SCHED_AUTOGROUP
640 struct autogroup *autogroup;
641#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700642 /*
643 * Cumulative resource counters for dead threads in the group,
644 * and for reaped dead child processes forked by this group.
645 * Live threads maintain their own counters and add to these
646 * in __exit_signal, except for the group leader.
647 */
Peter Zijlstra32bd6712009-02-05 12:24:15 +0100648 cputime_t utime, stime, cutime, cstime;
Laurent Vivier9ac52312007-10-15 17:00:19 +0200649 cputime_t gtime;
650 cputime_t cgtime;
Frederic Weisbecker9fbc42e2013-02-25 17:25:39 +0100651#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +0100652 struct cputime prev_cputime;
Hidetoshi Seto0cf55e12009-12-02 17:28:07 +0900653#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700654 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
655 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
Eric Dumazet6eaeeab2007-05-10 22:22:37 -0700656 unsigned long inblock, oublock, cinblock, coublock;
Jiri Pirko1f102062009-09-22 16:44:10 -0700657 unsigned long maxrss, cmaxrss;
Andrea Righi940389b2008-07-28 00:48:12 +0200658 struct task_io_accounting ioac;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700659
660 /*
Peter Zijlstra32bd6712009-02-05 12:24:15 +0100661 * Cumulative ns of schedule CPU time fo dead threads in the
662 * group, not including a zombie group leader, (This only differs
663 * from jiffies_to_ns(utime + stime) if sched_clock uses something
664 * other than jiffies.)
665 */
666 unsigned long long sum_sched_runtime;
667
668 /*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700669 * We don't bother to synchronize most readers of this at all,
670 * because there is no reader checking a limit that actually needs
671 * to get both rlim_cur and rlim_max atomically, and either one
672 * alone is a single word that can safely be read normally.
673 * getrlimit/setrlimit use task_lock(current->group_leader) to
674 * protect this instead of the siglock, because they really
675 * have no need to disable irqs.
676 */
677 struct rlimit rlim[RLIM_NLIMITS];
678
KaiGai Kohei0e464812006-06-25 05:49:24 -0700679#ifdef CONFIG_BSD_PROCESS_ACCT
680 struct pacct_struct pacct; /* per-process accounting information */
681#endif
Shailabh Nagarad4ecbc2006-07-14 00:24:44 -0700682#ifdef CONFIG_TASKSTATS
Shailabh Nagarad4ecbc2006-07-14 00:24:44 -0700683 struct taskstats *stats;
684#endif
Miloslav Trmac522ed772007-07-15 23:40:56 -0700685#ifdef CONFIG_AUDIT
686 unsigned audit_tty;
Richard Guy Briggs46e959e2013-05-03 14:03:50 -0400687 unsigned audit_tty_log_passwd;
Miloslav Trmac522ed772007-07-15 23:40:56 -0700688 struct tty_audit_buf *tty_audit_buf;
689#endif
Ben Blum4714d1d2011-05-26 16:25:18 -0700690#ifdef CONFIG_CGROUPS
691 /*
Tejun Heo77e4ef92011-12-12 18:12:21 -0800692 * group_rwsem prevents new tasks from entering the threadgroup and
693 * member tasks from exiting,a more specifically, setting of
694 * PF_EXITING. fork and exit paths are protected with this rwsem
695 * using threadgroup_change_begin/end(). Users which require
696 * threadgroup to remain stable should use threadgroup_[un]lock()
697 * which also takes care of exec path. Currently, cgroup is the
698 * only user.
Ben Blum4714d1d2011-05-26 16:25:18 -0700699 */
Tejun Heo257058a2011-12-12 18:12:21 -0800700 struct rw_semaphore group_rwsem;
Ben Blum4714d1d2011-05-26 16:25:18 -0700701#endif
KOSAKI Motohiro28b83c52009-09-21 17:03:13 -0700702
David Rientjese1e12d22012-12-11 16:02:56 -0800703 oom_flags_t oom_flags;
David Rientjesa9c58b902012-12-11 16:02:54 -0800704 short oom_score_adj; /* OOM kill score adjustment */
705 short oom_score_adj_min; /* OOM kill score adjustment min value.
706 * Only settable by CAP_SYS_RESOURCE. */
KOSAKI Motohiro9b1bf122010-10-27 15:34:08 -0700707
708 struct mutex cred_guard_mutex; /* guard against foreign influences on
709 * credential calculations
710 * (notably. ptrace) */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700711};
712
713/*
714 * Bits in flags field of signal_struct.
715 */
716#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
Oleg Nesterovee77f072011-04-01 20:12:38 +0200717#define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
718#define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
Oleg Nesterov403bad72013-04-30 15:28:10 -0700719#define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
Oleg Nesterove4420552008-04-30 00:52:44 -0700720/*
721 * Pending notifications to parent.
722 */
723#define SIGNAL_CLD_STOPPED 0x00000010
724#define SIGNAL_CLD_CONTINUED 0x00000020
725#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700726
Oleg Nesterovfae5fa42008-04-30 00:53:03 -0700727#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
728
Oleg Nesteroved5d2ca2008-02-04 22:27:24 -0800729/* If true, all threads except ->group_exit_task have pending SIGKILL */
730static inline int signal_group_exit(const struct signal_struct *sig)
731{
732 return (sig->flags & SIGNAL_GROUP_EXIT) ||
733 (sig->group_exit_task != NULL);
734}
735
Linus Torvalds1da177e2005-04-16 15:20:36 -0700736/*
737 * Some day this will be a full-fledged user tracking system..
738 */
739struct user_struct {
740 atomic_t __count; /* reference count */
741 atomic_t processes; /* How many processes does this user have? */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700742 atomic_t sigpending; /* How many pending signals does this user have? */
Amy Griffis2d9048e2006-06-01 13:10:59 -0700743#ifdef CONFIG_INOTIFY_USER
Robert Love0eeca282005-07-12 17:06:03 -0400744 atomic_t inotify_watches; /* How many inotify watches does this user have? */
745 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
746#endif
Eric Paris4afeff82010-10-28 17:21:58 -0400747#ifdef CONFIG_FANOTIFY
748 atomic_t fanotify_listeners;
749#endif
Davide Libenzi7ef99642008-12-01 13:13:55 -0800750#ifdef CONFIG_EPOLL
Robin Holt52bd19f72011-01-12 17:00:01 -0800751 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
Davide Libenzi7ef99642008-12-01 13:13:55 -0800752#endif
Alexey Dobriyan970a8642007-10-16 23:30:09 -0700753#ifdef CONFIG_POSIX_MQUEUE
Linus Torvalds1da177e2005-04-16 15:20:36 -0700754 /* protected by mq_lock */
755 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
Alexey Dobriyan970a8642007-10-16 23:30:09 -0700756#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700757 unsigned long locked_shm; /* How many pages of mlocked shm ? */
758
759#ifdef CONFIG_KEYS
760 struct key *uid_keyring; /* UID specific keyring */
761 struct key *session_keyring; /* UID's default session keyring */
762#endif
763
764 /* Hash table maintenance information */
Pavel Emelyanov735de222007-09-18 22:46:44 -0700765 struct hlist_node uidhash_node;
Eric W. Biederman7b44ab92011-11-16 23:20:58 -0800766 kuid_t uid;
Srivatsa Vaddagiri24e377a2007-10-15 17:00:09 +0200767
Ingo Molnarcdd6c482009-09-21 12:02:48 +0200768#ifdef CONFIG_PERF_EVENTS
Peter Zijlstra789f90f2009-05-15 15:19:27 +0200769 atomic_long_t locked_vm;
770#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700771};
772
Kay Sieverseb41d942007-11-02 13:47:53 +0100773extern int uids_sysfs_init(void);
Dhaval Giani5cb350b2007-10-15 17:00:14 +0200774
Eric W. Biederman7b44ab92011-11-16 23:20:58 -0800775extern struct user_struct *find_user(kuid_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700776
777extern struct user_struct root_user;
778#define INIT_USER (&root_user)
779
David Howellsb6dff3e2008-11-14 10:39:16 +1100780
Linus Torvalds1da177e2005-04-16 15:20:36 -0700781struct backing_dev_info;
782struct reclaim_state;
783
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700784#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700785struct sched_info {
786 /* cumulative counters */
Ingo Molnar2d723762007-10-15 17:00:12 +0200787 unsigned long pcount; /* # of times run on this cpu */
Ken Chen9c2c4802008-12-16 23:41:22 -0800788 unsigned long long run_delay; /* time spent waiting on a runqueue */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700789
790 /* timestamps */
Balbir Singh172ba842007-07-09 18:52:00 +0200791 unsigned long long last_arrival,/* when we last ran on a cpu */
792 last_queued; /* when we were last queued to run */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700793};
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700794#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700795
Shailabh Nagarca74e922006-07-14 00:24:36 -0700796#ifdef CONFIG_TASK_DELAY_ACCT
797struct task_delay_info {
798 spinlock_t lock;
799 unsigned int flags; /* Private per-task flags */
800
801 /* For each stat XXX, add following, aligned appropriately
802 *
803 * struct timespec XXX_start, XXX_end;
804 * u64 XXX_delay;
805 * u32 XXX_count;
806 *
807 * Atomicity of updates to XXX_delay, XXX_count protected by
808 * single lock above (split into XXX_lock if contention is an issue).
809 */
Shailabh Nagar0ff92242006-07-14 00:24:37 -0700810
811 /*
812 * XXX_count is incremented on every XXX operation, the delay
813 * associated with the operation is added to XXX_delay.
814 * XXX_delay contains the accumulated delay time in nanoseconds.
815 */
Thomas Gleixner9667a232014-07-16 21:04:35 +0000816 u64 blkio_start; /* Shared by blkio, swapin */
Shailabh Nagar0ff92242006-07-14 00:24:37 -0700817 u64 blkio_delay; /* wait for sync block io completion */
818 u64 swapin_delay; /* wait for swapin block io completion */
819 u32 blkio_count; /* total count of the number of sync block */
820 /* io operations performed */
821 u32 swapin_count; /* total count of the number of swapin block */
822 /* io operations performed */
Keika Kobayashi873b4772008-07-25 01:48:52 -0700823
Thomas Gleixner9667a232014-07-16 21:04:35 +0000824 u64 freepages_start;
Keika Kobayashi873b4772008-07-25 01:48:52 -0700825 u64 freepages_delay; /* wait for memory reclaim */
826 u32 freepages_count; /* total count of memory reclaim */
Shailabh Nagarca74e922006-07-14 00:24:36 -0700827};
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700828#endif /* CONFIG_TASK_DELAY_ACCT */
829
830static inline int sched_info_on(void)
831{
832#ifdef CONFIG_SCHEDSTATS
833 return 1;
834#elif defined(CONFIG_TASK_DELAY_ACCT)
835 extern int delayacct_on;
836 return delayacct_on;
837#else
838 return 0;
Shailabh Nagarca74e922006-07-14 00:24:36 -0700839#endif
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700840}
Shailabh Nagarca74e922006-07-14 00:24:36 -0700841
Ingo Molnard15bcfd2007-07-09 18:51:57 +0200842enum cpu_idle_type {
843 CPU_IDLE,
844 CPU_NOT_IDLE,
845 CPU_NEWLY_IDLE,
846 CPU_MAX_IDLE_TYPES
Linus Torvalds1da177e2005-04-16 15:20:36 -0700847};
848
849/*
Nicolas Pitreca8ce3d2014-05-26 18:19:39 -0400850 * Increase resolution of cpu_capacity calculations
Nikhil Rao1399fa72011-05-18 10:09:39 -0700851 */
Nicolas Pitreca8ce3d2014-05-26 18:19:39 -0400852#define SCHED_CAPACITY_SHIFT 10
853#define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700854
Nikhil Rao1399fa72011-05-18 10:09:39 -0700855/*
856 * sched-domains (multiprocessor balancing) declarations:
857 */
Peter Williams2dd73a42006-06-27 02:54:34 -0700858#ifdef CONFIG_SMP
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200859#define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
860#define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
861#define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
862#define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
Peter Zijlstrac88d5912009-09-10 13:50:02 +0200863#define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200864#define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
Nicolas Pitre5d4dfdd2014-05-27 13:50:41 -0400865#define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
Vincent Guittotd77b3ed2014-04-11 11:44:40 +0200866#define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200867#define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
868#define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
Michael Neuling532cb4c2010-06-08 14:57:02 +1000869#define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200870#define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
Peter Zijlstrae3589f62011-07-15 10:35:52 +0200871#define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
Mel Gorman3a7053b2013-10-07 11:29:00 +0100872#define SD_NUMA 0x4000 /* cross-node balancing */
Siddha, Suresh B5c45bf22006-06-27 02:54:42 -0700873
Vincent Guittot143e1e22014-04-11 11:44:37 +0200874#ifdef CONFIG_SCHED_SMT
Guenter Roeckb6220ad2014-06-24 18:05:29 -0700875static inline int cpu_smt_flags(void)
Vincent Guittot143e1e22014-04-11 11:44:37 +0200876{
Nicolas Pitre5d4dfdd2014-05-27 13:50:41 -0400877 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
Vincent Guittot143e1e22014-04-11 11:44:37 +0200878}
879#endif
880
881#ifdef CONFIG_SCHED_MC
Guenter Roeckb6220ad2014-06-24 18:05:29 -0700882static inline int cpu_core_flags(void)
Vincent Guittot143e1e22014-04-11 11:44:37 +0200883{
884 return SD_SHARE_PKG_RESOURCES;
885}
886#endif
887
888#ifdef CONFIG_NUMA
Guenter Roeckb6220ad2014-06-24 18:05:29 -0700889static inline int cpu_numa_flags(void)
Vincent Guittot143e1e22014-04-11 11:44:37 +0200890{
891 return SD_NUMA;
892}
893#endif
Michael Neuling532cb4c2010-06-08 14:57:02 +1000894
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +0900895struct sched_domain_attr {
896 int relax_domain_level;
897};
898
899#define SD_ATTR_INIT (struct sched_domain_attr) { \
900 .relax_domain_level = -1, \
901}
902
Peter Zijlstra60495e72011-04-07 14:10:04 +0200903extern int sched_domain_level_max;
904
Li Zefan5e6521e2013-03-05 16:06:23 +0800905struct sched_group;
906
Linus Torvalds1da177e2005-04-16 15:20:36 -0700907struct sched_domain {
908 /* These fields must be setup */
909 struct sched_domain *parent; /* top domain must be null terminated */
Siddha, Suresh B1a848872006-10-03 01:14:08 -0700910 struct sched_domain *child; /* bottom domain must be null terminated */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700911 struct sched_group *groups; /* the balancing groups of the domain */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700912 unsigned long min_interval; /* Minimum balance interval ms */
913 unsigned long max_interval; /* Maximum balance interval ms */
914 unsigned int busy_factor; /* less balancing by factor if busy */
915 unsigned int imbalance_pct; /* No balance until over watermark */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700916 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
Nick Piggin78979862005-06-25 14:57:13 -0700917 unsigned int busy_idx;
918 unsigned int idle_idx;
919 unsigned int newidle_idx;
920 unsigned int wake_idx;
Nick Piggin147cbb42005-06-25 14:57:19 -0700921 unsigned int forkexec_idx;
Peter Zijlstraa52bfd72009-09-01 10:34:35 +0200922 unsigned int smt_gain;
Vincent Guittot25f55d92013-04-23 16:59:02 +0200923
924 int nohz_idle; /* NOHZ IDLE status */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700925 int flags; /* See SD_* */
Peter Zijlstra60495e72011-04-07 14:10:04 +0200926 int level;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700927
928 /* Runtime fields. */
929 unsigned long last_balance; /* init to jiffies. units in jiffies */
930 unsigned int balance_interval; /* initialise to 1. units in ms. */
931 unsigned int nr_balance_failed; /* initialise to 0 */
932
Jason Lowf48627e2013-09-13 11:26:53 -0700933 /* idle_balance() stats */
Jason Low9bd721c2013-09-13 11:26:52 -0700934 u64 max_newidle_lb_cost;
Jason Lowf48627e2013-09-13 11:26:53 -0700935 unsigned long next_decay_max_lb_cost;
Peter Zijlstra2398f2c2008-06-27 13:41:35 +0200936
Linus Torvalds1da177e2005-04-16 15:20:36 -0700937#ifdef CONFIG_SCHEDSTATS
938 /* load_balance() stats */
Ken Chen480b9432007-10-18 21:32:56 +0200939 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
940 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
941 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
942 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
943 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
944 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
945 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
946 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700947
948 /* Active load balancing */
Ken Chen480b9432007-10-18 21:32:56 +0200949 unsigned int alb_count;
950 unsigned int alb_failed;
951 unsigned int alb_pushed;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700952
Nick Piggin68767a02005-06-25 14:57:20 -0700953 /* SD_BALANCE_EXEC stats */
Ken Chen480b9432007-10-18 21:32:56 +0200954 unsigned int sbe_count;
955 unsigned int sbe_balanced;
956 unsigned int sbe_pushed;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700957
Nick Piggin68767a02005-06-25 14:57:20 -0700958 /* SD_BALANCE_FORK stats */
Ken Chen480b9432007-10-18 21:32:56 +0200959 unsigned int sbf_count;
960 unsigned int sbf_balanced;
961 unsigned int sbf_pushed;
Nick Piggin68767a02005-06-25 14:57:20 -0700962
Linus Torvalds1da177e2005-04-16 15:20:36 -0700963 /* try_to_wake_up() stats */
Ken Chen480b9432007-10-18 21:32:56 +0200964 unsigned int ttwu_wake_remote;
965 unsigned int ttwu_move_affine;
966 unsigned int ttwu_move_balance;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700967#endif
Ingo Molnara5d8c342008-10-09 11:35:51 +0200968#ifdef CONFIG_SCHED_DEBUG
969 char *name;
970#endif
Peter Zijlstradce840a2011-04-07 14:09:50 +0200971 union {
972 void *private; /* used during construction */
973 struct rcu_head rcu; /* used during destruction */
974 };
Rusty Russell6c99e9a2008-11-25 02:35:04 +1030975
Peter Zijlstra669c55e2010-04-16 14:59:29 +0200976 unsigned int span_weight;
Ingo Molnar4200efd2009-05-19 09:22:19 +0200977 /*
978 * Span of all CPUs in this domain.
979 *
980 * NOTE: this field is variable length. (Allocated dynamically
981 * by attaching extra space to the end of the structure,
982 * depending on how many CPUs the kernel has booted up with)
Ingo Molnar4200efd2009-05-19 09:22:19 +0200983 */
984 unsigned long span[0];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700985};
986
Rusty Russell758b2cd2008-11-25 02:35:04 +1030987static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
988{
Rusty Russell6c99e9a2008-11-25 02:35:04 +1030989 return to_cpumask(sd->span);
Rusty Russell758b2cd2008-11-25 02:35:04 +1030990}
991
Rusty Russellacc3f5d2009-11-03 14:53:40 +1030992extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +0900993 struct sched_domain_attr *dattr_new);
Paul Jackson029190c2007-10-18 23:40:20 -0700994
Rusty Russellacc3f5d2009-11-03 14:53:40 +1030995/* Allocate an array of sched domains, for partition_sched_domains(). */
996cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
997void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
998
Peter Zijlstra39be3502012-01-26 12:44:34 +0100999bool cpus_share_cache(int this_cpu, int that_cpu);
1000
Vincent Guittot143e1e22014-04-11 11:44:37 +02001001typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
Guenter Roeckb6220ad2014-06-24 18:05:29 -07001002typedef int (*sched_domain_flags_f)(void);
Vincent Guittot143e1e22014-04-11 11:44:37 +02001003
1004#define SDTL_OVERLAP 0x01
1005
1006struct sd_data {
1007 struct sched_domain **__percpu sd;
1008 struct sched_group **__percpu sg;
Nicolas Pitre63b2ca32014-05-26 18:19:37 -04001009 struct sched_group_capacity **__percpu sgc;
Vincent Guittot143e1e22014-04-11 11:44:37 +02001010};
1011
1012struct sched_domain_topology_level {
1013 sched_domain_mask_f mask;
1014 sched_domain_flags_f sd_flags;
1015 int flags;
1016 int numa_level;
1017 struct sd_data data;
1018#ifdef CONFIG_SCHED_DEBUG
1019 char *name;
1020#endif
1021};
1022
1023extern struct sched_domain_topology_level *sched_domain_topology;
1024
1025extern void set_sched_topology(struct sched_domain_topology_level *tl);
1026
1027#ifdef CONFIG_SCHED_DEBUG
1028# define SD_INIT_NAME(type) .name = #type
1029#else
1030# define SD_INIT_NAME(type)
1031#endif
1032
Ingo Molnar1b427c12008-07-18 14:01:39 +02001033#else /* CONFIG_SMP */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001034
Ingo Molnar1b427c12008-07-18 14:01:39 +02001035struct sched_domain_attr;
1036
1037static inline void
Rusty Russellacc3f5d2009-11-03 14:53:40 +10301038partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
Ingo Molnar1b427c12008-07-18 14:01:39 +02001039 struct sched_domain_attr *dattr_new)
Con Kolivasd02c7a82007-07-26 13:40:43 +02001040{
Con Kolivasd02c7a82007-07-26 13:40:43 +02001041}
Peter Zijlstra39be3502012-01-26 12:44:34 +01001042
1043static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1044{
1045 return true;
1046}
1047
Ingo Molnar1b427c12008-07-18 14:01:39 +02001048#endif /* !CONFIG_SMP */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001049
Peter Zijlstra47fe38f2009-09-02 13:49:18 +02001050
Linus Torvalds1da177e2005-04-16 15:20:36 -07001051struct io_context; /* See blkdev.h */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001052
Linus Torvalds1da177e2005-04-16 15:20:36 -07001053
Chen, Kenneth W383f2832005-09-09 13:02:02 -07001054#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
Ingo Molnar36c8b582006-07-03 00:25:41 -07001055extern void prefetch_stack(struct task_struct *t);
Chen, Kenneth W383f2832005-09-09 13:02:02 -07001056#else
1057static inline void prefetch_stack(struct task_struct *t) { }
1058#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001059
1060struct audit_context; /* See audit.c */
1061struct mempolicy;
Jens Axboeb92ce552006-04-11 13:52:07 +02001062struct pipe_inode_info;
Serge E. Hallyn4865ecf2006-10-02 02:18:14 -07001063struct uts_namespace;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001064
Ingo Molnar20b8a592007-07-09 18:51:58 +02001065struct load_weight {
Peter Zijlstra9dbdb152013-11-18 18:27:06 +01001066 unsigned long weight;
1067 u32 inv_weight;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001068};
1069
Paul Turner9d85f212012-10-04 13:18:29 +02001070struct sched_avg {
1071 /*
1072 * These sums represent an infinite geometric series and so are bound
Kamalesh Babulal239003e2013-06-27 11:34:09 +05301073 * above by 1024/(1-y). Thus we only need a u32 to store them for all
Paul Turner9d85f212012-10-04 13:18:29 +02001074 * choices of y < 1-2^(-32)*1024.
1075 */
1076 u32 runnable_avg_sum, runnable_avg_period;
1077 u64 last_runnable_update;
Paul Turner9ee474f2012-10-04 13:18:30 +02001078 s64 decay_count;
Paul Turner2dac7542012-10-04 13:18:30 +02001079 unsigned long load_avg_contrib;
Paul Turner9d85f212012-10-04 13:18:29 +02001080};
1081
Ingo Molnar94c18222007-08-02 17:41:40 +02001082#ifdef CONFIG_SCHEDSTATS
Lucas De Marchi41acab82010-03-10 23:37:45 -03001083struct sched_statistics {
Ingo Molnar94c18222007-08-02 17:41:40 +02001084 u64 wait_start;
1085 u64 wait_max;
Arjan van de Ven6d082592008-01-25 21:08:35 +01001086 u64 wait_count;
1087 u64 wait_sum;
Arjan van de Ven8f0dfc32009-07-20 11:26:58 -07001088 u64 iowait_count;
1089 u64 iowait_sum;
Ingo Molnar94c18222007-08-02 17:41:40 +02001090
1091 u64 sleep_start;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001092 u64 sleep_max;
Ingo Molnar94c18222007-08-02 17:41:40 +02001093 s64 sum_sleep_runtime;
1094
1095 u64 block_start;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001096 u64 block_max;
1097 u64 exec_max;
Ingo Molnareba1ed42007-10-15 17:00:02 +02001098 u64 slice_max;
Ingo Molnarcc367732007-10-15 17:00:18 +02001099
Ingo Molnarcc367732007-10-15 17:00:18 +02001100 u64 nr_migrations_cold;
1101 u64 nr_failed_migrations_affine;
1102 u64 nr_failed_migrations_running;
1103 u64 nr_failed_migrations_hot;
1104 u64 nr_forced_migrations;
Ingo Molnarcc367732007-10-15 17:00:18 +02001105
1106 u64 nr_wakeups;
1107 u64 nr_wakeups_sync;
1108 u64 nr_wakeups_migrate;
1109 u64 nr_wakeups_local;
1110 u64 nr_wakeups_remote;
1111 u64 nr_wakeups_affine;
1112 u64 nr_wakeups_affine_attempts;
1113 u64 nr_wakeups_passive;
1114 u64 nr_wakeups_idle;
Lucas De Marchi41acab82010-03-10 23:37:45 -03001115};
1116#endif
1117
1118struct sched_entity {
1119 struct load_weight load; /* for load-balancing */
1120 struct rb_node run_node;
1121 struct list_head group_node;
1122 unsigned int on_rq;
1123
1124 u64 exec_start;
1125 u64 sum_exec_runtime;
1126 u64 vruntime;
1127 u64 prev_sum_exec_runtime;
1128
Lucas De Marchi41acab82010-03-10 23:37:45 -03001129 u64 nr_migrations;
1130
Lucas De Marchi41acab82010-03-10 23:37:45 -03001131#ifdef CONFIG_SCHEDSTATS
1132 struct sched_statistics statistics;
Ingo Molnar94c18222007-08-02 17:41:40 +02001133#endif
1134
Ingo Molnar20b8a592007-07-09 18:51:58 +02001135#ifdef CONFIG_FAIR_GROUP_SCHED
Peter Zijlstrafed14d42012-02-11 06:05:00 +01001136 int depth;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001137 struct sched_entity *parent;
1138 /* rq on which this entity is (to be) queued: */
1139 struct cfs_rq *cfs_rq;
1140 /* rq "owned" by this entity/group: */
1141 struct cfs_rq *my_q;
1142#endif
Clark Williams8bd75c72013-02-07 09:47:07 -06001143
Alex Shi141965c2013-06-26 13:05:39 +08001144#ifdef CONFIG_SMP
Paul Turnerf4e26b12012-10-04 13:18:32 +02001145 /* Per-entity load-tracking */
Paul Turner9d85f212012-10-04 13:18:29 +02001146 struct sched_avg avg;
1147#endif
Ingo Molnar20b8a592007-07-09 18:51:58 +02001148};
Ingo Molnar70b97a72006-07-03 00:25:42 -07001149
Peter Zijlstrafa717062008-01-25 21:08:27 +01001150struct sched_rt_entity {
1151 struct list_head run_list;
Peter Zijlstra78f2c7d2008-01-25 21:08:27 +01001152 unsigned long timeout;
Ying Xue57d2aa02012-07-17 15:03:43 +08001153 unsigned long watchdog_stamp;
Richard Kennedybee367e2008-08-01 13:24:08 +01001154 unsigned int time_slice;
Peter Zijlstra6f505b12008-01-25 21:08:30 +01001155
Peter Zijlstra58d6c2d2008-04-19 19:45:00 +02001156 struct sched_rt_entity *back;
Peter Zijlstra052f1dc2008-02-13 15:45:40 +01001157#ifdef CONFIG_RT_GROUP_SCHED
Peter Zijlstra6f505b12008-01-25 21:08:30 +01001158 struct sched_rt_entity *parent;
1159 /* rq on which this entity is (to be) queued: */
1160 struct rt_rq *rt_rq;
1161 /* rq "owned" by this entity/group: */
1162 struct rt_rq *my_q;
1163#endif
Peter Zijlstrafa717062008-01-25 21:08:27 +01001164};
1165
Dario Faggioliaab03e02013-11-28 11:14:43 +01001166struct sched_dl_entity {
1167 struct rb_node rb_node;
1168
1169 /*
1170 * Original scheduling parameters. Copied here from sched_attr
xiaofeng.yan4027d082014-05-09 03:21:27 +00001171 * during sched_setattr(), they will remain the same until
1172 * the next sched_setattr().
Dario Faggioliaab03e02013-11-28 11:14:43 +01001173 */
1174 u64 dl_runtime; /* maximum runtime for each instance */
1175 u64 dl_deadline; /* relative deadline of each instance */
Harald Gustafsson755378a2013-11-07 14:43:40 +01001176 u64 dl_period; /* separation of two instances (period) */
Dario Faggioli332ac172013-11-07 14:43:45 +01001177 u64 dl_bw; /* dl_runtime / dl_deadline */
Dario Faggioliaab03e02013-11-28 11:14:43 +01001178
1179 /*
1180 * Actual scheduling parameters. Initialized with the values above,
1181 * they are continously updated during task execution. Note that
1182 * the remaining runtime could be < 0 in case we are in overrun.
1183 */
1184 s64 runtime; /* remaining runtime for this instance */
1185 u64 deadline; /* absolute deadline for this instance */
1186 unsigned int flags; /* specifying the scheduler behaviour */
1187
1188 /*
1189 * Some bool flags:
1190 *
1191 * @dl_throttled tells if we exhausted the runtime. If so, the
1192 * task has to wait for a replenishment to be performed at the
1193 * next firing of dl_timer.
1194 *
1195 * @dl_new tells if a new instance arrived. If so we must
1196 * start executing it with full runtime and reset its absolute
1197 * deadline;
Dario Faggioli2d3d8912013-11-07 14:43:44 +01001198 *
1199 * @dl_boosted tells if we are boosted due to DI. If so we are
1200 * outside bandwidth enforcement mechanism (but only until we
Juri Lelli5bfd1262014-04-15 13:49:04 +02001201 * exit the critical section);
1202 *
1203 * @dl_yielded tells if task gave up the cpu before consuming
1204 * all its available runtime during the last job.
Dario Faggioliaab03e02013-11-28 11:14:43 +01001205 */
Juri Lelli5bfd1262014-04-15 13:49:04 +02001206 int dl_throttled, dl_new, dl_boosted, dl_yielded;
Dario Faggioliaab03e02013-11-28 11:14:43 +01001207
1208 /*
1209 * Bandwidth enforcement timer. Each -deadline task has its
1210 * own bandwidth to be enforced, thus we need one timer per task.
1211 */
1212 struct hrtimer dl_timer;
1213};
Clark Williams8bd75c72013-02-07 09:47:07 -06001214
Paul E. McKenney86848962009-08-27 15:00:12 -07001215struct rcu_node;
1216
Peter Zijlstra8dc85d52010-09-02 16:50:03 +02001217enum perf_event_task_context {
1218 perf_invalid_context = -1,
1219 perf_hw_context = 0,
Peter Zijlstra89a1e182010-09-07 17:34:50 +02001220 perf_sw_context,
Peter Zijlstra8dc85d52010-09-02 16:50:03 +02001221 perf_nr_task_contexts,
1222};
1223
Linus Torvalds1da177e2005-04-16 15:20:36 -07001224struct task_struct {
1225 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
Roman Zippelf7e42172007-05-09 02:35:17 -07001226 void *stack;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001227 atomic_t usage;
William Cohen97dc32c2007-05-08 00:23:41 -07001228 unsigned int flags; /* per process flags, defined below */
1229 unsigned int ptrace;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001230
Peter Williams2dd73a42006-06-27 02:54:34 -07001231#ifdef CONFIG_SMP
Peter Zijlstrafa14ff42011-09-12 13:06:17 +02001232 struct llist_node wake_entry;
Peter Zijlstra3ca7a442011-04-05 17:23:40 +02001233 int on_cpu;
Michael Wang62470412013-07-04 12:55:51 +08001234 struct task_struct *last_wakee;
1235 unsigned long wakee_flips;
1236 unsigned long wakee_flip_decay_ts;
Peter Zijlstraac66f542013-10-07 11:29:16 +01001237
1238 int wake_cpu;
Nick Piggin4866cde2005-06-25 14:57:23 -07001239#endif
Peter Zijlstrafd2f4412011-04-05 17:23:44 +02001240 int on_rq;
Ingo Molnar50e645a2007-07-09 18:52:00 +02001241
Ingo Molnarb29739f2006-06-27 02:54:51 -07001242 int prio, static_prio, normal_prio;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001243 unsigned int rt_priority;
Ingo Molnar5522d5d2007-10-15 17:00:12 +02001244 const struct sched_class *sched_class;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001245 struct sched_entity se;
Peter Zijlstrafa717062008-01-25 21:08:27 +01001246 struct sched_rt_entity rt;
Peter Zijlstra8323f262012-06-22 13:36:05 +02001247#ifdef CONFIG_CGROUP_SCHED
1248 struct task_group *sched_task_group;
1249#endif
Dario Faggioliaab03e02013-11-28 11:14:43 +01001250 struct sched_dl_entity dl;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001251
Avi Kivitye107be32007-07-26 13:40:43 +02001252#ifdef CONFIG_PREEMPT_NOTIFIERS
1253 /* list of struct preempt_notifier: */
1254 struct hlist_head preempt_notifiers;
1255#endif
1256
Alexey Dobriyan6c5c9342006-09-29 01:59:40 -07001257#ifdef CONFIG_BLK_DEV_IO_TRACE
Jens Axboe2056a782006-03-23 20:00:26 +01001258 unsigned int btrace_seq;
Alexey Dobriyan6c5c9342006-09-29 01:59:40 -07001259#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001260
William Cohen97dc32c2007-05-08 00:23:41 -07001261 unsigned int policy;
Peter Zijlstra29baa742012-04-23 12:11:21 +02001262 int nr_cpus_allowed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001263 cpumask_t cpus_allowed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001264
Paul E. McKenneya57eb942010-06-29 16:49:16 -07001265#ifdef CONFIG_PREEMPT_RCU
Paul E. McKenneye260be62008-01-25 21:08:24 +01001266 int rcu_read_lock_nesting;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001267 char rcu_read_unlock_special;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001268 struct list_head rcu_node_entry;
Paul E. McKenneya57eb942010-06-29 16:49:16 -07001269#endif /* #ifdef CONFIG_PREEMPT_RCU */
1270#ifdef CONFIG_TREE_PREEMPT_RCU
1271 struct rcu_node *rcu_blocked_node;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001272#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
Paul E. McKenney8315f422014-06-27 13:42:20 -07001273#ifdef CONFIG_TASKS_RCU
1274 unsigned long rcu_tasks_nvcsw;
1275 bool rcu_tasks_holdout;
1276 struct list_head rcu_tasks_holdout_list;
1277#endif /* #ifdef CONFIG_TASKS_RCU */
Paul E. McKenneye260be62008-01-25 21:08:24 +01001278
Chandra Seetharaman52f17b62006-07-14 00:24:38 -07001279#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001280 struct sched_info sched_info;
1281#endif
1282
1283 struct list_head tasks;
Dario Faggioli806c09a2010-11-30 19:51:33 +01001284#ifdef CONFIG_SMP
Gregory Haskins917b6272008-12-29 09:39:53 -05001285 struct plist_node pushable_tasks;
Juri Lelli1baca4c2013-11-07 14:43:38 +01001286 struct rb_node pushable_dl_tasks;
Dario Faggioli806c09a2010-11-30 19:51:33 +01001287#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001288
1289 struct mm_struct *mm, *active_mm;
Jiri Kosina4471a672011-04-14 15:22:09 -07001290#ifdef CONFIG_COMPAT_BRK
1291 unsigned brk_randomized:1;
1292#endif
Davidlohr Bueso615d6e82014-04-07 15:37:25 -07001293 /* per-thread vma caching */
1294 u32 vmacache_seqnum;
1295 struct vm_area_struct *vmacache[VMACACHE_SIZE];
KAMEZAWA Hiroyuki34e55232010-03-05 13:41:40 -08001296#if defined(SPLIT_RSS_COUNTING)
1297 struct task_rss_stat rss_stat;
1298#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001299/* task state */
William Cohen97dc32c2007-05-08 00:23:41 -07001300 int exit_state;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001301 int exit_code, exit_signal;
1302 int pdeath_signal; /* The signal sent when the parent dies */
Tejun Heoa8f072c2011-06-02 11:13:59 +02001303 unsigned int jobctl; /* JOBCTL_*, siglock protected */
Andrei Epure9b89f6b2013-04-11 20:30:29 +03001304
1305 /* Used for emulating ABI behavior of previous Linux versions */
William Cohen97dc32c2007-05-08 00:23:41 -07001306 unsigned int personality;
Andrei Epure9b89f6b2013-04-11 20:30:29 +03001307
Kentaro Takedaf9ce1f12009-02-05 17:18:11 +09001308 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1309 * execve */
Arjan van de Ven8f0dfc32009-07-20 11:26:58 -07001310 unsigned in_iowait:1;
1311
Lennart Poetteringca94c442009-06-15 17:17:47 +02001312 /* Revert to default priority/policy when forking */
1313 unsigned sched_reset_on_fork:1;
Peter Zijlstraa8e4f2e2011-04-05 17:23:49 +02001314 unsigned sched_contributes_to_load:1;
Lennart Poetteringca94c442009-06-15 17:17:47 +02001315
Kees Cook1d4457f2014-05-21 15:23:46 -07001316 unsigned long atomic_flags; /* Flags needing atomic access. */
1317
Linus Torvalds1da177e2005-04-16 15:20:36 -07001318 pid_t pid;
1319 pid_t tgid;
Arjan van de Ven0a425402006-09-26 10:52:38 +02001320
Hiroshi Shimamoto13145622009-08-18 15:06:02 +09001321#ifdef CONFIG_CC_STACKPROTECTOR
Arjan van de Ven0a425402006-09-26 10:52:38 +02001322 /* Canary value for the -fstack-protector gcc feature */
1323 unsigned long stack_canary;
Hiroshi Shimamoto13145622009-08-18 15:06:02 +09001324#endif
Oleg Nesterov4d1d61a2012-05-11 10:59:08 +10001325 /*
Linus Torvalds1da177e2005-04-16 15:20:36 -07001326 * pointers to (original) parent process, youngest child, younger sibling,
Oleg Nesterov4d1d61a2012-05-11 10:59:08 +10001327 * older sibling, respectively. (p->father can be replaced with
Roland McGrathf4700212008-03-24 18:36:23 -07001328 * p->real_parent->pid)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001329 */
Kees Cookabd63bc2011-12-14 14:39:26 -08001330 struct task_struct __rcu *real_parent; /* real parent process */
1331 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001332 /*
Roland McGrathf4700212008-03-24 18:36:23 -07001333 * children/sibling forms the list of my natural children
Linus Torvalds1da177e2005-04-16 15:20:36 -07001334 */
1335 struct list_head children; /* list of my children */
1336 struct list_head sibling; /* linkage in my parent's children list */
1337 struct task_struct *group_leader; /* threadgroup leader */
1338
Roland McGrathf4700212008-03-24 18:36:23 -07001339 /*
1340 * ptraced is the list of tasks this task is using ptrace on.
1341 * This includes both natural children and PTRACE_ATTACH targets.
1342 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1343 */
1344 struct list_head ptraced;
1345 struct list_head ptrace_entry;
1346
Linus Torvalds1da177e2005-04-16 15:20:36 -07001347 /* PID/PID hash table linkage. */
Eric W. Biederman92476d72006-03-31 02:31:42 -08001348 struct pid_link pids[PIDTYPE_MAX];
Oleg Nesterov47e65322006-03-28 16:11:25 -08001349 struct list_head thread_group;
Oleg Nesterov0c740d02014-01-21 15:49:56 -08001350 struct list_head thread_node;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001351
1352 struct completion *vfork_done; /* for vfork() */
1353 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1354 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1355
Michael Neulingc66f08b2007-10-18 03:06:34 -07001356 cputime_t utime, stime, utimescaled, stimescaled;
Laurent Vivier9ac52312007-10-15 17:00:19 +02001357 cputime_t gtime;
Frederic Weisbecker9fbc42e2013-02-25 17:25:39 +01001358#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
Frederic Weisbeckerd37f761d2012-11-22 00:58:35 +01001359 struct cputime prev_cputime;
Hidetoshi Setod99ca3b2009-12-02 17:26:47 +09001360#endif
Frederic Weisbecker6a616712012-12-16 20:00:34 +01001361#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1362 seqlock_t vtime_seqlock;
1363 unsigned long long vtime_snap;
1364 enum {
1365 VTIME_SLEEPING = 0,
1366 VTIME_USER,
1367 VTIME_SYS,
1368 } vtime_snap_whence;
1369#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001370 unsigned long nvcsw, nivcsw; /* context switch counts */
Thomas Gleixnerccbf62d2014-07-16 21:04:34 +00001371 u64 start_time; /* monotonic time in nsec */
Thomas Gleixner57e0be02014-07-16 21:04:32 +00001372 u64 real_start_time; /* boot based time in nsec */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001373/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1374 unsigned long min_flt, maj_flt;
1375
Frank Mayharf06febc2008-09-12 09:54:39 -07001376 struct task_cputime cputime_expires;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001377 struct list_head cpu_timers[3];
1378
1379/* process credentials */
Arnd Bergmann1b0ba1c2010-02-24 19:45:09 +01001380 const struct cred __rcu *real_cred; /* objective and real subjective task
David Howells3b11a1d2008-11-14 10:39:26 +11001381 * credentials (COW) */
Arnd Bergmann1b0ba1c2010-02-24 19:45:09 +01001382 const struct cred __rcu *cred; /* effective (overridable) subjective task
David Howells3b11a1d2008-11-14 10:39:26 +11001383 * credentials (COW) */
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -07001384 char comm[TASK_COMM_LEN]; /* executable name excluding path
1385 - access with [gs]et_task_comm (which lock
1386 it with task_lock())
Linus Torvalds221af7f2010-01-28 22:14:42 -08001387 - initialized normally by setup_new_exec */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001388/* file system info */
1389 int link_count, total_link_count;
Alexey Dobriyan3d5b6fc2006-09-29 01:59:40 -07001390#ifdef CONFIG_SYSVIPC
Linus Torvalds1da177e2005-04-16 15:20:36 -07001391/* ipc stuff */
1392 struct sysv_sem sysvsem;
Jack Millerab602f72014-08-08 14:23:19 -07001393 struct sysv_shm sysvshm;
Alexey Dobriyan3d5b6fc2006-09-29 01:59:40 -07001394#endif
Mandeep Singh Bainese162b392009-01-15 11:08:40 -08001395#ifdef CONFIG_DETECT_HUNG_TASK
Ingo Molnar82a1fcb2008-01-25 21:08:02 +01001396/* hung task detection */
Ingo Molnar82a1fcb2008-01-25 21:08:02 +01001397 unsigned long last_switch_count;
1398#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001399/* CPU-specific state of this task */
1400 struct thread_struct thread;
1401/* filesystem information */
1402 struct fs_struct *fs;
1403/* open file information */
1404 struct files_struct *files;
Serge E. Hallyn1651e142006-10-02 02:18:08 -07001405/* namespaces */
Serge E. Hallynab516012006-10-02 02:18:06 -07001406 struct nsproxy *nsproxy;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001407/* signal handlers */
1408 struct signal_struct *signal;
1409 struct sighand_struct *sighand;
1410
1411 sigset_t blocked, real_blocked;
Roland McGrathf3de2722008-04-30 00:53:09 -07001412 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001413 struct sigpending pending;
1414
1415 unsigned long sas_ss_sp;
1416 size_t sas_ss_size;
1417 int (*notifier)(void *priv);
1418 void *notifier_data;
1419 sigset_t *notifier_mask;
Al Viro67d12142012-06-27 11:07:19 +04001420 struct callback_head *task_works;
Oleg Nesterove73f8952012-05-11 10:59:07 +10001421
Linus Torvalds1da177e2005-04-16 15:20:36 -07001422 struct audit_context *audit_context;
Al Virobfef93a2008-01-10 04:53:18 -05001423#ifdef CONFIG_AUDITSYSCALL
Eric W. Biedermane1760bd2012-09-10 22:39:43 -07001424 kuid_t loginuid;
Eric Paris4746ec52008-01-08 10:06:53 -05001425 unsigned int sessionid;
Al Virobfef93a2008-01-10 04:53:18 -05001426#endif
Will Drewry932eceb2012-04-12 16:47:54 -05001427 struct seccomp seccomp;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001428
1429/* Thread group tracking */
1430 u32 parent_exec_id;
1431 u32 self_exec_id;
Miao Xie58568d22009-06-16 15:31:49 -07001432/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1433 * mempolicy */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001434 spinlock_t alloc_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001435
Ingo Molnarb29739f2006-06-27 02:54:51 -07001436 /* Protection of the PI data structures: */
Thomas Gleixner1d615482009-11-17 14:54:03 +01001437 raw_spinlock_t pi_lock;
Ingo Molnarb29739f2006-06-27 02:54:51 -07001438
Ingo Molnar23f78d4a2006-06-27 02:54:53 -07001439#ifdef CONFIG_RT_MUTEXES
1440 /* PI waiters blocked on a rt_mutex held by this task */
Peter Zijlstrafb00aca2013-11-07 14:43:43 +01001441 struct rb_root pi_waiters;
1442 struct rb_node *pi_waiters_leftmost;
Ingo Molnar23f78d4a2006-06-27 02:54:53 -07001443 /* Deadlock detection and priority inheritance handling */
1444 struct rt_mutex_waiter *pi_blocked_on;
Ingo Molnar23f78d4a2006-06-27 02:54:53 -07001445#endif
1446
Ingo Molnar408894e2006-01-09 15:59:20 -08001447#ifdef CONFIG_DEBUG_MUTEXES
1448 /* mutex deadlock detection */
1449 struct mutex_waiter *blocked_on;
1450#endif
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001451#ifdef CONFIG_TRACE_IRQFLAGS
1452 unsigned int irq_events;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001453 unsigned long hardirq_enable_ip;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001454 unsigned long hardirq_disable_ip;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001455 unsigned int hardirq_enable_event;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001456 unsigned int hardirq_disable_event;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001457 int hardirqs_enabled;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001458 int hardirq_context;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001459 unsigned long softirq_disable_ip;
1460 unsigned long softirq_enable_ip;
1461 unsigned int softirq_disable_event;
1462 unsigned int softirq_enable_event;
1463 int softirqs_enabled;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001464 int softirq_context;
1465#endif
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001466#ifdef CONFIG_LOCKDEP
Peter Zijlstrabdb94412008-02-25 23:02:48 +01001467# define MAX_LOCK_DEPTH 48UL
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001468 u64 curr_chain_key;
1469 int lockdep_depth;
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001470 unsigned int lockdep_recursion;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001471 struct held_lock held_locks[MAX_LOCK_DEPTH];
Nick Piggincf40bd12009-01-21 08:12:39 +01001472 gfp_t lockdep_reclaim_gfp;
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001473#endif
Ingo Molnar408894e2006-01-09 15:59:20 -08001474
Linus Torvalds1da177e2005-04-16 15:20:36 -07001475/* journalling filesystem info */
1476 void *journal_info;
1477
Neil Brownd89d8792007-05-01 09:53:42 +02001478/* stacked block device info */
Akinobu Mitabddd87c2010-02-23 08:55:42 +01001479 struct bio_list *bio_list;
Neil Brownd89d8792007-05-01 09:53:42 +02001480
Jens Axboe73c10102011-03-08 13:19:51 +01001481#ifdef CONFIG_BLOCK
1482/* stack plugging */
1483 struct blk_plug *plug;
1484#endif
1485
Linus Torvalds1da177e2005-04-16 15:20:36 -07001486/* VM state */
1487 struct reclaim_state *reclaim_state;
1488
Linus Torvalds1da177e2005-04-16 15:20:36 -07001489 struct backing_dev_info *backing_dev_info;
1490
1491 struct io_context *io_context;
1492
1493 unsigned long ptrace_message;
1494 siginfo_t *last_siginfo; /* For ptrace use. */
Andrew Morton7c3ab732006-12-10 02:19:19 -08001495 struct task_io_accounting ioac;
Jay Lan8f0ab512006-09-30 23:28:59 -07001496#if defined(CONFIG_TASK_XACCT)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001497 u64 acct_rss_mem1; /* accumulated rss usage */
1498 u64 acct_vm_mem1; /* accumulated virtual memory usage */
Jonathan Lim49b5cf32008-07-25 01:48:40 -07001499 cputime_t acct_timexpd; /* stime + utime since last update */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001500#endif
1501#ifdef CONFIG_CPUSETS
Miao Xie58568d22009-06-16 15:31:49 -07001502 nodemask_t mems_allowed; /* Protected by alloc_lock */
Mel Gormancc9a6c82012-03-21 16:34:11 -07001503 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
Paul Jackson825a46a2006-03-24 03:16:03 -08001504 int cpuset_mem_spread_rotor;
Jack Steiner6adef3e2010-05-26 14:42:49 -07001505 int cpuset_slab_spread_rotor;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001506#endif
Paul Menageddbcc7e2007-10-18 23:39:30 -07001507#ifdef CONFIG_CGROUPS
Paul Menage817929e2007-10-18 23:39:36 -07001508 /* Control Group info protected by css_set_lock */
Arnd Bergmann2c392b82010-02-24 19:41:39 +01001509 struct css_set __rcu *cgroups;
Paul Menage817929e2007-10-18 23:39:36 -07001510 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1511 struct list_head cg_list;
Paul Menageddbcc7e2007-10-18 23:39:30 -07001512#endif
Alexey Dobriyan42b2dd02007-10-16 23:27:30 -07001513#ifdef CONFIG_FUTEX
Ingo Molnar0771dfe2006-03-27 01:16:22 -08001514 struct robust_list_head __user *robust_list;
Ingo Molnar34f192c2006-03-27 01:16:24 -08001515#ifdef CONFIG_COMPAT
1516 struct compat_robust_list_head __user *compat_robust_list;
1517#endif
Ingo Molnarc87e2832006-06-27 02:54:58 -07001518 struct list_head pi_state_list;
1519 struct futex_pi_state *pi_state_cache;
Alexey Dobriyan42b2dd02007-10-16 23:27:30 -07001520#endif
Ingo Molnarcdd6c482009-09-21 12:02:48 +02001521#ifdef CONFIG_PERF_EVENTS
Peter Zijlstra8dc85d52010-09-02 16:50:03 +02001522 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
Ingo Molnarcdd6c482009-09-21 12:02:48 +02001523 struct mutex perf_event_mutex;
1524 struct list_head perf_event_list;
Paul Mackerrasa63eaf32009-05-22 14:17:31 +10001525#endif
Thomas Gleixner8f47b182014-02-07 20:58:39 +01001526#ifdef CONFIG_DEBUG_PREEMPT
1527 unsigned long preempt_disable_ip;
1528#endif
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001529#ifdef CONFIG_NUMA
Miao Xie58568d22009-06-16 15:31:49 -07001530 struct mempolicy *mempolicy; /* Protected by alloc_lock */
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001531 short il_next;
Eric Dumazet207205a2011-03-22 16:30:44 -07001532 short pref_node_fork;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001533#endif
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001534#ifdef CONFIG_NUMA_BALANCING
1535 int numa_scan_seq;
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001536 unsigned int numa_scan_period;
Mel Gorman598f0ec2013-10-07 11:28:55 +01001537 unsigned int numa_scan_period_max;
Rik van Rielde1c9ce2013-10-07 11:29:39 +01001538 int numa_preferred_nid;
Mel Gorman6b9a7462013-10-07 11:29:11 +01001539 unsigned long numa_migrate_retry;
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001540 u64 node_stamp; /* migration stamp */
Rik van Riel7e2703e2014-01-27 17:03:45 -05001541 u64 last_task_numa_placement;
1542 u64 last_sum_exec_runtime;
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001543 struct callback_head numa_work;
Mel Gormanf809ca92013-10-07 11:28:57 +01001544
Peter Zijlstra8c8a7432013-10-07 11:29:21 +01001545 struct list_head numa_entry;
1546 struct numa_group *numa_group;
1547
Mel Gorman745d6142013-10-07 11:28:59 +01001548 /*
1549 * Exponential decaying average of faults on a per-node basis.
1550 * Scheduling placement decisions are made based on the these counts.
1551 * The values remain static for the duration of a PTE scan
1552 */
Rik van Rielff1df892014-01-27 17:03:41 -05001553 unsigned long *numa_faults_memory;
Mel Gorman83e1d2c2013-10-07 11:29:27 +01001554 unsigned long total_numa_faults;
Mel Gorman745d6142013-10-07 11:28:59 +01001555
1556 /*
1557 * numa_faults_buffer records faults per node during the current
Rik van Rielff1df892014-01-27 17:03:41 -05001558 * scan window. When the scan completes, the counts in
1559 * numa_faults_memory decay and these values are copied.
Mel Gorman745d6142013-10-07 11:28:59 +01001560 */
Rik van Rielff1df892014-01-27 17:03:41 -05001561 unsigned long *numa_faults_buffer_memory;
Mel Gorman745d6142013-10-07 11:28:59 +01001562
Rik van Riel04bb2f92013-10-07 11:29:36 +01001563 /*
Rik van Riel50ec8a42014-01-27 17:03:42 -05001564 * Track the nodes the process was running on when a NUMA hinting
1565 * fault was incurred.
1566 */
1567 unsigned long *numa_faults_cpu;
1568 unsigned long *numa_faults_buffer_cpu;
1569
1570 /*
Rik van Riel04bb2f92013-10-07 11:29:36 +01001571 * numa_faults_locality tracks if faults recorded during the last
1572 * scan window were remote/local. The task scan period is adapted
1573 * based on the locality of the faults with different weights
1574 * depending on whether they were shared or private faults
1575 */
1576 unsigned long numa_faults_locality[2];
1577
Ingo Molnarb32e86b2013-10-07 11:29:30 +01001578 unsigned long numa_pages_migrated;
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001579#endif /* CONFIG_NUMA_BALANCING */
1580
Ingo Molnare56d0902006-01-08 01:01:37 -08001581 struct rcu_head rcu;
Jens Axboeb92ce552006-04-11 13:52:07 +02001582
1583 /*
1584 * cache last used pipe for splice
1585 */
1586 struct pipe_inode_info *splice_pipe;
Eric Dumazet5640f762012-09-23 23:04:42 +00001587
1588 struct page_frag task_frag;
1589
Shailabh Nagarca74e922006-07-14 00:24:36 -07001590#ifdef CONFIG_TASK_DELAY_ACCT
1591 struct task_delay_info *delays;
1592#endif
Akinobu Mitaf4f154f2006-12-08 02:39:47 -08001593#ifdef CONFIG_FAULT_INJECTION
1594 int make_it_fail;
1595#endif
Wu Fengguang9d823e82011-06-11 18:10:12 -06001596 /*
1597 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1598 * balance_dirty_pages() for some dirty throttling pause
1599 */
1600 int nr_dirtied;
1601 int nr_dirtied_pause;
Wu Fengguang83712352011-06-11 19:25:42 -06001602 unsigned long dirty_paused_when; /* start of a write-and-pause period */
Wu Fengguang9d823e82011-06-11 18:10:12 -06001603
Arjan van de Ven97455122008-01-25 21:08:34 +01001604#ifdef CONFIG_LATENCYTOP
1605 int latency_record_count;
1606 struct latency_record latency_record[LT_SAVECOUNT];
1607#endif
Arjan van de Ven69766752008-09-01 15:52:40 -07001608 /*
1609 * time slack values; these are used to round up poll() and
1610 * select() etc timeout values. These are in nanoseconds.
1611 */
1612 unsigned long timer_slack_ns;
1613 unsigned long default_timer_slack_ns;
David Millerf8d570a2008-11-06 00:37:40 -08001614
Frederic Weisbeckerfb526072008-11-25 21:07:04 +01001615#ifdef CONFIG_FUNCTION_GRAPH_TRACER
Daniel Mack3ad2f3f2010-02-03 08:01:28 +08001616 /* Index of current stored address in ret_stack */
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001617 int curr_ret_stack;
1618 /* Stack of return addresses for return function tracing */
1619 struct ftrace_ret_stack *ret_stack;
Steven Rostedt8aef2d22009-03-24 01:10:15 -04001620 /* time stamp for last schedule */
1621 unsigned long long ftrace_timestamp;
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001622 /*
1623 * Number of functions that haven't been traced
1624 * because of depth overrun.
1625 */
1626 atomic_t trace_overrun;
Frederic Weisbecker380c4b12008-12-06 03:43:41 +01001627 /* Pause for the tracing */
1628 atomic_t tracing_graph_pause;
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001629#endif
Steven Rostedtea4e2bc2008-12-03 15:36:57 -05001630#ifdef CONFIG_TRACING
1631 /* state flags for use by tracers */
1632 unsigned long trace;
Steven Rostedtb1cff0a2011-05-25 14:27:43 -04001633 /* bitmask and counter of trace recursion */
Steven Rostedt261842b2009-04-16 21:41:52 -04001634 unsigned long trace_recursion;
1635#endif /* CONFIG_TRACING */
Andrew Mortonc255a452012-07-31 16:43:02 -07001636#ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
Glauber Costa0e9d92f2012-12-18 14:22:42 -08001637 unsigned int memcg_kmem_skip_account;
Johannes Weiner519e5242013-09-12 15:13:42 -07001638 struct memcg_oom_info {
Johannes Weiner49426422013-10-16 13:46:59 -07001639 struct mem_cgroup *memcg;
1640 gfp_t gfp_mask;
1641 int order;
Johannes Weiner519e5242013-09-12 15:13:42 -07001642 unsigned int may_oom:1;
1643 } memcg_oom;
KAMEZAWA Hiroyuki569b8462009-12-15 16:47:03 -08001644#endif
Srikar Dronamraju0326f5a2012-03-13 23:30:11 +05301645#ifdef CONFIG_UPROBES
1646 struct uprobe_task *utask;
Srikar Dronamraju0326f5a2012-03-13 23:30:11 +05301647#endif
Kent Overstreetcafe5632013-03-23 16:11:31 -07001648#if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1649 unsigned int sequential_io;
1650 unsigned int sequential_io_avg;
1651#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001652};
1653
Rusty Russell76e6eee2009-03-12 14:35:43 -06001654/* Future-safe accessor for struct task_struct's cpus_allowed. */
Rusty Russella4636812009-12-17 11:43:29 -06001655#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
Rusty Russell76e6eee2009-03-12 14:35:43 -06001656
Peter Zijlstra6688cc02013-10-07 11:29:24 +01001657#define TNF_MIGRATED 0x01
1658#define TNF_NO_GROUP 0x02
Rik van Rieldabe1d92013-10-07 11:29:34 +01001659#define TNF_SHARED 0x04
Rik van Riel04bb2f92013-10-07 11:29:36 +01001660#define TNF_FAULT_LOCAL 0x08
Peter Zijlstra6688cc02013-10-07 11:29:24 +01001661
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001662#ifdef CONFIG_NUMA_BALANCING
Peter Zijlstra6688cc02013-10-07 11:29:24 +01001663extern void task_numa_fault(int last_node, int node, int pages, int flags);
Mel Gormane29cf082013-10-07 11:29:22 +01001664extern pid_t task_numa_group_id(struct task_struct *p);
Mel Gorman1a687c22012-11-22 11:16:36 +00001665extern void set_numabalancing_state(bool enabled);
Rik van Riel82727012013-10-07 11:29:28 +01001666extern void task_numa_free(struct task_struct *p);
Rik van Riel10f39042014-01-27 17:03:44 -05001667extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1668 int src_nid, int dst_cpu);
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001669#else
Mel Gormanac8e8952013-10-07 11:29:03 +01001670static inline void task_numa_fault(int last_node, int node, int pages,
Peter Zijlstra6688cc02013-10-07 11:29:24 +01001671 int flags)
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001672{
1673}
Mel Gormane29cf082013-10-07 11:29:22 +01001674static inline pid_t task_numa_group_id(struct task_struct *p)
1675{
1676 return 0;
1677}
Mel Gorman1a687c22012-11-22 11:16:36 +00001678static inline void set_numabalancing_state(bool enabled)
1679{
1680}
Rik van Riel82727012013-10-07 11:29:28 +01001681static inline void task_numa_free(struct task_struct *p)
1682{
1683}
Rik van Riel10f39042014-01-27 17:03:44 -05001684static inline bool should_numa_migrate_memory(struct task_struct *p,
1685 struct page *page, int src_nid, int dst_cpu)
1686{
1687 return true;
1688}
Peter Zijlstracbee9f82012-10-25 14:16:43 +02001689#endif
1690
Alexey Dobriyane8681712007-10-26 12:17:22 +04001691static inline struct pid *task_pid(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001692{
1693 return task->pids[PIDTYPE_PID].pid;
1694}
1695
Alexey Dobriyane8681712007-10-26 12:17:22 +04001696static inline struct pid *task_tgid(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001697{
1698 return task->group_leader->pids[PIDTYPE_PID].pid;
1699}
1700
Oleg Nesterov6dda81f2009-04-02 16:58:35 -07001701/*
1702 * Without tasklist or rcu lock it is not safe to dereference
1703 * the result of task_pgrp/task_session even if task == current,
1704 * we can race with another thread doing sys_setsid/sys_setpgid.
1705 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04001706static inline struct pid *task_pgrp(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001707{
1708 return task->group_leader->pids[PIDTYPE_PGID].pid;
1709}
1710
Alexey Dobriyane8681712007-10-26 12:17:22 +04001711static inline struct pid *task_session(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001712{
1713 return task->group_leader->pids[PIDTYPE_SID].pid;
1714}
1715
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001716struct pid_namespace;
1717
1718/*
1719 * the helpers to get the task's different pids as they are seen
1720 * from various namespaces
1721 *
1722 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
Eric W. Biederman44c4e1b2008-02-08 04:19:15 -08001723 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1724 * current.
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001725 * task_xid_nr_ns() : id seen from the ns specified;
1726 *
1727 * set_task_vxid() : assigns a virtual id to a task;
1728 *
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001729 * see also pid_nr() etc in include/linux/pid.h
1730 */
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001731pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1732 struct pid_namespace *ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001733
Alexey Dobriyane8681712007-10-26 12:17:22 +04001734static inline pid_t task_pid_nr(struct task_struct *tsk)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001735{
1736 return tsk->pid;
1737}
1738
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001739static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1740 struct pid_namespace *ns)
1741{
1742 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1743}
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001744
1745static inline pid_t task_pid_vnr(struct task_struct *tsk)
1746{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001747 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001748}
1749
1750
Alexey Dobriyane8681712007-10-26 12:17:22 +04001751static inline pid_t task_tgid_nr(struct task_struct *tsk)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001752{
1753 return tsk->tgid;
1754}
1755
Pavel Emelyanov2f2a3a42007-10-18 23:40:19 -07001756pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001757
1758static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1759{
1760 return pid_vnr(task_tgid(tsk));
1761}
1762
1763
Richard Guy Briggs80e0b6e2014-03-16 14:00:19 -04001764static inline int pid_alive(const struct task_struct *p);
Richard Guy Briggsad36d282013-08-15 18:05:12 -04001765static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1766{
1767 pid_t pid = 0;
1768
1769 rcu_read_lock();
1770 if (pid_alive(tsk))
1771 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1772 rcu_read_unlock();
1773
1774 return pid;
1775}
1776
1777static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1778{
1779 return task_ppid_nr_ns(tsk, &init_pid_ns);
1780}
1781
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001782static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1783 struct pid_namespace *ns)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001784{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001785 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001786}
1787
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001788static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1789{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001790 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001791}
1792
1793
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001794static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1795 struct pid_namespace *ns)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001796{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001797 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001798}
1799
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001800static inline pid_t task_session_vnr(struct task_struct *tsk)
1801{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001802 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001803}
1804
Oleg Nesterov1b0f7ff2009-04-02 16:58:39 -07001805/* obsolete, do not use */
1806static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1807{
1808 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1809}
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001810
Linus Torvalds1da177e2005-04-16 15:20:36 -07001811/**
1812 * pid_alive - check that a task structure is not stale
1813 * @p: Task structure to be checked.
1814 *
1815 * Test if a process is not yet dead (at most zombie state)
1816 * If pid_alive fails, then pointers within the task structure
1817 * can be stale and must not be dereferenced.
Yacine Belkadie69f6182013-07-12 20:45:47 +02001818 *
1819 * Return: 1 if the process is alive. 0 otherwise.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001820 */
Richard Guy Briggsad36d282013-08-15 18:05:12 -04001821static inline int pid_alive(const struct task_struct *p)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001822{
Eric W. Biederman92476d72006-03-31 02:31:42 -08001823 return p->pids[PIDTYPE_PID].pid != NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001824}
1825
Sukadev Bhattiproluf400e192006-09-29 02:00:07 -07001826/**
Serge E. Hallynb460cbc2007-10-18 23:39:52 -07001827 * is_global_init - check if a task structure is init
Henne32602592006-10-06 00:44:01 -07001828 * @tsk: Task structure to be checked.
1829 *
1830 * Check if a task structure is the first user space task the kernel created.
Yacine Belkadie69f6182013-07-12 20:45:47 +02001831 *
1832 * Return: 1 if the task structure is init. 0 otherwise.
Sukadev Bhattiproluf400e192006-09-29 02:00:07 -07001833 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04001834static inline int is_global_init(struct task_struct *tsk)
Pavel Emelyanovb461cc02007-10-18 23:40:09 -07001835{
1836 return tsk->pid == 1;
1837}
Serge E. Hallynb460cbc2007-10-18 23:39:52 -07001838
Cedric Le Goater9ec52092006-10-02 02:19:00 -07001839extern struct pid *cad_pid;
1840
Linus Torvalds1da177e2005-04-16 15:20:36 -07001841extern void free_task(struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001842#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
Ingo Molnare56d0902006-01-08 01:01:37 -08001843
Andrew Morton158d9eb2006-03-31 02:31:34 -08001844extern void __put_task_struct(struct task_struct *t);
Ingo Molnare56d0902006-01-08 01:01:37 -08001845
1846static inline void put_task_struct(struct task_struct *t)
1847{
1848 if (atomic_dec_and_test(&t->usage))
Eric W. Biederman8c7904a2006-03-31 02:31:37 -08001849 __put_task_struct(t);
Ingo Molnare56d0902006-01-08 01:01:37 -08001850}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001851
Frederic Weisbecker6a616712012-12-16 20:00:34 +01001852#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1853extern void task_cputime(struct task_struct *t,
1854 cputime_t *utime, cputime_t *stime);
1855extern void task_cputime_scaled(struct task_struct *t,
1856 cputime_t *utimescaled, cputime_t *stimescaled);
1857extern cputime_t task_gtime(struct task_struct *t);
1858#else
Frederic Weisbecker6fac4822012-11-13 14:20:55 +01001859static inline void task_cputime(struct task_struct *t,
1860 cputime_t *utime, cputime_t *stime)
1861{
1862 if (utime)
1863 *utime = t->utime;
1864 if (stime)
1865 *stime = t->stime;
1866}
1867
1868static inline void task_cputime_scaled(struct task_struct *t,
1869 cputime_t *utimescaled,
1870 cputime_t *stimescaled)
1871{
1872 if (utimescaled)
1873 *utimescaled = t->utimescaled;
1874 if (stimescaled)
1875 *stimescaled = t->stimescaled;
1876}
Frederic Weisbecker6a616712012-12-16 20:00:34 +01001877
1878static inline cputime_t task_gtime(struct task_struct *t)
1879{
1880 return t->gtime;
1881}
1882#endif
Frederic Weisbeckere80d0a12012-11-21 16:26:44 +01001883extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1884extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
Balbir Singh49048622008-09-05 18:12:23 +02001885
Linus Torvalds1da177e2005-04-16 15:20:36 -07001886/*
1887 * Per process flags
1888 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001889#define PF_EXITING 0x00000004 /* getting shut down */
Alexey Kuznetsov778e9a92007-06-08 13:47:00 -07001890#define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
Laurent Vivier94886b82007-10-15 17:00:19 +02001891#define PF_VCPU 0x00000010 /* I'm a virtual CPU */
Tejun Heo21aa9af2010-06-08 21:40:37 +02001892#define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001893#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
Andi Kleen4db96cf2009-09-16 11:50:14 +02001894#define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001895#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1896#define PF_DUMPCORE 0x00000200 /* dumped core */
1897#define PF_SIGNALED 0x00000400 /* killed by a signal */
1898#define PF_MEMALLOC 0x00000800 /* Allocating memory */
Vasiliy Kulikov72fa5992011-08-08 19:02:04 +04001899#define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001900#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
Tejun Heo774a1222013-01-15 18:52:51 -08001901#define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001902#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1903#define PF_FROZEN 0x00010000 /* frozen for system suspend */
1904#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1905#define PF_KSWAPD 0x00040000 /* I am kswapd */
Ming Lei21caf2f2013-02-22 16:34:08 -08001906#define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001907#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
Oleg Nesterov246bb0b2008-07-25 01:47:38 -07001908#define PF_KTHREAD 0x00200000 /* I am a kernel thread */
Jens Axboeb31dc662006-06-13 08:26:10 +02001909#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1910#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1911#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1912#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
Tejun Heo14a40ff2013-03-19 13:45:20 -07001913#define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
Andi Kleen4db96cf2009-09-16 11:50:14 +02001914#define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
Thomas Gleixner61a87122006-06-27 02:54:56 -07001915#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
Tejun Heo58a69cb2011-02-16 09:25:31 +01001916#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
Colin Cross2b44c4d2013-07-24 17:41:33 -07001917#define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001918
1919/*
1920 * Only the _current_ task can read/write to tsk->flags, but other
1921 * tasks can access tsk->flags in readonly mode for example
1922 * with tsk_used_math (like during threaded core dumping).
1923 * There is however an exception to this rule during ptrace
1924 * or during fork: the ptracer task is allowed to write to the
1925 * child->flags of its traced child (same goes for fork, the parent
1926 * can write to the child->flags), because we're guaranteed the
1927 * child is not running and in turn not changing child->flags
1928 * at the same time the parent does it.
1929 */
1930#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1931#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1932#define clear_used_math() clear_stopped_child_used_math(current)
1933#define set_used_math() set_stopped_child_used_math(current)
1934#define conditional_stopped_child_used_math(condition, child) \
1935 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1936#define conditional_used_math(condition) \
1937 conditional_stopped_child_used_math(condition, current)
1938#define copy_to_stopped_child_used_math(child) \
1939 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1940/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1941#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1942#define used_math() tsk_used_math(current)
1943
Ming Lei21caf2f2013-02-22 16:34:08 -08001944/* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1945static inline gfp_t memalloc_noio_flags(gfp_t flags)
1946{
1947 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1948 flags &= ~__GFP_IO;
1949 return flags;
1950}
1951
1952static inline unsigned int memalloc_noio_save(void)
1953{
1954 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1955 current->flags |= PF_MEMALLOC_NOIO;
1956 return flags;
1957}
1958
1959static inline void memalloc_noio_restore(unsigned int flags)
1960{
1961 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1962}
1963
Kees Cook1d4457f2014-05-21 15:23:46 -07001964/* Per-process atomic flags. */
1965#define PFA_NO_NEW_PRIVS 0x00000001 /* May not gain new privileges. */
1966
1967static inline bool task_no_new_privs(struct task_struct *p)
1968{
1969 return test_bit(PFA_NO_NEW_PRIVS, &p->atomic_flags);
1970}
1971
1972static inline void task_set_no_new_privs(struct task_struct *p)
1973{
1974 set_bit(PFA_NO_NEW_PRIVS, &p->atomic_flags);
1975}
1976
Tejun Heoe5c19022011-03-23 10:37:00 +01001977/*
Tejun Heoa8f072c2011-06-02 11:13:59 +02001978 * task->jobctl flags
Tejun Heoe5c19022011-03-23 10:37:00 +01001979 */
Tejun Heoa8f072c2011-06-02 11:13:59 +02001980#define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
Tejun Heoe5c19022011-03-23 10:37:00 +01001981
Tejun Heoa8f072c2011-06-02 11:13:59 +02001982#define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1983#define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1984#define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
Tejun Heo73ddff22011-06-14 11:20:14 +02001985#define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
Tejun Heofb1d9102011-06-14 11:20:17 +02001986#define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
Tejun Heoa8f072c2011-06-02 11:13:59 +02001987#define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
Tejun Heo544b2c92011-06-14 11:20:18 +02001988#define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
Tejun Heoa8f072c2011-06-02 11:13:59 +02001989
1990#define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1991#define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1992#define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
Tejun Heo73ddff22011-06-14 11:20:14 +02001993#define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
Tejun Heofb1d9102011-06-14 11:20:17 +02001994#define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
Tejun Heoa8f072c2011-06-02 11:13:59 +02001995#define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
Tejun Heo544b2c92011-06-14 11:20:18 +02001996#define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
Tejun Heoa8f072c2011-06-02 11:13:59 +02001997
Tejun Heofb1d9102011-06-14 11:20:17 +02001998#define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
Tejun Heo73ddff22011-06-14 11:20:14 +02001999#define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
Tejun Heo3759a0d2011-06-02 11:14:00 +02002000
Tejun Heo7dd3db52011-06-02 11:14:00 +02002001extern bool task_set_jobctl_pending(struct task_struct *task,
2002 unsigned int mask);
Tejun Heo73ddff22011-06-14 11:20:14 +02002003extern void task_clear_jobctl_trapping(struct task_struct *task);
Tejun Heo3759a0d2011-06-02 11:14:00 +02002004extern void task_clear_jobctl_pending(struct task_struct *task,
2005 unsigned int mask);
Tejun Heo39efa3e2011-03-23 10:37:00 +01002006
Paul E. McKenneya57eb942010-06-29 16:49:16 -07002007#ifdef CONFIG_PREEMPT_RCU
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07002008#define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
Paul E. McKenney1aa03f12012-01-11 17:25:17 -08002009#define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
Paul E. McKenney8315f422014-06-27 13:42:20 -07002010#endif /* #ifdef CONFIG_PREEMPT_RCU */
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07002011
2012static inline void rcu_copy_process(struct task_struct *p)
2013{
Paul E. McKenney8315f422014-06-27 13:42:20 -07002014#ifdef CONFIG_PREEMPT_RCU
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07002015 p->rcu_read_lock_nesting = 0;
2016 p->rcu_read_unlock_special = 0;
Paul E. McKenneydd5d19b2009-08-27 14:58:16 -07002017 p->rcu_blocked_node = NULL;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07002018 INIT_LIST_HEAD(&p->rcu_node_entry);
Paul E. McKenney8315f422014-06-27 13:42:20 -07002019#endif /* #ifdef CONFIG_PREEMPT_RCU */
2020#ifdef CONFIG_TASKS_RCU
2021 p->rcu_tasks_holdout = false;
2022 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2023#endif /* #ifdef CONFIG_TASKS_RCU */
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07002024}
2025
Mel Gorman907aed42012-07-31 16:44:07 -07002026static inline void tsk_restore_flags(struct task_struct *task,
2027 unsigned long orig_flags, unsigned long flags)
2028{
2029 task->flags &= ~flags;
2030 task->flags |= orig_flags & flags;
2031}
2032
Linus Torvalds1da177e2005-04-16 15:20:36 -07002033#ifdef CONFIG_SMP
KOSAKI Motohiro1e1b6c52011-05-19 15:08:58 +09002034extern void do_set_cpus_allowed(struct task_struct *p,
2035 const struct cpumask *new_mask);
2036
Mike Traviscd8ba7c2008-03-26 14:23:49 -07002037extern int set_cpus_allowed_ptr(struct task_struct *p,
Rusty Russell96f874e2008-11-25 02:35:14 +10302038 const struct cpumask *new_mask);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002039#else
KOSAKI Motohiro1e1b6c52011-05-19 15:08:58 +09002040static inline void do_set_cpus_allowed(struct task_struct *p,
2041 const struct cpumask *new_mask)
2042{
2043}
Mike Traviscd8ba7c2008-03-26 14:23:49 -07002044static inline int set_cpus_allowed_ptr(struct task_struct *p,
Rusty Russell96f874e2008-11-25 02:35:14 +10302045 const struct cpumask *new_mask)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002046{
Rusty Russell96f874e2008-11-25 02:35:14 +10302047 if (!cpumask_test_cpu(0, new_mask))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002048 return -EINVAL;
2049 return 0;
2050}
2051#endif
Rusty Russelle0ad9552009-09-24 09:34:38 -06002052
Frederic Weisbecker3451d022011-08-10 23:21:01 +02002053#ifdef CONFIG_NO_HZ_COMMON
Peter Zijlstra5167e8d2012-06-22 15:52:09 +02002054void calc_load_enter_idle(void);
2055void calc_load_exit_idle(void);
2056#else
2057static inline void calc_load_enter_idle(void) { }
2058static inline void calc_load_exit_idle(void) { }
Frederic Weisbecker3451d022011-08-10 23:21:01 +02002059#endif /* CONFIG_NO_HZ_COMMON */
Peter Zijlstra5167e8d2012-06-22 15:52:09 +02002060
Rusty Russelle0ad9552009-09-24 09:34:38 -06002061#ifndef CONFIG_CPUMASK_OFFSTACK
Mike Traviscd8ba7c2008-03-26 14:23:49 -07002062static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2063{
2064 return set_cpus_allowed_ptr(p, &new_mask);
2065}
Rusty Russelle0ad9552009-09-24 09:34:38 -06002066#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002067
Ingo Molnarb3425012009-02-26 20:20:29 +01002068/*
Peter Zijlstrac6763292010-05-25 10:48:51 +02002069 * Do not use outside of architecture code which knows its limitations.
2070 *
2071 * sched_clock() has no promise of monotonicity or bounded drift between
2072 * CPUs, use (which you should not) requires disabling IRQs.
2073 *
2074 * Please use one of the three interfaces below.
Ingo Molnarb3425012009-02-26 20:20:29 +01002075 */
Mike Frysinger1bbfa6f2009-12-09 20:07:03 -05002076extern unsigned long long notrace sched_clock(void);
Peter Zijlstrac6763292010-05-25 10:48:51 +02002077/*
Hiroshi Shimamoto489a71b2012-04-02 17:00:44 +09002078 * See the comment in kernel/sched/clock.c
Peter Zijlstrac6763292010-05-25 10:48:51 +02002079 */
2080extern u64 cpu_clock(int cpu);
2081extern u64 local_clock(void);
2082extern u64 sched_clock_cpu(int cpu);
2083
Ingo Molnare436d802007-07-19 21:28:35 +02002084
Peter Zijlstrac1955a32008-08-11 08:59:03 +02002085extern void sched_clock_init(void);
Peter Zijlstrac1955a32008-08-11 08:59:03 +02002086
Peter Zijlstra3e51f332008-05-03 18:29:28 +02002087#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
Peter Zijlstra3e51f332008-05-03 18:29:28 +02002088static inline void sched_clock_tick(void)
2089{
2090}
2091
2092static inline void sched_clock_idle_sleep_event(void)
2093{
2094}
2095
2096static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2097{
2098}
2099#else
Peter Zijlstrac6763292010-05-25 10:48:51 +02002100/*
2101 * Architectures can set this to 1 if they have specified
2102 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2103 * but then during bootup it turns out that sched_clock()
2104 * is reliable after all:
2105 */
Peter Zijlstra35af99e2013-11-28 19:38:42 +01002106extern int sched_clock_stable(void);
2107extern void set_sched_clock_stable(void);
2108extern void clear_sched_clock_stable(void);
Peter Zijlstrac6763292010-05-25 10:48:51 +02002109
Peter Zijlstra3e51f332008-05-03 18:29:28 +02002110extern void sched_clock_tick(void);
2111extern void sched_clock_idle_sleep_event(void);
2112extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2113#endif
2114
Venkatesh Pallipadib52bfee2010-10-04 17:03:19 -07002115#ifdef CONFIG_IRQ_TIME_ACCOUNTING
2116/*
2117 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2118 * The reason for this explicit opt-in is not to have perf penalty with
2119 * slow sched_clocks.
2120 */
2121extern void enable_sched_clock_irqtime(void);
2122extern void disable_sched_clock_irqtime(void);
2123#else
2124static inline void enable_sched_clock_irqtime(void) {}
2125static inline void disable_sched_clock_irqtime(void) {}
2126#endif
2127
Ingo Molnar36c8b582006-07-03 00:25:41 -07002128extern unsigned long long
Ingo Molnar41b86e92007-07-09 18:51:58 +02002129task_sched_runtime(struct task_struct *task);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002130
2131/* sched_exec is called by processes performing an exec */
2132#ifdef CONFIG_SMP
2133extern void sched_exec(void);
2134#else
2135#define sched_exec() {}
2136#endif
2137
Ingo Molnar2aa44d02007-08-23 15:18:02 +02002138extern void sched_clock_idle_sleep_event(void);
2139extern void sched_clock_idle_wakeup_event(u64 delta_ns);
Ingo Molnarbb29ab22007-07-09 18:51:59 +02002140
Linus Torvalds1da177e2005-04-16 15:20:36 -07002141#ifdef CONFIG_HOTPLUG_CPU
2142extern void idle_task_exit(void);
2143#else
2144static inline void idle_task_exit(void) {}
2145#endif
2146
Frederic Weisbecker3451d022011-08-10 23:21:01 +02002147#if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
Frederic Weisbecker1c200912011-08-10 23:21:01 +02002148extern void wake_up_nohz_cpu(int cpu);
Thomas Gleixner06d83082008-03-22 09:20:24 +01002149#else
Frederic Weisbecker1c200912011-08-10 23:21:01 +02002150static inline void wake_up_nohz_cpu(int cpu) { }
Thomas Gleixner06d83082008-03-22 09:20:24 +01002151#endif
2152
Frederic Weisbeckerce831b32013-04-20 15:15:35 +02002153#ifdef CONFIG_NO_HZ_FULL
2154extern bool sched_can_stop_tick(void);
Frederic Weisbecker265f22a2013-05-03 03:39:05 +02002155extern u64 scheduler_tick_max_deferment(void);
Frederic Weisbeckerce831b32013-04-20 15:15:35 +02002156#else
2157static inline bool sched_can_stop_tick(void) { return false; }
Ingo Molnarbf0f6f22007-07-09 18:51:58 +02002158#endif
Ingo Molnarbf0f6f22007-07-09 18:51:58 +02002159
Mike Galbraith5091faa2010-11-30 14:18:03 +01002160#ifdef CONFIG_SCHED_AUTOGROUP
Mike Galbraith5091faa2010-11-30 14:18:03 +01002161extern void sched_autogroup_create_attach(struct task_struct *p);
2162extern void sched_autogroup_detach(struct task_struct *p);
2163extern void sched_autogroup_fork(struct signal_struct *sig);
2164extern void sched_autogroup_exit(struct signal_struct *sig);
2165#ifdef CONFIG_PROC_FS
2166extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
Hiroshi Shimamoto2e5b5b32012-02-23 17:41:27 +09002167extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
Mike Galbraith5091faa2010-11-30 14:18:03 +01002168#endif
2169#else
2170static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2171static inline void sched_autogroup_detach(struct task_struct *p) { }
2172static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2173static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2174#endif
2175
Dan Carpenterfa933842014-05-23 13:20:42 +03002176extern int yield_to(struct task_struct *p, bool preempt);
Ingo Molnar36c8b582006-07-03 00:25:41 -07002177extern void set_user_nice(struct task_struct *p, long nice);
2178extern int task_prio(const struct task_struct *p);
Dongsheng Yangd0ea0262014-01-27 22:00:45 -05002179/**
2180 * task_nice - return the nice value of a given task.
2181 * @p: the task in question.
2182 *
2183 * Return: The nice value [ -20 ... 0 ... 19 ].
2184 */
2185static inline int task_nice(const struct task_struct *p)
2186{
2187 return PRIO_TO_NICE((p)->static_prio);
2188}
Ingo Molnar36c8b582006-07-03 00:25:41 -07002189extern int can_nice(const struct task_struct *p, const int nice);
2190extern int task_curr(const struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002191extern int idle_cpu(int cpu);
KOSAKI Motohirofe7de492010-10-20 16:01:12 -07002192extern int sched_setscheduler(struct task_struct *, int,
2193 const struct sched_param *);
Rusty Russell961ccdd2008-06-23 13:55:38 +10002194extern int sched_setscheduler_nocheck(struct task_struct *, int,
KOSAKI Motohirofe7de492010-10-20 16:01:12 -07002195 const struct sched_param *);
Dario Faggiolid50dde52013-11-07 14:43:36 +01002196extern int sched_setattr(struct task_struct *,
2197 const struct sched_attr *);
Ingo Molnar36c8b582006-07-03 00:25:41 -07002198extern struct task_struct *idle_task(int cpu);
Paul E. McKenneyc4f30602011-11-10 12:41:56 -08002199/**
2200 * is_idle_task - is the specified task an idle task?
Randy Dunlapfa757282012-01-21 11:03:13 -08002201 * @p: the task in question.
Yacine Belkadie69f6182013-07-12 20:45:47 +02002202 *
2203 * Return: 1 if @p is an idle task. 0 otherwise.
Paul E. McKenneyc4f30602011-11-10 12:41:56 -08002204 */
Paul E. McKenney7061ca32011-12-20 08:20:46 -08002205static inline bool is_idle_task(const struct task_struct *p)
Paul E. McKenneyc4f30602011-11-10 12:41:56 -08002206{
2207 return p->pid == 0;
2208}
Ingo Molnar36c8b582006-07-03 00:25:41 -07002209extern struct task_struct *curr_task(int cpu);
2210extern void set_curr_task(int cpu, struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002211
2212void yield(void);
2213
2214/*
2215 * The default (Linux) execution domain.
2216 */
2217extern struct exec_domain default_exec_domain;
2218
2219union thread_union {
2220 struct thread_info thread_info;
2221 unsigned long stack[THREAD_SIZE/sizeof(long)];
2222};
2223
2224#ifndef __HAVE_ARCH_KSTACK_END
2225static inline int kstack_end(void *addr)
2226{
2227 /* Reliable end of stack detection:
2228 * Some APM bios versions misalign the stack
2229 */
2230 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2231}
2232#endif
2233
2234extern union thread_union init_thread_union;
2235extern struct task_struct init_task;
2236
2237extern struct mm_struct init_mm;
2238
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002239extern struct pid_namespace init_pid_ns;
2240
2241/*
2242 * find a task by one of its numerical ids
2243 *
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002244 * find_task_by_pid_ns():
2245 * finds a task by its pid in the specified namespace
Pavel Emelyanov228ebcb2007-10-18 23:40:16 -07002246 * find_task_by_vpid():
2247 * finds a task by its virtual pid
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002248 *
Pavel Emelyanove49859e2008-07-25 01:48:36 -07002249 * see also find_vpid() etc in include/linux/pid.h
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002250 */
2251
Pavel Emelyanov228ebcb2007-10-18 23:40:16 -07002252extern struct task_struct *find_task_by_vpid(pid_t nr);
2253extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2254 struct pid_namespace *ns);
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002255
Linus Torvalds1da177e2005-04-16 15:20:36 -07002256/* per-UID process charging. */
Eric W. Biederman7b44ab92011-11-16 23:20:58 -08002257extern struct user_struct * alloc_uid(kuid_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002258static inline struct user_struct *get_uid(struct user_struct *u)
2259{
2260 atomic_inc(&u->__count);
2261 return u;
2262}
2263extern void free_uid(struct user_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002264
2265#include <asm/current.h>
2266
Torben Hohnf0af911a92011-01-27 15:59:10 +01002267extern void xtime_update(unsigned long ticks);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002268
Harvey Harrisonb3c97522008-02-13 15:03:15 -08002269extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2270extern int wake_up_process(struct task_struct *tsk);
Samir Bellabes3e51e3e2011-05-11 18:18:05 +02002271extern void wake_up_new_task(struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002272#ifdef CONFIG_SMP
2273 extern void kick_process(struct task_struct *tsk);
2274#else
2275 static inline void kick_process(struct task_struct *tsk) { }
2276#endif
Dario Faggioliaab03e02013-11-28 11:14:43 +01002277extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
Ingo Molnarad46c2c2007-07-09 18:52:00 +02002278extern void sched_dead(struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002279
Linus Torvalds1da177e2005-04-16 15:20:36 -07002280extern void proc_caches_init(void);
2281extern void flush_signals(struct task_struct *);
David Howells3bcac022009-04-29 13:45:05 +01002282extern void __flush_signals(struct task_struct *);
Oleg Nesterov10ab8252007-05-09 02:34:37 -07002283extern void ignore_signals(struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002284extern void flush_signal_handlers(struct task_struct *, int force_default);
2285extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2286
2287static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2288{
2289 unsigned long flags;
2290 int ret;
2291
2292 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2293 ret = dequeue_signal(tsk, mask, info);
2294 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2295
2296 return ret;
Oleg Nesterov53c8f9f2011-06-22 23:08:18 +02002297}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002298
2299extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2300 sigset_t *mask);
2301extern void unblock_all_signals(void);
2302extern void release_task(struct task_struct * p);
2303extern int send_sig_info(int, struct siginfo *, struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002304extern int force_sigsegv(int, struct task_struct *);
2305extern int force_sig_info(int, struct siginfo *, struct task_struct *);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002306extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002307extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
Serge Hallynd178bc32011-09-26 10:45:18 -05002308extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2309 const struct cred *, u32);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002310extern int kill_pgrp(struct pid *pid, int sig, int priv);
2311extern int kill_pid(struct pid *pid, int sig, int priv);
Matthew Wilcoxc3de4b32007-02-09 08:11:47 -07002312extern int kill_proc_info(int, struct siginfo *, pid_t);
Oleg Nesterov86773472011-06-22 23:09:09 +02002313extern __must_check bool do_notify_parent(struct task_struct *, int);
Oleg Nesterova7f07652009-09-23 15:56:44 -07002314extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002315extern void force_sig(int, struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002316extern int send_sig(int, struct task_struct *, int);
Oleg Nesterov09faef12010-05-26 14:43:11 -07002317extern int zap_other_threads(struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002318extern struct sigqueue *sigqueue_alloc(void);
2319extern void sigqueue_free(struct sigqueue *);
Oleg Nesterovac5c2152008-04-30 00:52:57 -07002320extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
Oleg Nesterov9ac95f22006-02-09 22:41:50 +03002321extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002322
Al Viro51a7b442012-05-21 23:33:55 -04002323static inline void restore_saved_sigmask(void)
2324{
2325 if (test_and_clear_restore_sigmask())
Al Viro77097ae2012-04-27 13:58:59 -04002326 __set_current_blocked(&current->saved_sigmask);
Al Viro51a7b442012-05-21 23:33:55 -04002327}
2328
Al Virob7f9a112012-05-02 09:59:21 -04002329static inline sigset_t *sigmask_to_save(void)
2330{
2331 sigset_t *res = &current->blocked;
2332 if (unlikely(test_restore_sigmask()))
2333 res = &current->saved_sigmask;
2334 return res;
2335}
2336
Cedric Le Goater9ec52092006-10-02 02:19:00 -07002337static inline int kill_cad_pid(int sig, int priv)
2338{
2339 return kill_pid(cad_pid, sig, priv);
2340}
2341
Linus Torvalds1da177e2005-04-16 15:20:36 -07002342/* These can be the second arg to send_sig_info/send_group_sig_info. */
2343#define SEND_SIG_NOINFO ((struct siginfo *) 0)
2344#define SEND_SIG_PRIV ((struct siginfo *) 1)
2345#define SEND_SIG_FORCED ((struct siginfo *) 2)
2346
Sebastian Andrzej Siewior2a855dd2009-10-25 15:37:58 +01002347/*
2348 * True if we are on the alternate signal stack.
2349 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002350static inline int on_sig_stack(unsigned long sp)
2351{
Sebastian Andrzej Siewior2a855dd2009-10-25 15:37:58 +01002352#ifdef CONFIG_STACK_GROWSUP
2353 return sp >= current->sas_ss_sp &&
2354 sp - current->sas_ss_sp < current->sas_ss_size;
2355#else
2356 return sp > current->sas_ss_sp &&
2357 sp - current->sas_ss_sp <= current->sas_ss_size;
2358#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002359}
2360
2361static inline int sas_ss_flags(unsigned long sp)
2362{
Richard Weinberger72f15c02014-03-05 15:15:22 +01002363 if (!current->sas_ss_size)
2364 return SS_DISABLE;
2365
2366 return on_sig_stack(sp) ? SS_ONSTACK : 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002367}
2368
Al Viro5a1b98d2012-11-06 13:28:21 -05002369static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2370{
2371 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2372#ifdef CONFIG_STACK_GROWSUP
2373 return current->sas_ss_sp;
2374#else
2375 return current->sas_ss_sp + current->sas_ss_size;
2376#endif
2377 return sp;
2378}
2379
Linus Torvalds1da177e2005-04-16 15:20:36 -07002380/*
2381 * Routines for handling mm_structs
2382 */
2383extern struct mm_struct * mm_alloc(void);
2384
2385/* mmdrop drops the mm and the page tables */
Harvey Harrisonb3c97522008-02-13 15:03:15 -08002386extern void __mmdrop(struct mm_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002387static inline void mmdrop(struct mm_struct * mm)
2388{
Ingo Molnar6fb43d72007-07-09 18:52:01 +02002389 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002390 __mmdrop(mm);
2391}
2392
2393/* mmput gets rid of the mappings and all user-space */
2394extern void mmput(struct mm_struct *);
2395/* Grab a reference to a task's mm, if it is not already going away */
2396extern struct mm_struct *get_task_mm(struct task_struct *task);
Christopher Yeoh8cdb8782012-02-02 11:34:09 +10302397/*
2398 * Grab a reference to a task's mm, if it is not already going away
2399 * and ptrace_may_access with the mode parameter passed to it
2400 * succeeds.
2401 */
2402extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002403/* Remove the current tasks stale references to the old mm_struct */
2404extern void mm_release(struct task_struct *, struct mm_struct *);
2405
Alexey Dobriyan6f2c55b2009-04-02 16:56:59 -07002406extern int copy_thread(unsigned long, unsigned long, unsigned long,
Al Viroafa86fc2012-10-22 22:51:14 -04002407 struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002408extern void flush_thread(void);
2409extern void exit_thread(void);
2410
Linus Torvalds1da177e2005-04-16 15:20:36 -07002411extern void exit_files(struct task_struct *);
Oleg Nesterova7e53282006-03-28 16:11:27 -08002412extern void __cleanup_sighand(struct sighand_struct *);
Oleg Nesterovcbaffba2008-05-26 20:55:42 +04002413
Linus Torvalds1da177e2005-04-16 15:20:36 -07002414extern void exit_itimers(struct signal_struct *);
Oleg Nesterovcbaffba2008-05-26 20:55:42 +04002415extern void flush_itimer_signals(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002416
Joe Perches9402c952012-01-12 17:17:17 -08002417extern void do_group_exit(int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002418
Linus Torvaldsc4ad8f92014-02-05 12:54:53 -08002419extern int do_execve(struct filename *,
David Howellsd7627462010-08-17 23:52:56 +01002420 const char __user * const __user *,
Al Viroda3d4c52012-10-20 21:49:33 -04002421 const char __user * const __user *);
Al Viroe80d6662012-10-22 23:10:08 -04002422extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
Ingo Molnar36c8b582006-07-03 00:25:41 -07002423struct task_struct *fork_idle(int);
Al Viro2aa3a7f2012-09-21 19:55:31 -04002424extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002425
Adrian Hunter82b89772014-05-28 11:45:04 +03002426extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2427static inline void set_task_comm(struct task_struct *tsk, const char *from)
2428{
2429 __set_task_comm(tsk, from, false);
2430}
Andrew Morton59714d62008-02-04 22:27:21 -08002431extern char *get_task_comm(char *to, struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002432
2433#ifdef CONFIG_SMP
Peter Zijlstra317f3942011-04-05 17:23:58 +02002434void scheduler_ipi(void);
Roland McGrath85ba2d82008-07-25 19:45:58 -07002435extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002436#else
Peter Zijlstra184748c2011-04-05 17:23:39 +02002437static inline void scheduler_ipi(void) { }
Roland McGrath85ba2d82008-07-25 19:45:58 -07002438static inline unsigned long wait_task_inactive(struct task_struct *p,
2439 long match_state)
2440{
2441 return 1;
2442}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002443#endif
2444
Jiri Pirko05725f72009-04-14 20:17:16 +02002445#define next_task(p) \
2446 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002447
2448#define for_each_process(p) \
2449 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2450
Oleg Nesterov5bb459b2009-07-10 03:48:23 +02002451extern bool current_is_single_threaded(void);
David Howellsd84f4f92008-11-14 10:39:23 +11002452
Linus Torvalds1da177e2005-04-16 15:20:36 -07002453/*
2454 * Careful: do_each_thread/while_each_thread is a double loop so
2455 * 'break' will not work as expected - use goto instead.
2456 */
2457#define do_each_thread(g, t) \
2458 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2459
2460#define while_each_thread(g, t) \
2461 while ((t = next_thread(t)) != g)
2462
Oleg Nesterov0c740d02014-01-21 15:49:56 -08002463#define __for_each_thread(signal, t) \
2464 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2465
2466#define for_each_thread(p, t) \
2467 __for_each_thread((p)->signal, t)
2468
2469/* Careful: this is a double loop, 'break' won't work as expected. */
2470#define for_each_process_thread(p, t) \
2471 for_each_process(p) for_each_thread(p, t)
2472
Oleg Nesterov7e498272010-05-26 14:43:22 -07002473static inline int get_nr_threads(struct task_struct *tsk)
2474{
Oleg Nesterovb3ac0222010-05-26 14:43:24 -07002475 return tsk->signal->nr_threads;
Oleg Nesterov7e498272010-05-26 14:43:22 -07002476}
2477
Oleg Nesterov087806b2011-06-22 23:10:26 +02002478static inline bool thread_group_leader(struct task_struct *p)
2479{
2480 return p->exit_signal >= 0;
2481}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002482
Eric W. Biederman0804ef42006-10-02 02:17:04 -07002483/* Do to the insanities of de_thread it is possible for a process
2484 * to have the pid of the thread group leader without actually being
2485 * the thread group leader. For iteration through the pids in proc
2486 * all we care about is that we have a task with the appropriate
2487 * pid, we don't actually care if we have the right task.
2488 */
Oleg Nesterove1403b82013-09-11 14:20:06 -07002489static inline bool has_group_leader_pid(struct task_struct *p)
Eric W. Biederman0804ef42006-10-02 02:17:04 -07002490{
Oleg Nesterove1403b82013-09-11 14:20:06 -07002491 return task_pid(p) == p->signal->leader_pid;
Eric W. Biederman0804ef42006-10-02 02:17:04 -07002492}
2493
Pavel Emelyanovbac0abd2007-10-18 23:40:18 -07002494static inline
Oleg Nesterove1403b82013-09-11 14:20:06 -07002495bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
Pavel Emelyanovbac0abd2007-10-18 23:40:18 -07002496{
Oleg Nesterove1403b82013-09-11 14:20:06 -07002497 return p1->signal == p2->signal;
Pavel Emelyanovbac0abd2007-10-18 23:40:18 -07002498}
2499
Ingo Molnar36c8b582006-07-03 00:25:41 -07002500static inline struct task_struct *next_thread(const struct task_struct *p)
Oleg Nesterov47e65322006-03-28 16:11:25 -08002501{
Jiri Pirko05725f72009-04-14 20:17:16 +02002502 return list_entry_rcu(p->thread_group.next,
2503 struct task_struct, thread_group);
Oleg Nesterov47e65322006-03-28 16:11:25 -08002504}
2505
Alexey Dobriyane8681712007-10-26 12:17:22 +04002506static inline int thread_group_empty(struct task_struct *p)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002507{
Oleg Nesterov47e65322006-03-28 16:11:25 -08002508 return list_empty(&p->thread_group);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002509}
2510
2511#define delay_group_leader(p) \
2512 (thread_group_leader(p) && !thread_group_empty(p))
2513
Linus Torvalds1da177e2005-04-16 15:20:36 -07002514/*
Eric W. Biederman260ea102006-06-23 02:05:18 -07002515 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
Jens Axboe22e2c502005-06-27 10:55:12 +02002516 * subscriptions and synchronises with wait4(). Also used in procfs. Also
Paul Menageddbcc7e2007-10-18 23:39:30 -07002517 * pins the final release of task.io_context. Also protects ->cpuset and
Oleg Nesterovd68b46f2012-03-05 14:59:13 -08002518 * ->cgroup.subsys[]. And ->vfork_done.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002519 *
2520 * Nests both inside and outside of read_lock(&tasklist_lock).
2521 * It must not be nested with write_lock_irq(&tasklist_lock),
2522 * neither inside nor outside.
2523 */
2524static inline void task_lock(struct task_struct *p)
2525{
2526 spin_lock(&p->alloc_lock);
2527}
2528
2529static inline void task_unlock(struct task_struct *p)
2530{
2531 spin_unlock(&p->alloc_lock);
2532}
2533
Namhyung Kimb8ed3742010-10-27 15:34:06 -07002534extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
Oleg Nesterovf63ee722006-03-28 16:11:13 -08002535 unsigned long *flags);
2536
Anton Vorontsov9388dc32012-02-09 20:45:19 +04002537static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2538 unsigned long *flags)
2539{
2540 struct sighand_struct *ret;
2541
2542 ret = __lock_task_sighand(tsk, flags);
2543 (void)__cond_lock(&tsk->sighand->siglock, ret);
2544 return ret;
2545}
Namhyung Kimb8ed3742010-10-27 15:34:06 -07002546
Oleg Nesterovf63ee722006-03-28 16:11:13 -08002547static inline void unlock_task_sighand(struct task_struct *tsk,
2548 unsigned long *flags)
2549{
2550 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2551}
2552
Ben Blum4714d1d2011-05-26 16:25:18 -07002553#ifdef CONFIG_CGROUPS
Tejun Heo257058a2011-12-12 18:12:21 -08002554static inline void threadgroup_change_begin(struct task_struct *tsk)
Ben Blum4714d1d2011-05-26 16:25:18 -07002555{
Tejun Heo257058a2011-12-12 18:12:21 -08002556 down_read(&tsk->signal->group_rwsem);
Ben Blum4714d1d2011-05-26 16:25:18 -07002557}
Tejun Heo257058a2011-12-12 18:12:21 -08002558static inline void threadgroup_change_end(struct task_struct *tsk)
Ben Blum4714d1d2011-05-26 16:25:18 -07002559{
Tejun Heo257058a2011-12-12 18:12:21 -08002560 up_read(&tsk->signal->group_rwsem);
Ben Blum4714d1d2011-05-26 16:25:18 -07002561}
Tejun Heo77e4ef92011-12-12 18:12:21 -08002562
2563/**
2564 * threadgroup_lock - lock threadgroup
2565 * @tsk: member task of the threadgroup to lock
2566 *
2567 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2568 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
Oleg Nesterove56fb282013-04-30 15:28:20 -07002569 * change ->group_leader/pid. This is useful for cases where the threadgroup
2570 * needs to stay stable across blockable operations.
Tejun Heo77e4ef92011-12-12 18:12:21 -08002571 *
2572 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2573 * synchronization. While held, no new task will be added to threadgroup
2574 * and no existing live task will have its PF_EXITING set.
2575 *
Oleg Nesterove56fb282013-04-30 15:28:20 -07002576 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2577 * sub-thread becomes a new leader.
Tejun Heo77e4ef92011-12-12 18:12:21 -08002578 */
Tejun Heo257058a2011-12-12 18:12:21 -08002579static inline void threadgroup_lock(struct task_struct *tsk)
Ben Blum4714d1d2011-05-26 16:25:18 -07002580{
Tejun Heo257058a2011-12-12 18:12:21 -08002581 down_write(&tsk->signal->group_rwsem);
Ben Blum4714d1d2011-05-26 16:25:18 -07002582}
Tejun Heo77e4ef92011-12-12 18:12:21 -08002583
2584/**
2585 * threadgroup_unlock - unlock threadgroup
2586 * @tsk: member task of the threadgroup to unlock
2587 *
2588 * Reverse threadgroup_lock().
2589 */
Tejun Heo257058a2011-12-12 18:12:21 -08002590static inline void threadgroup_unlock(struct task_struct *tsk)
Ben Blum4714d1d2011-05-26 16:25:18 -07002591{
Tejun Heo257058a2011-12-12 18:12:21 -08002592 up_write(&tsk->signal->group_rwsem);
Ben Blum4714d1d2011-05-26 16:25:18 -07002593}
2594#else
Tejun Heo257058a2011-12-12 18:12:21 -08002595static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2596static inline void threadgroup_change_end(struct task_struct *tsk) {}
2597static inline void threadgroup_lock(struct task_struct *tsk) {}
2598static inline void threadgroup_unlock(struct task_struct *tsk) {}
Ben Blum4714d1d2011-05-26 16:25:18 -07002599#endif
2600
Al Virof0373602005-11-13 16:06:57 -08002601#ifndef __HAVE_THREAD_FUNCTIONS
2602
Roman Zippelf7e42172007-05-09 02:35:17 -07002603#define task_thread_info(task) ((struct thread_info *)(task)->stack)
2604#define task_stack_page(task) ((task)->stack)
Al Viroa1261f52005-11-13 16:06:55 -08002605
Al Viro10ebffd2005-11-13 16:06:56 -08002606static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2607{
2608 *task_thread_info(p) = *task_thread_info(org);
2609 task_thread_info(p)->task = p;
2610}
2611
2612static inline unsigned long *end_of_stack(struct task_struct *p)
2613{
Roman Zippelf7e42172007-05-09 02:35:17 -07002614 return (unsigned long *)(task_thread_info(p) + 1);
Al Viro10ebffd2005-11-13 16:06:56 -08002615}
2616
Al Virof0373602005-11-13 16:06:57 -08002617#endif
2618
FUJITA Tomonori8b05c7e2008-07-23 21:26:53 -07002619static inline int object_is_on_stack(void *obj)
2620{
2621 void *stack = task_stack_page(current);
2622
2623 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2624}
2625
Benjamin Herrenschmidt8c9843e2008-04-18 16:56:15 +10002626extern void thread_info_cache_init(void);
2627
Eric Sandeen7c9f8862008-04-22 16:38:23 -05002628#ifdef CONFIG_DEBUG_STACK_USAGE
2629static inline unsigned long stack_not_used(struct task_struct *p)
2630{
2631 unsigned long *n = end_of_stack(p);
2632
2633 do { /* Skip over canary */
2634 n++;
2635 } while (!*n);
2636
2637 return (unsigned long)n - (unsigned long)end_of_stack(p);
2638}
2639#endif
2640
Linus Torvalds1da177e2005-04-16 15:20:36 -07002641/* set thread flags in other task's structures
2642 * - see asm/thread_info.h for TIF_xxxx flags available
2643 */
2644static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2645{
Al Viroa1261f52005-11-13 16:06:55 -08002646 set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002647}
2648
2649static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2650{
Al Viroa1261f52005-11-13 16:06:55 -08002651 clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002652}
2653
2654static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2655{
Al Viroa1261f52005-11-13 16:06:55 -08002656 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002657}
2658
2659static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2660{
Al Viroa1261f52005-11-13 16:06:55 -08002661 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002662}
2663
2664static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2665{
Al Viroa1261f52005-11-13 16:06:55 -08002666 return test_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002667}
2668
2669static inline void set_tsk_need_resched(struct task_struct *tsk)
2670{
2671 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2672}
2673
2674static inline void clear_tsk_need_resched(struct task_struct *tsk)
2675{
2676 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2677}
2678
Gregory Haskins8ae121a2008-04-23 07:13:29 -04002679static inline int test_tsk_need_resched(struct task_struct *tsk)
2680{
2681 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2682}
2683
Eric W. Biederman690cc3f2009-05-13 16:55:10 +00002684static inline int restart_syscall(void)
2685{
2686 set_tsk_thread_flag(current, TIF_SIGPENDING);
2687 return -ERESTARTNOINTR;
2688}
2689
Linus Torvalds1da177e2005-04-16 15:20:36 -07002690static inline int signal_pending(struct task_struct *p)
2691{
2692 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2693}
Matthew Wilcoxf776d122007-12-06 11:15:50 -05002694
Roland McGrathd9588722009-09-23 15:57:04 -07002695static inline int __fatal_signal_pending(struct task_struct *p)
2696{
2697 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2698}
Matthew Wilcoxf776d122007-12-06 11:15:50 -05002699
2700static inline int fatal_signal_pending(struct task_struct *p)
2701{
2702 return signal_pending(p) && __fatal_signal_pending(p);
2703}
2704
Oleg Nesterov16882c12008-06-08 21:20:41 +04002705static inline int signal_pending_state(long state, struct task_struct *p)
2706{
2707 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2708 return 0;
2709 if (!signal_pending(p))
2710 return 0;
2711
Oleg Nesterov16882c12008-06-08 21:20:41 +04002712 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2713}
2714
Linus Torvalds1da177e2005-04-16 15:20:36 -07002715/*
2716 * cond_resched() and cond_resched_lock(): latency reduction via
2717 * explicit rescheduling in places that are safe. The return
2718 * value indicates whether a reschedule was done in fact.
2719 * cond_resched_lock() will drop the spinlock before scheduling,
2720 * cond_resched_softirq() will enable bhs before scheduling.
2721 */
Linus Torvaldsc3921ab2008-05-11 16:04:48 -07002722extern int _cond_resched(void);
Frederic Weisbecker6f80bd92009-07-16 15:44:29 +02002723
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002724#define cond_resched() ({ \
2725 __might_sleep(__FILE__, __LINE__, 0); \
2726 _cond_resched(); \
2727})
Frederic Weisbecker6f80bd92009-07-16 15:44:29 +02002728
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002729extern int __cond_resched_lock(spinlock_t *lock);
2730
Frederic Weisbeckerbdd4e852011-06-08 01:13:27 +02002731#ifdef CONFIG_PREEMPT_COUNT
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002732#define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
Herbert Xu02b67cc32008-01-25 21:08:28 +01002733#else
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002734#define PREEMPT_LOCK_OFFSET 0
Herbert Xu02b67cc32008-01-25 21:08:28 +01002735#endif
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002736
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002737#define cond_resched_lock(lock) ({ \
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002738 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002739 __cond_resched_lock(lock); \
2740})
2741
2742extern int __cond_resched_softirq(void);
2743
Venkatesh Pallipadi75e10562010-10-04 17:03:16 -07002744#define cond_resched_softirq() ({ \
2745 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2746 __cond_resched_softirq(); \
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002747})
Linus Torvalds1da177e2005-04-16 15:20:36 -07002748
Simon Hormanf6f3c432013-05-22 14:50:31 +09002749static inline void cond_resched_rcu(void)
2750{
2751#if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2752 rcu_read_unlock();
2753 cond_resched();
2754 rcu_read_lock();
2755#endif
2756}
2757
Linus Torvalds1da177e2005-04-16 15:20:36 -07002758/*
2759 * Does a critical section need to be broken due to another
Nick Piggin95c354f2008-01-30 13:31:20 +01002760 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2761 * but a general need for low latency)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002762 */
Nick Piggin95c354f2008-01-30 13:31:20 +01002763static inline int spin_needbreak(spinlock_t *lock)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002764{
Nick Piggin95c354f2008-01-30 13:31:20 +01002765#ifdef CONFIG_PREEMPT
2766 return spin_is_contended(lock);
2767#else
Linus Torvalds1da177e2005-04-16 15:20:36 -07002768 return 0;
Nick Piggin95c354f2008-01-30 13:31:20 +01002769#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002770}
2771
Roland McGrath7bb44ad2007-05-23 13:57:44 -07002772/*
Thomas Gleixneree761f62013-03-21 22:49:32 +01002773 * Idle thread specific functions to determine the need_resched
Peter Zijlstra69dd0f82014-04-09 14:30:10 +02002774 * polling state.
Thomas Gleixneree761f62013-03-21 22:49:32 +01002775 */
Peter Zijlstra69dd0f82014-04-09 14:30:10 +02002776#ifdef TIF_POLLING_NRFLAG
Thomas Gleixneree761f62013-03-21 22:49:32 +01002777static inline int tsk_is_polling(struct task_struct *p)
2778{
2779 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2780}
Peter Zijlstraea811742013-09-11 12:43:13 +02002781
2782static inline void __current_set_polling(void)
Thomas Gleixner3a98f872013-03-21 22:49:33 +01002783{
2784 set_thread_flag(TIF_POLLING_NRFLAG);
2785}
2786
Peter Zijlstraea811742013-09-11 12:43:13 +02002787static inline bool __must_check current_set_polling_and_test(void)
2788{
2789 __current_set_polling();
2790
2791 /*
2792 * Polling state must be visible before we test NEED_RESCHED,
Kirill Tkhai88751252014-06-29 00:03:57 +04002793 * paired by resched_curr()
Peter Zijlstraea811742013-09-11 12:43:13 +02002794 */
Peter Zijlstra4e857c52014-03-17 18:06:10 +01002795 smp_mb__after_atomic();
Peter Zijlstraea811742013-09-11 12:43:13 +02002796
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,
Kirill Tkhai88751252014-06-29 00:03:57 +04002811 * paired by resched_curr()
Peter Zijlstraea811742013-09-11 12:43:13 +02002812 */
Peter Zijlstra4e857c52014-03-17 18:06:10 +01002813 smp_mb__after_atomic();
Peter Zijlstraea811742013-09-11 12:43:13 +02002814
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 */
Kirill Tkhai88751252014-06-29 00:03:57 +04002843 smp_mb(); /* paired with resched_curr() */
Peter Zijlstra8cb75e02013-11-20 12:22:37 +01002844
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
Oleg Nesterovf98bafa2014-06-04 16:07:34 -07002966#ifdef CONFIG_MEMCG
Balbir Singhcf475ad2008-04-29 01:00:16 -07002967extern void mm_update_next_owner(struct mm_struct *mm);
Balbir Singhcf475ad2008-04-29 01:00:16 -07002968#else
2969static inline void mm_update_next_owner(struct mm_struct *mm)
2970{
2971}
Oleg Nesterovf98bafa2014-06-04 16:07:34 -07002972#endif /* CONFIG_MEMCG */
Balbir Singhcf475ad2008-04-29 01:00:16 -07002973
Jiri Slaby3e10e712009-11-19 17:16:37 +01002974static inline unsigned long task_rlimit(const struct task_struct *tsk,
2975 unsigned int limit)
2976{
2977 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2978}
2979
2980static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2981 unsigned int limit)
2982{
2983 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2984}
2985
2986static inline unsigned long rlimit(unsigned int limit)
2987{
2988 return task_rlimit(current, limit);
2989}
2990
2991static inline unsigned long rlimit_max(unsigned int limit)
2992{
2993 return task_rlimit_max(current, limit);
2994}
2995
Linus Torvalds1da177e2005-04-16 15:20:36 -07002996#endif