wait: fix reparent_leader() vs EXIT_DEAD->EXIT_ZOMBIE race
[deliverable/linux.git] / include / linux / sched.h
CommitLineData
1da177e4
LT
1#ifndef _LINUX_SCHED_H
2#define _LINUX_SCHED_H
3
607ca46e 4#include <uapi/linux/sched.h>
b7b3c76a 5
5c228079
DY
6#include <linux/sched/prio.h>
7
b7b3c76a
DW
8
9struct sched_param {
10 int sched_priority;
11};
12
1da177e4
LT
13#include <asm/param.h> /* for HZ */
14
1da177e4
LT
15#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>
fb00aca4 21#include <linux/plist.h>
1da177e4
LT
22#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>
c92ff1bd 27#include <linux/mm_types.h>
00d1a39e 28#include <linux/preempt_mask.h>
1da177e4 29
1da177e4
LT
30#include <asm/page.h>
31#include <asm/ptrace.h>
bfc3f028 32#include <linux/cputime.h>
1da177e4
LT
33
34#include <linux/smp.h>
35#include <linux/sem.h>
36#include <linux/signal.h>
1da177e4
LT
37#include <linux/compiler.h>
38#include <linux/completion.h>
39#include <linux/pid.h>
40#include <linux/percpu.h>
41#include <linux/topology.h>
3e26c149 42#include <linux/proportions.h>
1da177e4 43#include <linux/seccomp.h>
e56d0903 44#include <linux/rcupdate.h>
05725f7e 45#include <linux/rculist.h>
23f78d4a 46#include <linux/rtmutex.h>
1da177e4 47
a3b6714e
DW
48#include <linux/time.h>
49#include <linux/param.h>
50#include <linux/resource.h>
51#include <linux/timer.h>
52#include <linux/hrtimer.h>
7c3ab738 53#include <linux/task_io_accounting.h>
9745512c 54#include <linux/latencytop.h>
9e2b2dc4 55#include <linux/cred.h>
fa14ff4a 56#include <linux/llist.h>
7b44ab97 57#include <linux/uidgid.h>
21caf2fc 58#include <linux/gfp.h>
a3b6714e
DW
59
60#include <asm/processor.h>
36d57ac4 61
d50dde5a
DF
62#define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
63
64/*
65 * Extended scheduling parameters data structure.
66 *
67 * This is needed because the original struct sched_param can not be
68 * altered without introducing ABI issues with legacy applications
69 * (e.g., in sched_getparam()).
70 *
71 * However, the possibility of specifying more than just a priority for
72 * the tasks may be useful for a wide variety of application fields, e.g.,
73 * multimedia, streaming, automation and control, and many others.
74 *
75 * This variant (sched_attr) is meant at describing a so-called
76 * sporadic time-constrained task. In such model a task is specified by:
77 * - the activation period or minimum instance inter-arrival time;
78 * - the maximum (or average, depending on the actual scheduling
79 * discipline) computation time of all instances, a.k.a. runtime;
80 * - the deadline (relative to the actual activation time) of each
81 * instance.
82 * Very briefly, a periodic (sporadic) task asks for the execution of
83 * some specific computation --which is typically called an instance--
84 * (at most) every period. Moreover, each instance typically lasts no more
85 * than the runtime and must be completed by time instant t equal to
86 * the instance activation time + the deadline.
87 *
88 * This is reflected by the actual fields of the sched_attr structure:
89 *
90 * @size size of the structure, for fwd/bwd compat.
91 *
92 * @sched_policy task's scheduling policy
93 * @sched_flags for customizing the scheduler behaviour
94 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
95 * @sched_priority task's static priority (SCHED_FIFO/RR)
96 * @sched_deadline representative of the task's deadline
97 * @sched_runtime representative of the task's runtime
98 * @sched_period representative of the task's period
99 *
100 * Given this task model, there are a multiplicity of scheduling algorithms
101 * and policies, that can be used to ensure all the tasks will make their
102 * timing constraints.
aab03e05
DF
103 *
104 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
105 * only user of this new interface. More information about the algorithm
106 * available in the scheduling class file or in Documentation/.
d50dde5a
DF
107 */
108struct sched_attr {
109 u32 size;
110
111 u32 sched_policy;
112 u64 sched_flags;
113
114 /* SCHED_NORMAL, SCHED_BATCH */
115 s32 sched_nice;
116
117 /* SCHED_FIFO, SCHED_RR */
118 u32 sched_priority;
119
120 /* SCHED_DEADLINE */
121 u64 sched_runtime;
122 u64 sched_deadline;
123 u64 sched_period;
124};
125
1da177e4 126struct exec_domain;
c87e2837 127struct futex_pi_state;
286100a6 128struct robust_list_head;
bddd87c7 129struct bio_list;
5ad4e53b 130struct fs_struct;
cdd6c482 131struct perf_event_context;
73c10101 132struct blk_plug;
c4ad8f98 133struct filename;
1da177e4 134
615d6e87
DB
135#define VMACACHE_BITS 2
136#define VMACACHE_SIZE (1U << VMACACHE_BITS)
137#define VMACACHE_MASK (VMACACHE_SIZE - 1)
138
1da177e4
LT
139/*
140 * List of flags we want to share for kernel threads,
141 * if only because they are not used by them anyway.
142 */
143#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
144
145/*
146 * These are the constant used to fake the fixed-point load-average
147 * counting. Some notes:
148 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
149 * a load-average precision of 10 bits integer + 11 bits fractional
150 * - if you want to count load-averages more often, you need more
151 * precision, or rounding will get you. With 2-second counting freq,
152 * the EXP_n values would be 1981, 2034 and 2043 if still using only
153 * 11 bit fractions.
154 */
155extern unsigned long avenrun[]; /* Load averages */
2d02494f 156extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
1da177e4
LT
157
158#define FSHIFT 11 /* nr of bits of precision */
159#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
0c2043ab 160#define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
1da177e4
LT
161#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
162#define EXP_5 2014 /* 1/exp(5sec/5min) */
163#define EXP_15 2037 /* 1/exp(5sec/15min) */
164
165#define CALC_LOAD(load,exp,n) \
166 load *= exp; \
167 load += n*(FIXED_1-exp); \
168 load >>= FSHIFT;
169
170extern unsigned long total_forks;
171extern int nr_threads;
1da177e4
LT
172DECLARE_PER_CPU(unsigned long, process_counts);
173extern int nr_processes(void);
174extern unsigned long nr_running(void);
1da177e4 175extern unsigned long nr_iowait(void);
8c215bd3 176extern unsigned long nr_iowait_cpu(int cpu);
69d25870
AV
177extern unsigned long this_cpu_load(void);
178
179
0f004f5a 180extern void calc_global_load(unsigned long ticks);
5aaa0b7a 181extern void update_cpu_load_nohz(void);
1da177e4 182
7e49fcce
SR
183extern unsigned long get_parent_ip(unsigned long addr);
184
b637a328
PM
185extern void dump_cpu_task(int cpu);
186
43ae34cb
IM
187struct seq_file;
188struct cfs_rq;
4cf86d77 189struct task_group;
43ae34cb
IM
190#ifdef CONFIG_SCHED_DEBUG
191extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
192extern void proc_sched_set_task(struct task_struct *p);
193extern void
5cef9eca 194print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
43ae34cb 195#endif
1da177e4 196
4a8342d2
LT
197/*
198 * Task state bitmask. NOTE! These bits are also
199 * encoded in fs/proc/array.c: get_task_state().
200 *
201 * We have two separate sets of flags: task->state
202 * is about runnability, while task->exit_state are
203 * about the task exiting. Confusing, but this way
204 * modifying one set can't modify the other one by
205 * mistake.
206 */
1da177e4
LT
207#define TASK_RUNNING 0
208#define TASK_INTERRUPTIBLE 1
209#define TASK_UNINTERRUPTIBLE 2
f021a3c2
MW
210#define __TASK_STOPPED 4
211#define __TASK_TRACED 8
4a8342d2
LT
212/* in tsk->exit_state */
213#define EXIT_ZOMBIE 16
214#define EXIT_DEAD 32
215/* in tsk->state again */
af927232 216#define TASK_DEAD 64
f021a3c2 217#define TASK_WAKEKILL 128
e9c84311 218#define TASK_WAKING 256
f2530dc7
TG
219#define TASK_PARKED 512
220#define TASK_STATE_MAX 1024
f021a3c2 221
f2530dc7 222#define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
73342151 223
e1781538
PZ
224extern char ___assert_task_state[1 - 2*!!(
225 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
f021a3c2
MW
226
227/* Convenience macros for the sake of set_task_state */
228#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
229#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
230#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
1da177e4 231
92a1f4bc
MW
232/* Convenience macros for the sake of wake_up */
233#define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
f021a3c2 234#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
92a1f4bc
MW
235
236/* get_task_state() */
237#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
f021a3c2 238 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
74e37200 239 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
92a1f4bc 240
f021a3c2
MW
241#define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
242#define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
92a1f4bc 243#define task_is_stopped_or_traced(task) \
f021a3c2 244 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
92a1f4bc 245#define task_contributes_to_load(task) \
e3c8ca83 246 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
376fede8 247 (task->flags & PF_FROZEN) == 0)
1da177e4
LT
248
249#define __set_task_state(tsk, state_value) \
250 do { (tsk)->state = (state_value); } while (0)
251#define set_task_state(tsk, state_value) \
252 set_mb((tsk)->state, (state_value))
253
498d0c57
AM
254/*
255 * set_current_state() includes a barrier so that the write of current->state
256 * is correctly serialised wrt the caller's subsequent test of whether to
257 * actually sleep:
258 *
259 * set_current_state(TASK_UNINTERRUPTIBLE);
260 * if (do_i_need_to_sleep())
261 * schedule();
262 *
263 * If the caller does not need such serialisation then use __set_current_state()
264 */
1da177e4
LT
265#define __set_current_state(state_value) \
266 do { current->state = (state_value); } while (0)
267#define set_current_state(state_value) \
268 set_mb(current->state, (state_value))
269
270/* Task command name length */
271#define TASK_COMM_LEN 16
272
1da177e4
LT
273#include <linux/spinlock.h>
274
275/*
276 * This serializes "schedule()" and also protects
277 * the run-queue from deletions/modifications (but
278 * _adding_ to the beginning of the run-queue has
279 * a separate lock).
280 */
281extern rwlock_t tasklist_lock;
282extern spinlock_t mmlist_lock;
283
36c8b586 284struct task_struct;
1da177e4 285
db1466b3
PM
286#ifdef CONFIG_PROVE_RCU
287extern int lockdep_tasklist_lock_is_held(void);
288#endif /* #ifdef CONFIG_PROVE_RCU */
289
1da177e4
LT
290extern void sched_init(void);
291extern void sched_init_smp(void);
2d07b255 292extern asmlinkage void schedule_tail(struct task_struct *prev);
36c8b586 293extern void init_idle(struct task_struct *idle, int cpu);
1df21055 294extern void init_idle_bootup_task(struct task_struct *idle);
1da177e4 295
89f19f04 296extern int runqueue_is_locked(int cpu);
017730c1 297
3451d024 298#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
c1cc017c 299extern void nohz_balance_enter_idle(int cpu);
69e1e811 300extern void set_cpu_sd_state_idle(void);
6201b4d6 301extern int get_nohz_timer_target(int pinned);
46cb4b7c 302#else
c1cc017c 303static inline void nohz_balance_enter_idle(int cpu) { }
fdaabd80 304static inline void set_cpu_sd_state_idle(void) { }
6201b4d6
VK
305static inline int get_nohz_timer_target(int pinned)
306{
307 return smp_processor_id();
308}
46cb4b7c 309#endif
1da177e4 310
e59e2ae2 311/*
39bc89fd 312 * Only dump TASK_* tasks. (0 for all tasks)
e59e2ae2
IM
313 */
314extern void show_state_filter(unsigned long state_filter);
315
316static inline void show_state(void)
317{
39bc89fd 318 show_state_filter(0);
e59e2ae2
IM
319}
320
1da177e4
LT
321extern void show_regs(struct pt_regs *);
322
323/*
324 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
325 * task), SP is the stack pointer of the first frame that should be shown in the back
326 * trace (or NULL if the entire call-chain of the task should be shown).
327 */
328extern void show_stack(struct task_struct *task, unsigned long *sp);
329
330void io_schedule(void);
331long io_schedule_timeout(long timeout);
332
333extern void cpu_init (void);
334extern void trap_init(void);
335extern void update_process_times(int user);
336extern void scheduler_tick(void);
337
82a1fcb9
IM
338extern void sched_show_task(struct task_struct *p);
339
19cc36c0 340#ifdef CONFIG_LOCKUP_DETECTOR
8446f1d3 341extern void touch_softlockup_watchdog(void);
d6ad3e28 342extern void touch_softlockup_watchdog_sync(void);
04c9167f 343extern void touch_all_softlockup_watchdogs(void);
332fbdbc
DZ
344extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
345 void __user *buffer,
346 size_t *lenp, loff_t *ppos);
9c44bc03 347extern unsigned int softlockup_panic;
004417a6 348void lockup_detector_init(void);
8446f1d3 349#else
8446f1d3
IM
350static inline void touch_softlockup_watchdog(void)
351{
352}
d6ad3e28
JW
353static inline void touch_softlockup_watchdog_sync(void)
354{
355}
04c9167f
JF
356static inline void touch_all_softlockup_watchdogs(void)
357{
358}
004417a6
PZ
359static inline void lockup_detector_init(void)
360{
361}
8446f1d3
IM
362#endif
363
8b414521
MT
364#ifdef CONFIG_DETECT_HUNG_TASK
365void reset_hung_task_detector(void);
366#else
367static inline void reset_hung_task_detector(void)
368{
369}
370#endif
371
1da177e4
LT
372/* Attach to any functions which should be ignored in wchan output. */
373#define __sched __attribute__((__section__(".sched.text")))
deaf2227
IM
374
375/* Linker adds these: start and end of __sched functions */
376extern char __sched_text_start[], __sched_text_end[];
377
1da177e4
LT
378/* Is this address in the __sched functions? */
379extern int in_sched_functions(unsigned long addr);
380
381#define MAX_SCHEDULE_TIMEOUT LONG_MAX
b3c97528 382extern signed long schedule_timeout(signed long timeout);
64ed93a2 383extern signed long schedule_timeout_interruptible(signed long timeout);
294d5cc2 384extern signed long schedule_timeout_killable(signed long timeout);
64ed93a2 385extern signed long schedule_timeout_uninterruptible(signed long timeout);
1da177e4 386asmlinkage void schedule(void);
c5491ea7 387extern void schedule_preempt_disabled(void);
1da177e4 388
ab516013 389struct nsproxy;
acce292c 390struct user_namespace;
1da177e4 391
efc1a3b1
DH
392#ifdef CONFIG_MMU
393extern void arch_pick_mmap_layout(struct mm_struct *mm);
1da177e4
LT
394extern unsigned long
395arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
396 unsigned long, unsigned long);
397extern unsigned long
398arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
399 unsigned long len, unsigned long pgoff,
400 unsigned long flags);
efc1a3b1
DH
401#else
402static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
403#endif
1da177e4 404
d049f74f
KC
405#define SUID_DUMP_DISABLE 0 /* No setuid dumping */
406#define SUID_DUMP_USER 1 /* Dump as user of process */
407#define SUID_DUMP_ROOT 2 /* Dump as root */
408
6c5d5238 409/* mm flags */
f8af4da3 410
7288e118 411/* for SUID_DUMP_* above */
3cb4a0bb 412#define MMF_DUMPABLE_BITS 2
f8af4da3 413#define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
3cb4a0bb 414
942be387
ON
415extern void set_dumpable(struct mm_struct *mm, int value);
416/*
417 * This returns the actual value of the suid_dumpable flag. For things
418 * that are using this for checking for privilege transitions, it must
419 * test against SUID_DUMP_USER rather than treating it as a boolean
420 * value.
421 */
422static inline int __get_dumpable(unsigned long mm_flags)
423{
424 return mm_flags & MMF_DUMPABLE_MASK;
425}
426
427static inline int get_dumpable(struct mm_struct *mm)
428{
429 return __get_dumpable(mm->flags);
430}
431
3cb4a0bb
KH
432/* coredump filter bits */
433#define MMF_DUMP_ANON_PRIVATE 2
434#define MMF_DUMP_ANON_SHARED 3
435#define MMF_DUMP_MAPPED_PRIVATE 4
436#define MMF_DUMP_MAPPED_SHARED 5
82df3973 437#define MMF_DUMP_ELF_HEADERS 6
e575f111
KM
438#define MMF_DUMP_HUGETLB_PRIVATE 7
439#define MMF_DUMP_HUGETLB_SHARED 8
f8af4da3 440
3cb4a0bb 441#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
e575f111 442#define MMF_DUMP_FILTER_BITS 7
3cb4a0bb
KH
443#define MMF_DUMP_FILTER_MASK \
444 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
445#define MMF_DUMP_FILTER_DEFAULT \
e575f111 446 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
656eb2cd
RM
447 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
448
449#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
450# define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
451#else
452# define MMF_DUMP_MASK_DEFAULT_ELF 0
453#endif
f8af4da3
HD
454 /* leave room for more dump flags */
455#define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
ba76149f 456#define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
bafb282d 457#define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
f8af4da3 458
9f68f672
ON
459#define MMF_HAS_UPROBES 19 /* has uprobes */
460#define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
f8ac4ec9 461
f8af4da3 462#define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
6c5d5238 463
1da177e4
LT
464struct sighand_struct {
465 atomic_t count;
466 struct k_sigaction action[_NSIG];
467 spinlock_t siglock;
b8fceee1 468 wait_queue_head_t signalfd_wqh;
1da177e4
LT
469};
470
0e464814 471struct pacct_struct {
f6ec29a4
KK
472 int ac_flag;
473 long ac_exitcode;
0e464814 474 unsigned long ac_mem;
77787bfb
KK
475 cputime_t ac_utime, ac_stime;
476 unsigned long ac_minflt, ac_majflt;
0e464814
KK
477};
478
42c4ab41
SG
479struct cpu_itimer {
480 cputime_t expires;
481 cputime_t incr;
8356b5f9
SG
482 u32 error;
483 u32 incr_error;
42c4ab41
SG
484};
485
d37f761d
FW
486/**
487 * struct cputime - snaphsot of system and user cputime
488 * @utime: time spent in user mode
489 * @stime: time spent in system mode
490 *
491 * Gathers a generic snapshot of user and system time.
492 */
493struct cputime {
494 cputime_t utime;
495 cputime_t stime;
496};
497
f06febc9
FM
498/**
499 * struct task_cputime - collected CPU time counts
500 * @utime: time spent in user mode, in &cputime_t units
501 * @stime: time spent in kernel mode, in &cputime_t units
502 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
5ce73a4a 503 *
d37f761d
FW
504 * This is an extension of struct cputime that includes the total runtime
505 * spent by the task from the scheduler point of view.
506 *
507 * As a result, this structure groups together three kinds of CPU time
508 * that are tracked for threads and thread groups. Most things considering
f06febc9
FM
509 * CPU time want to group these counts together and treat all three
510 * of them in parallel.
511 */
512struct task_cputime {
513 cputime_t utime;
514 cputime_t stime;
515 unsigned long long sum_exec_runtime;
516};
517/* Alternate field names when used to cache expirations. */
518#define prof_exp stime
519#define virt_exp utime
520#define sched_exp sum_exec_runtime
521
4cd4c1b4
PZ
522#define INIT_CPUTIME \
523 (struct task_cputime) { \
64861634
MS
524 .utime = 0, \
525 .stime = 0, \
4cd4c1b4
PZ
526 .sum_exec_runtime = 0, \
527 }
528
a233f112
PZ
529#ifdef CONFIG_PREEMPT_COUNT
530#define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
531#else
532#define PREEMPT_DISABLED PREEMPT_ENABLED
533#endif
534
c99e6efe
PZ
535/*
536 * Disable preemption until the scheduler is running.
537 * Reset by start_kernel()->sched_init()->init_idle().
d86ee480
PZ
538 *
539 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
540 * before the scheduler is active -- see should_resched().
c99e6efe 541 */
a233f112 542#define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
c99e6efe 543
f06febc9 544/**
4cd4c1b4
PZ
545 * struct thread_group_cputimer - thread group interval timer counts
546 * @cputime: thread group interval timers.
547 * @running: non-zero when there are timers running and
548 * @cputime receives updates.
549 * @lock: lock for fields in this struct.
f06febc9
FM
550 *
551 * This structure contains the version of task_cputime, above, that is
4cd4c1b4 552 * used for thread group CPU timer calculations.
f06febc9 553 */
4cd4c1b4
PZ
554struct thread_group_cputimer {
555 struct task_cputime cputime;
556 int running;
ee30a7b2 557 raw_spinlock_t lock;
f06febc9 558};
f06febc9 559
4714d1d3 560#include <linux/rwsem.h>
5091faa4
MG
561struct autogroup;
562
1da177e4 563/*
e815f0a8 564 * NOTE! "signal_struct" does not have its own
1da177e4
LT
565 * locking, because a shared signal_struct always
566 * implies a shared sighand_struct, so locking
567 * sighand_struct is always a proper superset of
568 * the locking of signal_struct.
569 */
570struct signal_struct {
ea6d290c 571 atomic_t sigcnt;
1da177e4 572 atomic_t live;
b3ac022c 573 int nr_threads;
0c740d0a 574 struct list_head thread_head;
1da177e4
LT
575
576 wait_queue_head_t wait_chldexit; /* for wait4() */
577
578 /* current thread group signal load-balancing target: */
36c8b586 579 struct task_struct *curr_target;
1da177e4
LT
580
581 /* shared signal handling: */
582 struct sigpending shared_pending;
583
584 /* thread group exit support */
585 int group_exit_code;
586 /* overloaded:
587 * - notify group_exit_task when ->count is equal to notify_count
588 * - everyone except group_exit_task is stopped during signal delivery
589 * of fatal signals, group_exit_task processes the signal.
590 */
1da177e4 591 int notify_count;
07dd20e0 592 struct task_struct *group_exit_task;
1da177e4
LT
593
594 /* thread group stop support, overloads group_exit_code too */
595 int group_stop_count;
596 unsigned int flags; /* see SIGNAL_* flags below */
597
ebec18a6
LP
598 /*
599 * PR_SET_CHILD_SUBREAPER marks a process, like a service
600 * manager, to re-parent orphan (double-forking) child processes
601 * to this process instead of 'init'. The service manager is
602 * able to receive SIGCHLD signals and is able to investigate
603 * the process until it calls wait(). All children of this
604 * process will inherit a flag if they should look for a
605 * child_subreaper process at exit.
606 */
607 unsigned int is_child_subreaper:1;
608 unsigned int has_child_subreaper:1;
609
1da177e4 610 /* POSIX.1b Interval Timers */
5ed67f05
PE
611 int posix_timer_id;
612 struct list_head posix_timers;
1da177e4
LT
613
614 /* ITIMER_REAL timer for the process */
2ff678b8 615 struct hrtimer real_timer;
fea9d175 616 struct pid *leader_pid;
2ff678b8 617 ktime_t it_real_incr;
1da177e4 618
42c4ab41
SG
619 /*
620 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
621 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
622 * values are defined to 0 and 1 respectively
623 */
624 struct cpu_itimer it[2];
1da177e4 625
f06febc9 626 /*
4cd4c1b4
PZ
627 * Thread group totals for process CPU timers.
628 * See thread_group_cputimer(), et al, for details.
f06febc9 629 */
4cd4c1b4 630 struct thread_group_cputimer cputimer;
f06febc9
FM
631
632 /* Earliest-expiration cache. */
633 struct task_cputime cputime_expires;
634
635 struct list_head cpu_timers[3];
636
ab521dc0 637 struct pid *tty_old_pgrp;
1ec320af 638
1da177e4
LT
639 /* boolean value for session group leader */
640 int leader;
641
642 struct tty_struct *tty; /* NULL if no tty */
643
5091faa4
MG
644#ifdef CONFIG_SCHED_AUTOGROUP
645 struct autogroup *autogroup;
646#endif
1da177e4
LT
647 /*
648 * Cumulative resource counters for dead threads in the group,
649 * and for reaped dead child processes forked by this group.
650 * Live threads maintain their own counters and add to these
651 * in __exit_signal, except for the group leader.
652 */
32bd671d 653 cputime_t utime, stime, cutime, cstime;
9ac52315
LV
654 cputime_t gtime;
655 cputime_t cgtime;
9fbc42ea 656#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
d37f761d 657 struct cputime prev_cputime;
0cf55e1e 658#endif
1da177e4
LT
659 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
660 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
6eaeeaba 661 unsigned long inblock, oublock, cinblock, coublock;
1f10206c 662 unsigned long maxrss, cmaxrss;
940389b8 663 struct task_io_accounting ioac;
1da177e4 664
32bd671d
PZ
665 /*
666 * Cumulative ns of schedule CPU time fo dead threads in the
667 * group, not including a zombie group leader, (This only differs
668 * from jiffies_to_ns(utime + stime) if sched_clock uses something
669 * other than jiffies.)
670 */
671 unsigned long long sum_sched_runtime;
672
1da177e4
LT
673 /*
674 * We don't bother to synchronize most readers of this at all,
675 * because there is no reader checking a limit that actually needs
676 * to get both rlim_cur and rlim_max atomically, and either one
677 * alone is a single word that can safely be read normally.
678 * getrlimit/setrlimit use task_lock(current->group_leader) to
679 * protect this instead of the siglock, because they really
680 * have no need to disable irqs.
681 */
682 struct rlimit rlim[RLIM_NLIMITS];
683
0e464814
KK
684#ifdef CONFIG_BSD_PROCESS_ACCT
685 struct pacct_struct pacct; /* per-process accounting information */
686#endif
ad4ecbcb 687#ifdef CONFIG_TASKSTATS
ad4ecbcb
SN
688 struct taskstats *stats;
689#endif
522ed776
MT
690#ifdef CONFIG_AUDIT
691 unsigned audit_tty;
46e959ea 692 unsigned audit_tty_log_passwd;
522ed776
MT
693 struct tty_audit_buf *tty_audit_buf;
694#endif
4714d1d3
BB
695#ifdef CONFIG_CGROUPS
696 /*
77e4ef99
TH
697 * group_rwsem prevents new tasks from entering the threadgroup and
698 * member tasks from exiting,a more specifically, setting of
699 * PF_EXITING. fork and exit paths are protected with this rwsem
700 * using threadgroup_change_begin/end(). Users which require
701 * threadgroup to remain stable should use threadgroup_[un]lock()
702 * which also takes care of exec path. Currently, cgroup is the
703 * only user.
4714d1d3 704 */
257058ae 705 struct rw_semaphore group_rwsem;
4714d1d3 706#endif
28b83c51 707
e1e12d2f 708 oom_flags_t oom_flags;
a9c58b90
DR
709 short oom_score_adj; /* OOM kill score adjustment */
710 short oom_score_adj_min; /* OOM kill score adjustment min value.
711 * Only settable by CAP_SYS_RESOURCE. */
9b1bf12d
KM
712
713 struct mutex cred_guard_mutex; /* guard against foreign influences on
714 * credential calculations
715 * (notably. ptrace) */
1da177e4
LT
716};
717
718/*
719 * Bits in flags field of signal_struct.
720 */
721#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
ee77f075
ON
722#define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
723#define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
403bad72 724#define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
e4420551
ON
725/*
726 * Pending notifications to parent.
727 */
728#define SIGNAL_CLD_STOPPED 0x00000010
729#define SIGNAL_CLD_CONTINUED 0x00000020
730#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
1da177e4 731
fae5fa44
ON
732#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
733
ed5d2cac
ON
734/* If true, all threads except ->group_exit_task have pending SIGKILL */
735static inline int signal_group_exit(const struct signal_struct *sig)
736{
737 return (sig->flags & SIGNAL_GROUP_EXIT) ||
738 (sig->group_exit_task != NULL);
739}
740
1da177e4
LT
741/*
742 * Some day this will be a full-fledged user tracking system..
743 */
744struct user_struct {
745 atomic_t __count; /* reference count */
746 atomic_t processes; /* How many processes does this user have? */
747 atomic_t files; /* How many open files does this user have? */
748 atomic_t sigpending; /* How many pending signals does this user have? */
2d9048e2 749#ifdef CONFIG_INOTIFY_USER
0eeca283
RL
750 atomic_t inotify_watches; /* How many inotify watches does this user have? */
751 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
752#endif
4afeff85
EP
753#ifdef CONFIG_FANOTIFY
754 atomic_t fanotify_listeners;
755#endif
7ef9964e 756#ifdef CONFIG_EPOLL
52bd19f7 757 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
7ef9964e 758#endif
970a8645 759#ifdef CONFIG_POSIX_MQUEUE
1da177e4
LT
760 /* protected by mq_lock */
761 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
970a8645 762#endif
1da177e4
LT
763 unsigned long locked_shm; /* How many pages of mlocked shm ? */
764
765#ifdef CONFIG_KEYS
766 struct key *uid_keyring; /* UID specific keyring */
767 struct key *session_keyring; /* UID's default session keyring */
768#endif
769
770 /* Hash table maintenance information */
735de223 771 struct hlist_node uidhash_node;
7b44ab97 772 kuid_t uid;
24e377a8 773
cdd6c482 774#ifdef CONFIG_PERF_EVENTS
789f90fc
PZ
775 atomic_long_t locked_vm;
776#endif
1da177e4
LT
777};
778
eb41d946 779extern int uids_sysfs_init(void);
5cb350ba 780
7b44ab97 781extern struct user_struct *find_user(kuid_t);
1da177e4
LT
782
783extern struct user_struct root_user;
784#define INIT_USER (&root_user)
785
b6dff3ec 786
1da177e4
LT
787struct backing_dev_info;
788struct reclaim_state;
789
52f17b6c 790#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1da177e4
LT
791struct sched_info {
792 /* cumulative counters */
2d72376b 793 unsigned long pcount; /* # of times run on this cpu */
9c2c4802 794 unsigned long long run_delay; /* time spent waiting on a runqueue */
1da177e4
LT
795
796 /* timestamps */
172ba844
BS
797 unsigned long long last_arrival,/* when we last ran on a cpu */
798 last_queued; /* when we were last queued to run */
1da177e4 799};
52f17b6c 800#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
1da177e4 801
ca74e92b
SN
802#ifdef CONFIG_TASK_DELAY_ACCT
803struct task_delay_info {
804 spinlock_t lock;
805 unsigned int flags; /* Private per-task flags */
806
807 /* For each stat XXX, add following, aligned appropriately
808 *
809 * struct timespec XXX_start, XXX_end;
810 * u64 XXX_delay;
811 * u32 XXX_count;
812 *
813 * Atomicity of updates to XXX_delay, XXX_count protected by
814 * single lock above (split into XXX_lock if contention is an issue).
815 */
0ff92245
SN
816
817 /*
818 * XXX_count is incremented on every XXX operation, the delay
819 * associated with the operation is added to XXX_delay.
820 * XXX_delay contains the accumulated delay time in nanoseconds.
821 */
822 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
823 u64 blkio_delay; /* wait for sync block io completion */
824 u64 swapin_delay; /* wait for swapin block io completion */
825 u32 blkio_count; /* total count of the number of sync block */
826 /* io operations performed */
827 u32 swapin_count; /* total count of the number of swapin block */
828 /* io operations performed */
873b4771
KK
829
830 struct timespec freepages_start, freepages_end;
831 u64 freepages_delay; /* wait for memory reclaim */
832 u32 freepages_count; /* total count of memory reclaim */
ca74e92b 833};
52f17b6c
CS
834#endif /* CONFIG_TASK_DELAY_ACCT */
835
836static inline int sched_info_on(void)
837{
838#ifdef CONFIG_SCHEDSTATS
839 return 1;
840#elif defined(CONFIG_TASK_DELAY_ACCT)
841 extern int delayacct_on;
842 return delayacct_on;
843#else
844 return 0;
ca74e92b 845#endif
52f17b6c 846}
ca74e92b 847
d15bcfdb
IM
848enum cpu_idle_type {
849 CPU_IDLE,
850 CPU_NOT_IDLE,
851 CPU_NEWLY_IDLE,
852 CPU_MAX_IDLE_TYPES
1da177e4
LT
853};
854
1399fa78
NR
855/*
856 * Increase resolution of cpu_power calculations
857 */
858#define SCHED_POWER_SHIFT 10
859#define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
1da177e4 860
1399fa78
NR
861/*
862 * sched-domains (multiprocessor balancing) declarations:
863 */
2dd73a4f 864#ifdef CONFIG_SMP
b5d978e0
PZ
865#define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
866#define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
867#define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
868#define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
c88d5910 869#define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
b5d978e0 870#define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
b5d978e0 871#define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
b5d978e0
PZ
872#define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
873#define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
532cb4c4 874#define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
b5d978e0 875#define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
e3589f6c 876#define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
3a7053b3 877#define SD_NUMA 0x4000 /* cross-node balancing */
5c45bf27 878
532cb4c4
MN
879extern int __weak arch_sd_sibiling_asym_packing(void);
880
1d3504fc
HS
881struct sched_domain_attr {
882 int relax_domain_level;
883};
884
885#define SD_ATTR_INIT (struct sched_domain_attr) { \
886 .relax_domain_level = -1, \
887}
888
60495e77
PZ
889extern int sched_domain_level_max;
890
5e6521ea
LZ
891struct sched_group;
892
1da177e4
LT
893struct sched_domain {
894 /* These fields must be setup */
895 struct sched_domain *parent; /* top domain must be null terminated */
1a848870 896 struct sched_domain *child; /* bottom domain must be null terminated */
1da177e4 897 struct sched_group *groups; /* the balancing groups of the domain */
1da177e4
LT
898 unsigned long min_interval; /* Minimum balance interval ms */
899 unsigned long max_interval; /* Maximum balance interval ms */
900 unsigned int busy_factor; /* less balancing by factor if busy */
901 unsigned int imbalance_pct; /* No balance until over watermark */
1da177e4 902 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
7897986b
NP
903 unsigned int busy_idx;
904 unsigned int idle_idx;
905 unsigned int newidle_idx;
906 unsigned int wake_idx;
147cbb4b 907 unsigned int forkexec_idx;
a52bfd73 908 unsigned int smt_gain;
25f55d9d
VG
909
910 int nohz_idle; /* NOHZ IDLE status */
1da177e4 911 int flags; /* See SD_* */
60495e77 912 int level;
1da177e4
LT
913
914 /* Runtime fields. */
915 unsigned long last_balance; /* init to jiffies. units in jiffies */
916 unsigned int balance_interval; /* initialise to 1. units in ms. */
917 unsigned int nr_balance_failed; /* initialise to 0 */
918
f48627e6 919 /* idle_balance() stats */
9bd721c5 920 u64 max_newidle_lb_cost;
f48627e6 921 unsigned long next_decay_max_lb_cost;
2398f2c6 922
1da177e4
LT
923#ifdef CONFIG_SCHEDSTATS
924 /* load_balance() stats */
480b9434
KC
925 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
926 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
927 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
928 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
929 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
930 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
931 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
932 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1da177e4
LT
933
934 /* Active load balancing */
480b9434
KC
935 unsigned int alb_count;
936 unsigned int alb_failed;
937 unsigned int alb_pushed;
1da177e4 938
68767a0a 939 /* SD_BALANCE_EXEC stats */
480b9434
KC
940 unsigned int sbe_count;
941 unsigned int sbe_balanced;
942 unsigned int sbe_pushed;
1da177e4 943
68767a0a 944 /* SD_BALANCE_FORK stats */
480b9434
KC
945 unsigned int sbf_count;
946 unsigned int sbf_balanced;
947 unsigned int sbf_pushed;
68767a0a 948
1da177e4 949 /* try_to_wake_up() stats */
480b9434
KC
950 unsigned int ttwu_wake_remote;
951 unsigned int ttwu_move_affine;
952 unsigned int ttwu_move_balance;
1da177e4 953#endif
a5d8c348
IM
954#ifdef CONFIG_SCHED_DEBUG
955 char *name;
956#endif
dce840a0
PZ
957 union {
958 void *private; /* used during construction */
959 struct rcu_head rcu; /* used during destruction */
960 };
6c99e9ad 961
669c55e9 962 unsigned int span_weight;
4200efd9
IM
963 /*
964 * Span of all CPUs in this domain.
965 *
966 * NOTE: this field is variable length. (Allocated dynamically
967 * by attaching extra space to the end of the structure,
968 * depending on how many CPUs the kernel has booted up with)
4200efd9
IM
969 */
970 unsigned long span[0];
1da177e4
LT
971};
972
758b2cdc
RR
973static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
974{
6c99e9ad 975 return to_cpumask(sd->span);
758b2cdc
RR
976}
977
acc3f5d7 978extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1d3504fc 979 struct sched_domain_attr *dattr_new);
029190c5 980
acc3f5d7
RR
981/* Allocate an array of sched domains, for partition_sched_domains(). */
982cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
983void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
984
39be3501
PZ
985bool cpus_share_cache(int this_cpu, int that_cpu);
986
1b427c15 987#else /* CONFIG_SMP */
1da177e4 988
1b427c15 989struct sched_domain_attr;
d02c7a8c 990
1b427c15 991static inline void
acc3f5d7 992partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1b427c15
IM
993 struct sched_domain_attr *dattr_new)
994{
d02c7a8c 995}
39be3501
PZ
996
997static inline bool cpus_share_cache(int this_cpu, int that_cpu)
998{
999 return true;
1000}
1001
1b427c15 1002#endif /* !CONFIG_SMP */
1da177e4 1003
47fe38fc 1004
1da177e4 1005struct io_context; /* See blkdev.h */
1da177e4 1006
1da177e4 1007
383f2835 1008#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
36c8b586 1009extern void prefetch_stack(struct task_struct *t);
383f2835
CK
1010#else
1011static inline void prefetch_stack(struct task_struct *t) { }
1012#endif
1da177e4
LT
1013
1014struct audit_context; /* See audit.c */
1015struct mempolicy;
b92ce558 1016struct pipe_inode_info;
4865ecf1 1017struct uts_namespace;
1da177e4 1018
20b8a59f 1019struct load_weight {
9dbdb155
PZ
1020 unsigned long weight;
1021 u32 inv_weight;
20b8a59f
IM
1022};
1023
9d85f21c
PT
1024struct sched_avg {
1025 /*
1026 * These sums represent an infinite geometric series and so are bound
239003ea 1027 * above by 1024/(1-y). Thus we only need a u32 to store them for all
9d85f21c
PT
1028 * choices of y < 1-2^(-32)*1024.
1029 */
1030 u32 runnable_avg_sum, runnable_avg_period;
1031 u64 last_runnable_update;
9ee474f5 1032 s64 decay_count;
2dac754e 1033 unsigned long load_avg_contrib;
9d85f21c
PT
1034};
1035
94c18227 1036#ifdef CONFIG_SCHEDSTATS
41acab88 1037struct sched_statistics {
20b8a59f 1038 u64 wait_start;
94c18227 1039 u64 wait_max;
6d082592
AV
1040 u64 wait_count;
1041 u64 wait_sum;
8f0dfc34
AV
1042 u64 iowait_count;
1043 u64 iowait_sum;
94c18227 1044
20b8a59f 1045 u64 sleep_start;
20b8a59f 1046 u64 sleep_max;
94c18227
IM
1047 s64 sum_sleep_runtime;
1048
1049 u64 block_start;
20b8a59f
IM
1050 u64 block_max;
1051 u64 exec_max;
eba1ed4b 1052 u64 slice_max;
cc367732 1053
cc367732
IM
1054 u64 nr_migrations_cold;
1055 u64 nr_failed_migrations_affine;
1056 u64 nr_failed_migrations_running;
1057 u64 nr_failed_migrations_hot;
1058 u64 nr_forced_migrations;
cc367732
IM
1059
1060 u64 nr_wakeups;
1061 u64 nr_wakeups_sync;
1062 u64 nr_wakeups_migrate;
1063 u64 nr_wakeups_local;
1064 u64 nr_wakeups_remote;
1065 u64 nr_wakeups_affine;
1066 u64 nr_wakeups_affine_attempts;
1067 u64 nr_wakeups_passive;
1068 u64 nr_wakeups_idle;
41acab88
LDM
1069};
1070#endif
1071
1072struct sched_entity {
1073 struct load_weight load; /* for load-balancing */
1074 struct rb_node run_node;
1075 struct list_head group_node;
1076 unsigned int on_rq;
1077
1078 u64 exec_start;
1079 u64 sum_exec_runtime;
1080 u64 vruntime;
1081 u64 prev_sum_exec_runtime;
1082
41acab88
LDM
1083 u64 nr_migrations;
1084
41acab88
LDM
1085#ifdef CONFIG_SCHEDSTATS
1086 struct sched_statistics statistics;
94c18227
IM
1087#endif
1088
20b8a59f 1089#ifdef CONFIG_FAIR_GROUP_SCHED
fed14d45 1090 int depth;
20b8a59f
IM
1091 struct sched_entity *parent;
1092 /* rq on which this entity is (to be) queued: */
1093 struct cfs_rq *cfs_rq;
1094 /* rq "owned" by this entity/group: */
1095 struct cfs_rq *my_q;
1096#endif
8bd75c77 1097
141965c7 1098#ifdef CONFIG_SMP
f4e26b12 1099 /* Per-entity load-tracking */
9d85f21c
PT
1100 struct sched_avg avg;
1101#endif
20b8a59f 1102};
70b97a7f 1103
fa717060
PZ
1104struct sched_rt_entity {
1105 struct list_head run_list;
78f2c7db 1106 unsigned long timeout;
57d2aa00 1107 unsigned long watchdog_stamp;
bee367ed 1108 unsigned int time_slice;
6f505b16 1109
58d6c2d7 1110 struct sched_rt_entity *back;
052f1dc7 1111#ifdef CONFIG_RT_GROUP_SCHED
6f505b16
PZ
1112 struct sched_rt_entity *parent;
1113 /* rq on which this entity is (to be) queued: */
1114 struct rt_rq *rt_rq;
1115 /* rq "owned" by this entity/group: */
1116 struct rt_rq *my_q;
1117#endif
fa717060
PZ
1118};
1119
aab03e05
DF
1120struct sched_dl_entity {
1121 struct rb_node rb_node;
1122
1123 /*
1124 * Original scheduling parameters. Copied here from sched_attr
1125 * during sched_setscheduler2(), they will remain the same until
1126 * the next sched_setscheduler2().
1127 */
1128 u64 dl_runtime; /* maximum runtime for each instance */
1129 u64 dl_deadline; /* relative deadline of each instance */
755378a4 1130 u64 dl_period; /* separation of two instances (period) */
332ac17e 1131 u64 dl_bw; /* dl_runtime / dl_deadline */
aab03e05
DF
1132
1133 /*
1134 * Actual scheduling parameters. Initialized with the values above,
1135 * they are continously updated during task execution. Note that
1136 * the remaining runtime could be < 0 in case we are in overrun.
1137 */
1138 s64 runtime; /* remaining runtime for this instance */
1139 u64 deadline; /* absolute deadline for this instance */
1140 unsigned int flags; /* specifying the scheduler behaviour */
1141
1142 /*
1143 * Some bool flags:
1144 *
1145 * @dl_throttled tells if we exhausted the runtime. If so, the
1146 * task has to wait for a replenishment to be performed at the
1147 * next firing of dl_timer.
1148 *
1149 * @dl_new tells if a new instance arrived. If so we must
1150 * start executing it with full runtime and reset its absolute
1151 * deadline;
2d3d891d
DF
1152 *
1153 * @dl_boosted tells if we are boosted due to DI. If so we are
1154 * outside bandwidth enforcement mechanism (but only until we
1155 * exit the critical section).
aab03e05 1156 */
2d3d891d 1157 int dl_throttled, dl_new, dl_boosted;
aab03e05
DF
1158
1159 /*
1160 * Bandwidth enforcement timer. Each -deadline task has its
1161 * own bandwidth to be enforced, thus we need one timer per task.
1162 */
1163 struct hrtimer dl_timer;
1164};
8bd75c77 1165
86848966
PM
1166struct rcu_node;
1167
8dc85d54
PZ
1168enum perf_event_task_context {
1169 perf_invalid_context = -1,
1170 perf_hw_context = 0,
89a1e187 1171 perf_sw_context,
8dc85d54
PZ
1172 perf_nr_task_contexts,
1173};
1174
1da177e4
LT
1175struct task_struct {
1176 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
f7e4217b 1177 void *stack;
1da177e4 1178 atomic_t usage;
97dc32cd
WC
1179 unsigned int flags; /* per process flags, defined below */
1180 unsigned int ptrace;
1da177e4 1181
2dd73a4f 1182#ifdef CONFIG_SMP
fa14ff4a 1183 struct llist_node wake_entry;
3ca7a440 1184 int on_cpu;
62470419
MW
1185 struct task_struct *last_wakee;
1186 unsigned long wakee_flips;
1187 unsigned long wakee_flip_decay_ts;
ac66f547
PZ
1188
1189 int wake_cpu;
2dd73a4f 1190#endif
fd2f4419 1191 int on_rq;
50e645a8 1192
b29739f9 1193 int prio, static_prio, normal_prio;
c7aceaba 1194 unsigned int rt_priority;
5522d5d5 1195 const struct sched_class *sched_class;
20b8a59f 1196 struct sched_entity se;
fa717060 1197 struct sched_rt_entity rt;
8323f26c
PZ
1198#ifdef CONFIG_CGROUP_SCHED
1199 struct task_group *sched_task_group;
1200#endif
aab03e05 1201 struct sched_dl_entity dl;
1da177e4 1202
e107be36
AK
1203#ifdef CONFIG_PREEMPT_NOTIFIERS
1204 /* list of struct preempt_notifier: */
1205 struct hlist_head preempt_notifiers;
1206#endif
1207
6c5c9341 1208#ifdef CONFIG_BLK_DEV_IO_TRACE
2056a782 1209 unsigned int btrace_seq;
6c5c9341 1210#endif
1da177e4 1211
97dc32cd 1212 unsigned int policy;
29baa747 1213 int nr_cpus_allowed;
1da177e4 1214 cpumask_t cpus_allowed;
1da177e4 1215
a57eb940 1216#ifdef CONFIG_PREEMPT_RCU
e260be67 1217 int rcu_read_lock_nesting;
f41d911f 1218 char rcu_read_unlock_special;
f41d911f 1219 struct list_head rcu_node_entry;
a57eb940
PM
1220#endif /* #ifdef CONFIG_PREEMPT_RCU */
1221#ifdef CONFIG_TREE_PREEMPT_RCU
1222 struct rcu_node *rcu_blocked_node;
f41d911f 1223#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
24278d14
PM
1224#ifdef CONFIG_RCU_BOOST
1225 struct rt_mutex *rcu_boost_mutex;
1226#endif /* #ifdef CONFIG_RCU_BOOST */
e260be67 1227
52f17b6c 1228#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1da177e4
LT
1229 struct sched_info sched_info;
1230#endif
1231
1232 struct list_head tasks;
806c09a7 1233#ifdef CONFIG_SMP
917b627d 1234 struct plist_node pushable_tasks;
1baca4ce 1235 struct rb_node pushable_dl_tasks;
806c09a7 1236#endif
1da177e4
LT
1237
1238 struct mm_struct *mm, *active_mm;
4471a675
JK
1239#ifdef CONFIG_COMPAT_BRK
1240 unsigned brk_randomized:1;
1241#endif
615d6e87
DB
1242 /* per-thread vma caching */
1243 u32 vmacache_seqnum;
1244 struct vm_area_struct *vmacache[VMACACHE_SIZE];
34e55232
KH
1245#if defined(SPLIT_RSS_COUNTING)
1246 struct task_rss_stat rss_stat;
1247#endif
1da177e4 1248/* task state */
97dc32cd 1249 int exit_state;
1da177e4
LT
1250 int exit_code, exit_signal;
1251 int pdeath_signal; /* The signal sent when the parent dies */
a8f072c1 1252 unsigned int jobctl; /* JOBCTL_*, siglock protected */
9b89f6ba
AE
1253
1254 /* Used for emulating ABI behavior of previous Linux versions */
97dc32cd 1255 unsigned int personality;
9b89f6ba 1256
f9ce1f1c
KT
1257 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1258 * execve */
8f0dfc34
AV
1259 unsigned in_iowait:1;
1260
259e5e6c
AL
1261 /* task may not gain privileges */
1262 unsigned no_new_privs:1;
ca94c442
LP
1263
1264 /* Revert to default priority/policy when forking */
1265 unsigned sched_reset_on_fork:1;
a8e4f2ea 1266 unsigned sched_contributes_to_load:1;
ca94c442 1267
1da177e4
LT
1268 pid_t pid;
1269 pid_t tgid;
0a425405 1270
1314562a 1271#ifdef CONFIG_CC_STACKPROTECTOR
0a425405
AV
1272 /* Canary value for the -fstack-protector gcc feature */
1273 unsigned long stack_canary;
1314562a 1274#endif
4d1d61a6 1275 /*
1da177e4 1276 * pointers to (original) parent process, youngest child, younger sibling,
4d1d61a6 1277 * older sibling, respectively. (p->father can be replaced with
f470021a 1278 * p->real_parent->pid)
1da177e4 1279 */
abd63bc3
KC
1280 struct task_struct __rcu *real_parent; /* real parent process */
1281 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1da177e4 1282 /*
f470021a 1283 * children/sibling forms the list of my natural children
1da177e4
LT
1284 */
1285 struct list_head children; /* list of my children */
1286 struct list_head sibling; /* linkage in my parent's children list */
1287 struct task_struct *group_leader; /* threadgroup leader */
1288
f470021a
RM
1289 /*
1290 * ptraced is the list of tasks this task is using ptrace on.
1291 * This includes both natural children and PTRACE_ATTACH targets.
1292 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1293 */
1294 struct list_head ptraced;
1295 struct list_head ptrace_entry;
1296
1da177e4 1297 /* PID/PID hash table linkage. */
92476d7f 1298 struct pid_link pids[PIDTYPE_MAX];
47e65328 1299 struct list_head thread_group;
0c740d0a 1300 struct list_head thread_node;
1da177e4
LT
1301
1302 struct completion *vfork_done; /* for vfork() */
1303 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1304 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1305
c66f08be 1306 cputime_t utime, stime, utimescaled, stimescaled;
9ac52315 1307 cputime_t gtime;
9fbc42ea 1308#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
d37f761d 1309 struct cputime prev_cputime;
6a61671b
FW
1310#endif
1311#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1312 seqlock_t vtime_seqlock;
1313 unsigned long long vtime_snap;
1314 enum {
1315 VTIME_SLEEPING = 0,
1316 VTIME_USER,
1317 VTIME_SYS,
1318 } vtime_snap_whence;
d99ca3b9 1319#endif
1da177e4 1320 unsigned long nvcsw, nivcsw; /* context switch counts */
924b42d5
TJ
1321 struct timespec start_time; /* monotonic time */
1322 struct timespec real_start_time; /* boot based time */
1da177e4
LT
1323/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1324 unsigned long min_flt, maj_flt;
1325
f06febc9 1326 struct task_cputime cputime_expires;
1da177e4
LT
1327 struct list_head cpu_timers[3];
1328
1329/* process credentials */
1b0ba1c9 1330 const struct cred __rcu *real_cred; /* objective and real subjective task
3b11a1de 1331 * credentials (COW) */
1b0ba1c9 1332 const struct cred __rcu *cred; /* effective (overridable) subjective task
3b11a1de 1333 * credentials (COW) */
36772092
PBG
1334 char comm[TASK_COMM_LEN]; /* executable name excluding path
1335 - access with [gs]et_task_comm (which lock
1336 it with task_lock())
221af7f8 1337 - initialized normally by setup_new_exec */
1da177e4
LT
1338/* file system info */
1339 int link_count, total_link_count;
3d5b6fcc 1340#ifdef CONFIG_SYSVIPC
1da177e4
LT
1341/* ipc stuff */
1342 struct sysv_sem sysvsem;
3d5b6fcc 1343#endif
e162b39a 1344#ifdef CONFIG_DETECT_HUNG_TASK
82a1fcb9 1345/* hung task detection */
82a1fcb9
IM
1346 unsigned long last_switch_count;
1347#endif
1da177e4
LT
1348/* CPU-specific state of this task */
1349 struct thread_struct thread;
1350/* filesystem information */
1351 struct fs_struct *fs;
1352/* open file information */
1353 struct files_struct *files;
1651e14e 1354/* namespaces */
ab516013 1355 struct nsproxy *nsproxy;
1da177e4
LT
1356/* signal handlers */
1357 struct signal_struct *signal;
1358 struct sighand_struct *sighand;
1359
1360 sigset_t blocked, real_blocked;
f3de272b 1361 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1da177e4
LT
1362 struct sigpending pending;
1363
1364 unsigned long sas_ss_sp;
1365 size_t sas_ss_size;
1366 int (*notifier)(void *priv);
1367 void *notifier_data;
1368 sigset_t *notifier_mask;
67d12145 1369 struct callback_head *task_works;
e73f8959 1370
1da177e4 1371 struct audit_context *audit_context;
bfef93a5 1372#ifdef CONFIG_AUDITSYSCALL
e1760bd5 1373 kuid_t loginuid;
4746ec5b 1374 unsigned int sessionid;
bfef93a5 1375#endif
932ecebb 1376 struct seccomp seccomp;
1da177e4
LT
1377
1378/* Thread group tracking */
1379 u32 parent_exec_id;
1380 u32 self_exec_id;
58568d2a
MX
1381/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1382 * mempolicy */
1da177e4 1383 spinlock_t alloc_lock;
1da177e4 1384
b29739f9 1385 /* Protection of the PI data structures: */
1d615482 1386 raw_spinlock_t pi_lock;
b29739f9 1387
23f78d4a
IM
1388#ifdef CONFIG_RT_MUTEXES
1389 /* PI waiters blocked on a rt_mutex held by this task */
fb00aca4
PZ
1390 struct rb_root pi_waiters;
1391 struct rb_node *pi_waiters_leftmost;
23f78d4a
IM
1392 /* Deadlock detection and priority inheritance handling */
1393 struct rt_mutex_waiter *pi_blocked_on;
2d3d891d
DF
1394 /* Top pi_waiters task */
1395 struct task_struct *pi_top_task;
23f78d4a
IM
1396#endif
1397
408894ee
IM
1398#ifdef CONFIG_DEBUG_MUTEXES
1399 /* mutex deadlock detection */
1400 struct mutex_waiter *blocked_on;
1401#endif
de30a2b3
IM
1402#ifdef CONFIG_TRACE_IRQFLAGS
1403 unsigned int irq_events;
de30a2b3 1404 unsigned long hardirq_enable_ip;
de30a2b3 1405 unsigned long hardirq_disable_ip;
fa1452e8 1406 unsigned int hardirq_enable_event;
de30a2b3 1407 unsigned int hardirq_disable_event;
fa1452e8
HS
1408 int hardirqs_enabled;
1409 int hardirq_context;
de30a2b3 1410 unsigned long softirq_disable_ip;
de30a2b3 1411 unsigned long softirq_enable_ip;
fa1452e8 1412 unsigned int softirq_disable_event;
de30a2b3 1413 unsigned int softirq_enable_event;
fa1452e8 1414 int softirqs_enabled;
de30a2b3
IM
1415 int softirq_context;
1416#endif
fbb9ce95 1417#ifdef CONFIG_LOCKDEP
bdb9441e 1418# define MAX_LOCK_DEPTH 48UL
fbb9ce95
IM
1419 u64 curr_chain_key;
1420 int lockdep_depth;
fbb9ce95 1421 unsigned int lockdep_recursion;
c7aceaba 1422 struct held_lock held_locks[MAX_LOCK_DEPTH];
cf40bd16 1423 gfp_t lockdep_reclaim_gfp;
fbb9ce95 1424#endif
408894ee 1425
1da177e4
LT
1426/* journalling filesystem info */
1427 void *journal_info;
1428
d89d8796 1429/* stacked block device info */
bddd87c7 1430 struct bio_list *bio_list;
d89d8796 1431
73c10101
JA
1432#ifdef CONFIG_BLOCK
1433/* stack plugging */
1434 struct blk_plug *plug;
1435#endif
1436
1da177e4
LT
1437/* VM state */
1438 struct reclaim_state *reclaim_state;
1439
1da177e4
LT
1440 struct backing_dev_info *backing_dev_info;
1441
1442 struct io_context *io_context;
1443
1444 unsigned long ptrace_message;
1445 siginfo_t *last_siginfo; /* For ptrace use. */
7c3ab738 1446 struct task_io_accounting ioac;
8f0ab514 1447#if defined(CONFIG_TASK_XACCT)
1da177e4
LT
1448 u64 acct_rss_mem1; /* accumulated rss usage */
1449 u64 acct_vm_mem1; /* accumulated virtual memory usage */
49b5cf34 1450 cputime_t acct_timexpd; /* stime + utime since last update */
1da177e4
LT
1451#endif
1452#ifdef CONFIG_CPUSETS
58568d2a 1453 nodemask_t mems_allowed; /* Protected by alloc_lock */
cc9a6c87 1454 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
825a46af 1455 int cpuset_mem_spread_rotor;
6adef3eb 1456 int cpuset_slab_spread_rotor;
1da177e4 1457#endif
ddbcc7e8 1458#ifdef CONFIG_CGROUPS
817929ec 1459 /* Control Group info protected by css_set_lock */
2c392b8c 1460 struct css_set __rcu *cgroups;
817929ec
PM
1461 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1462 struct list_head cg_list;
ddbcc7e8 1463#endif
42b2dd0a 1464#ifdef CONFIG_FUTEX
0771dfef 1465 struct robust_list_head __user *robust_list;
34f192c6
IM
1466#ifdef CONFIG_COMPAT
1467 struct compat_robust_list_head __user *compat_robust_list;
1468#endif
c87e2837
IM
1469 struct list_head pi_state_list;
1470 struct futex_pi_state *pi_state_cache;
c7aceaba 1471#endif
cdd6c482 1472#ifdef CONFIG_PERF_EVENTS
8dc85d54 1473 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
cdd6c482
IM
1474 struct mutex perf_event_mutex;
1475 struct list_head perf_event_list;
a63eaf34 1476#endif
8f47b187
TG
1477#ifdef CONFIG_DEBUG_PREEMPT
1478 unsigned long preempt_disable_ip;
1479#endif
c7aceaba 1480#ifdef CONFIG_NUMA
58568d2a 1481 struct mempolicy *mempolicy; /* Protected by alloc_lock */
c7aceaba 1482 short il_next;
207205a2 1483 short pref_node_fork;
42b2dd0a 1484#endif
cbee9f88
PZ
1485#ifdef CONFIG_NUMA_BALANCING
1486 int numa_scan_seq;
cbee9f88 1487 unsigned int numa_scan_period;
598f0ec0 1488 unsigned int numa_scan_period_max;
de1c9ce6 1489 int numa_preferred_nid;
6b9a7460 1490 unsigned long numa_migrate_retry;
cbee9f88 1491 u64 node_stamp; /* migration stamp */
7e2703e6
RR
1492 u64 last_task_numa_placement;
1493 u64 last_sum_exec_runtime;
cbee9f88 1494 struct callback_head numa_work;
f809ca9a 1495
8c8a743c
PZ
1496 struct list_head numa_entry;
1497 struct numa_group *numa_group;
1498
745d6147
MG
1499 /*
1500 * Exponential decaying average of faults on a per-node basis.
1501 * Scheduling placement decisions are made based on the these counts.
1502 * The values remain static for the duration of a PTE scan
1503 */
ff1df896 1504 unsigned long *numa_faults_memory;
83e1d2cd 1505 unsigned long total_numa_faults;
745d6147
MG
1506
1507 /*
1508 * numa_faults_buffer records faults per node during the current
ff1df896
RR
1509 * scan window. When the scan completes, the counts in
1510 * numa_faults_memory decay and these values are copied.
745d6147 1511 */
ff1df896 1512 unsigned long *numa_faults_buffer_memory;
745d6147 1513
50ec8a40
RR
1514 /*
1515 * Track the nodes the process was running on when a NUMA hinting
1516 * fault was incurred.
1517 */
1518 unsigned long *numa_faults_cpu;
1519 unsigned long *numa_faults_buffer_cpu;
1520
04bb2f94
RR
1521 /*
1522 * numa_faults_locality tracks if faults recorded during the last
1523 * scan window were remote/local. The task scan period is adapted
1524 * based on the locality of the faults with different weights
1525 * depending on whether they were shared or private faults
1526 */
1527 unsigned long numa_faults_locality[2];
1528
b32e86b4 1529 unsigned long numa_pages_migrated;
cbee9f88
PZ
1530#endif /* CONFIG_NUMA_BALANCING */
1531
e56d0903 1532 struct rcu_head rcu;
b92ce558
JA
1533
1534 /*
1535 * cache last used pipe for splice
1536 */
1537 struct pipe_inode_info *splice_pipe;
5640f768
ED
1538
1539 struct page_frag task_frag;
1540
ca74e92b
SN
1541#ifdef CONFIG_TASK_DELAY_ACCT
1542 struct task_delay_info *delays;
f4f154fd
AM
1543#endif
1544#ifdef CONFIG_FAULT_INJECTION
1545 int make_it_fail;
ca74e92b 1546#endif
9d823e8f
WF
1547 /*
1548 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1549 * balance_dirty_pages() for some dirty throttling pause
1550 */
1551 int nr_dirtied;
1552 int nr_dirtied_pause;
83712358 1553 unsigned long dirty_paused_when; /* start of a write-and-pause period */
9d823e8f 1554
9745512c
AV
1555#ifdef CONFIG_LATENCYTOP
1556 int latency_record_count;
1557 struct latency_record latency_record[LT_SAVECOUNT];
1558#endif
6976675d
AV
1559 /*
1560 * time slack values; these are used to round up poll() and
1561 * select() etc timeout values. These are in nanoseconds.
1562 */
1563 unsigned long timer_slack_ns;
1564 unsigned long default_timer_slack_ns;
f8d570a4 1565
fb52607a 1566#ifdef CONFIG_FUNCTION_GRAPH_TRACER
3ad2f3fb 1567 /* Index of current stored address in ret_stack */
f201ae23
FW
1568 int curr_ret_stack;
1569 /* Stack of return addresses for return function tracing */
1570 struct ftrace_ret_stack *ret_stack;
8aef2d28
SR
1571 /* time stamp for last schedule */
1572 unsigned long long ftrace_timestamp;
f201ae23
FW
1573 /*
1574 * Number of functions that haven't been traced
1575 * because of depth overrun.
1576 */
1577 atomic_t trace_overrun;
380c4b14
FW
1578 /* Pause for the tracing */
1579 atomic_t tracing_graph_pause;
f201ae23 1580#endif
ea4e2bc4
SR
1581#ifdef CONFIG_TRACING
1582 /* state flags for use by tracers */
1583 unsigned long trace;
b1cff0ad 1584 /* bitmask and counter of trace recursion */
261842b7
SR
1585 unsigned long trace_recursion;
1586#endif /* CONFIG_TRACING */
c255a458 1587#ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
569b846d
KH
1588 struct memcg_batch_info {
1589 int do_batch; /* incremented when batch uncharge started */
1590 struct mem_cgroup *memcg; /* target memcg of uncharge */
7ffd4ca7
JW
1591 unsigned long nr_pages; /* uncharged usage */
1592 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
569b846d 1593 } memcg_batch;
0e9d92f2 1594 unsigned int memcg_kmem_skip_account;
519e5247 1595 struct memcg_oom_info {
49426420
JW
1596 struct mem_cgroup *memcg;
1597 gfp_t gfp_mask;
1598 int order;
519e5247
JW
1599 unsigned int may_oom:1;
1600 } memcg_oom;
569b846d 1601#endif
0326f5a9
SD
1602#ifdef CONFIG_UPROBES
1603 struct uprobe_task *utask;
0326f5a9 1604#endif
cafe5635
KO
1605#if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1606 unsigned int sequential_io;
1607 unsigned int sequential_io_avg;
1608#endif
1da177e4
LT
1609};
1610
76e6eee0 1611/* Future-safe accessor for struct task_struct's cpus_allowed. */
a4636818 1612#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
76e6eee0 1613
6688cc05
PZ
1614#define TNF_MIGRATED 0x01
1615#define TNF_NO_GROUP 0x02
dabe1d99 1616#define TNF_SHARED 0x04
04bb2f94 1617#define TNF_FAULT_LOCAL 0x08
6688cc05 1618
cbee9f88 1619#ifdef CONFIG_NUMA_BALANCING
6688cc05 1620extern void task_numa_fault(int last_node, int node, int pages, int flags);
e29cf08b 1621extern pid_t task_numa_group_id(struct task_struct *p);
1a687c2e 1622extern void set_numabalancing_state(bool enabled);
82727018 1623extern void task_numa_free(struct task_struct *p);
10f39042
RR
1624extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1625 int src_nid, int dst_cpu);
cbee9f88 1626#else
ac8e895b 1627static inline void task_numa_fault(int last_node, int node, int pages,
6688cc05 1628 int flags)
cbee9f88
PZ
1629{
1630}
e29cf08b
MG
1631static inline pid_t task_numa_group_id(struct task_struct *p)
1632{
1633 return 0;
1634}
1a687c2e
MG
1635static inline void set_numabalancing_state(bool enabled)
1636{
1637}
82727018
RR
1638static inline void task_numa_free(struct task_struct *p)
1639{
1640}
10f39042
RR
1641static inline bool should_numa_migrate_memory(struct task_struct *p,
1642 struct page *page, int src_nid, int dst_cpu)
1643{
1644 return true;
1645}
cbee9f88
PZ
1646#endif
1647
e868171a 1648static inline struct pid *task_pid(struct task_struct *task)
22c935f4
EB
1649{
1650 return task->pids[PIDTYPE_PID].pid;
1651}
1652
e868171a 1653static inline struct pid *task_tgid(struct task_struct *task)
22c935f4
EB
1654{
1655 return task->group_leader->pids[PIDTYPE_PID].pid;
1656}
1657
6dda81f4
ON
1658/*
1659 * Without tasklist or rcu lock it is not safe to dereference
1660 * the result of task_pgrp/task_session even if task == current,
1661 * we can race with another thread doing sys_setsid/sys_setpgid.
1662 */
e868171a 1663static inline struct pid *task_pgrp(struct task_struct *task)
22c935f4
EB
1664{
1665 return task->group_leader->pids[PIDTYPE_PGID].pid;
1666}
1667
e868171a 1668static inline struct pid *task_session(struct task_struct *task)
22c935f4
EB
1669{
1670 return task->group_leader->pids[PIDTYPE_SID].pid;
1671}
1672
7af57294
PE
1673struct pid_namespace;
1674
1675/*
1676 * the helpers to get the task's different pids as they are seen
1677 * from various namespaces
1678 *
1679 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
44c4e1b2
EB
1680 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1681 * current.
7af57294
PE
1682 * task_xid_nr_ns() : id seen from the ns specified;
1683 *
1684 * set_task_vxid() : assigns a virtual id to a task;
1685 *
7af57294
PE
1686 * see also pid_nr() etc in include/linux/pid.h
1687 */
52ee2dfd
ON
1688pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1689 struct pid_namespace *ns);
7af57294 1690
e868171a 1691static inline pid_t task_pid_nr(struct task_struct *tsk)
7af57294
PE
1692{
1693 return tsk->pid;
1694}
1695
52ee2dfd
ON
1696static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1697 struct pid_namespace *ns)
1698{
1699 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1700}
7af57294
PE
1701
1702static inline pid_t task_pid_vnr(struct task_struct *tsk)
1703{
52ee2dfd 1704 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
7af57294
PE
1705}
1706
1707
e868171a 1708static inline pid_t task_tgid_nr(struct task_struct *tsk)
7af57294
PE
1709{
1710 return tsk->tgid;
1711}
1712
2f2a3a46 1713pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
7af57294
PE
1714
1715static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1716{
1717 return pid_vnr(task_tgid(tsk));
1718}
1719
1720
52ee2dfd
ON
1721static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1722 struct pid_namespace *ns)
7af57294 1723{
52ee2dfd 1724 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
7af57294
PE
1725}
1726
7af57294
PE
1727static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1728{
52ee2dfd 1729 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
7af57294
PE
1730}
1731
1732
52ee2dfd
ON
1733static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1734 struct pid_namespace *ns)
7af57294 1735{
52ee2dfd 1736 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
7af57294
PE
1737}
1738
7af57294
PE
1739static inline pid_t task_session_vnr(struct task_struct *tsk)
1740{
52ee2dfd 1741 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
7af57294
PE
1742}
1743
1b0f7ffd
ON
1744/* obsolete, do not use */
1745static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1746{
1747 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1748}
7af57294 1749
1da177e4
LT
1750/**
1751 * pid_alive - check that a task structure is not stale
1752 * @p: Task structure to be checked.
1753 *
1754 * Test if a process is not yet dead (at most zombie state)
1755 * If pid_alive fails, then pointers within the task structure
1756 * can be stale and must not be dereferenced.
e69f6186
YB
1757 *
1758 * Return: 1 if the process is alive. 0 otherwise.
1da177e4 1759 */
e868171a 1760static inline int pid_alive(struct task_struct *p)
1da177e4 1761{
92476d7f 1762 return p->pids[PIDTYPE_PID].pid != NULL;
1da177e4
LT
1763}
1764
f400e198 1765/**
b460cbc5 1766 * is_global_init - check if a task structure is init
3260259f
H
1767 * @tsk: Task structure to be checked.
1768 *
1769 * Check if a task structure is the first user space task the kernel created.
e69f6186
YB
1770 *
1771 * Return: 1 if the task structure is init. 0 otherwise.
b460cbc5 1772 */
e868171a 1773static inline int is_global_init(struct task_struct *tsk)
b461cc03
PE
1774{
1775 return tsk->pid == 1;
1776}
b460cbc5 1777
9ec52099
CLG
1778extern struct pid *cad_pid;
1779
1da177e4 1780extern void free_task(struct task_struct *tsk);
1da177e4 1781#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
e56d0903 1782
158d9ebd 1783extern void __put_task_struct(struct task_struct *t);
e56d0903
IM
1784
1785static inline void put_task_struct(struct task_struct *t)
1786{
1787 if (atomic_dec_and_test(&t->usage))
8c7904a0 1788 __put_task_struct(t);
e56d0903 1789}
1da177e4 1790
6a61671b
FW
1791#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1792extern void task_cputime(struct task_struct *t,
1793 cputime_t *utime, cputime_t *stime);
1794extern void task_cputime_scaled(struct task_struct *t,
1795 cputime_t *utimescaled, cputime_t *stimescaled);
1796extern cputime_t task_gtime(struct task_struct *t);
1797#else
6fac4829
FW
1798static inline void task_cputime(struct task_struct *t,
1799 cputime_t *utime, cputime_t *stime)
1800{
1801 if (utime)
1802 *utime = t->utime;
1803 if (stime)
1804 *stime = t->stime;
1805}
1806
1807static inline void task_cputime_scaled(struct task_struct *t,
1808 cputime_t *utimescaled,
1809 cputime_t *stimescaled)
1810{
1811 if (utimescaled)
1812 *utimescaled = t->utimescaled;
1813 if (stimescaled)
1814 *stimescaled = t->stimescaled;
1815}
6a61671b
FW
1816
1817static inline cputime_t task_gtime(struct task_struct *t)
1818{
1819 return t->gtime;
1820}
1821#endif
e80d0a1a
FW
1822extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1823extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
49048622 1824
1da177e4
LT
1825/*
1826 * Per process flags
1827 */
1da177e4 1828#define PF_EXITING 0x00000004 /* getting shut down */
778e9a9c 1829#define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
94886b84 1830#define PF_VCPU 0x00000010 /* I'm a virtual CPU */
21aa9af0 1831#define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1da177e4 1832#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
4db96cf0 1833#define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1da177e4
LT
1834#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1835#define PF_DUMPCORE 0x00000200 /* dumped core */
1836#define PF_SIGNALED 0x00000400 /* killed by a signal */
1837#define PF_MEMALLOC 0x00000800 /* Allocating memory */
72fa5997 1838#define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1da177e4 1839#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
774a1221 1840#define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1da177e4
LT
1841#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1842#define PF_FROZEN 0x00010000 /* frozen for system suspend */
1843#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1844#define PF_KSWAPD 0x00040000 /* I am kswapd */
21caf2fc 1845#define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1da177e4 1846#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
246bb0b1 1847#define PF_KTHREAD 0x00200000 /* I am a kernel thread */
b31dc66a
JA
1848#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1849#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1850#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1851#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
14a40ffc 1852#define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
4db96cf0 1853#define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
61a87122 1854#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
58a69cb4 1855#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2b44c4db 1856#define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1da177e4
LT
1857
1858/*
1859 * Only the _current_ task can read/write to tsk->flags, but other
1860 * tasks can access tsk->flags in readonly mode for example
1861 * with tsk_used_math (like during threaded core dumping).
1862 * There is however an exception to this rule during ptrace
1863 * or during fork: the ptracer task is allowed to write to the
1864 * child->flags of its traced child (same goes for fork, the parent
1865 * can write to the child->flags), because we're guaranteed the
1866 * child is not running and in turn not changing child->flags
1867 * at the same time the parent does it.
1868 */
1869#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1870#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1871#define clear_used_math() clear_stopped_child_used_math(current)
1872#define set_used_math() set_stopped_child_used_math(current)
1873#define conditional_stopped_child_used_math(condition, child) \
1874 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1875#define conditional_used_math(condition) \
1876 conditional_stopped_child_used_math(condition, current)
1877#define copy_to_stopped_child_used_math(child) \
1878 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1879/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1880#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1881#define used_math() tsk_used_math(current)
1882
21caf2fc
ML
1883/* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1884static inline gfp_t memalloc_noio_flags(gfp_t flags)
1885{
1886 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1887 flags &= ~__GFP_IO;
1888 return flags;
1889}
1890
1891static inline unsigned int memalloc_noio_save(void)
1892{
1893 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1894 current->flags |= PF_MEMALLOC_NOIO;
1895 return flags;
1896}
1897
1898static inline void memalloc_noio_restore(unsigned int flags)
1899{
1900 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1901}
1902
e5c1902e 1903/*
a8f072c1 1904 * task->jobctl flags
e5c1902e 1905 */
a8f072c1 1906#define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
e5c1902e 1907
a8f072c1
TH
1908#define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1909#define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1910#define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
73ddff2b 1911#define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
fb1d910c 1912#define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
a8f072c1 1913#define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
544b2c91 1914#define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
a8f072c1
TH
1915
1916#define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1917#define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1918#define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
73ddff2b 1919#define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
fb1d910c 1920#define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
a8f072c1 1921#define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
544b2c91 1922#define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
a8f072c1 1923
fb1d910c 1924#define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
73ddff2b 1925#define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
3759a0d9 1926
7dd3db54
TH
1927extern bool task_set_jobctl_pending(struct task_struct *task,
1928 unsigned int mask);
73ddff2b 1929extern void task_clear_jobctl_trapping(struct task_struct *task);
3759a0d9
TH
1930extern void task_clear_jobctl_pending(struct task_struct *task,
1931 unsigned int mask);
39efa3ef 1932
a57eb940 1933#ifdef CONFIG_PREEMPT_RCU
f41d911f
PM
1934
1935#define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1aa03f11 1936#define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
f41d911f
PM
1937
1938static inline void rcu_copy_process(struct task_struct *p)
1939{
1940 p->rcu_read_lock_nesting = 0;
1941 p->rcu_read_unlock_special = 0;
a57eb940 1942#ifdef CONFIG_TREE_PREEMPT_RCU
dd5d19ba 1943 p->rcu_blocked_node = NULL;
24278d14
PM
1944#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1945#ifdef CONFIG_RCU_BOOST
1946 p->rcu_boost_mutex = NULL;
1947#endif /* #ifdef CONFIG_RCU_BOOST */
f41d911f
PM
1948 INIT_LIST_HEAD(&p->rcu_node_entry);
1949}
1950
f41d911f
PM
1951#else
1952
1953static inline void rcu_copy_process(struct task_struct *p)
1954{
1955}
1956
1957#endif
1958
907aed48
MG
1959static inline void tsk_restore_flags(struct task_struct *task,
1960 unsigned long orig_flags, unsigned long flags)
1961{
1962 task->flags &= ~flags;
1963 task->flags |= orig_flags & flags;
1964}
1965
1da177e4 1966#ifdef CONFIG_SMP
1e1b6c51
KM
1967extern void do_set_cpus_allowed(struct task_struct *p,
1968 const struct cpumask *new_mask);
1969
cd8ba7cd 1970extern int set_cpus_allowed_ptr(struct task_struct *p,
96f874e2 1971 const struct cpumask *new_mask);
1da177e4 1972#else
1e1b6c51
KM
1973static inline void do_set_cpus_allowed(struct task_struct *p,
1974 const struct cpumask *new_mask)
1975{
1976}
cd8ba7cd 1977static inline int set_cpus_allowed_ptr(struct task_struct *p,
96f874e2 1978 const struct cpumask *new_mask)
1da177e4 1979{
96f874e2 1980 if (!cpumask_test_cpu(0, new_mask))
1da177e4
LT
1981 return -EINVAL;
1982 return 0;
1983}
1984#endif
e0ad9556 1985
3451d024 1986#ifdef CONFIG_NO_HZ_COMMON
5167e8d5
PZ
1987void calc_load_enter_idle(void);
1988void calc_load_exit_idle(void);
1989#else
1990static inline void calc_load_enter_idle(void) { }
1991static inline void calc_load_exit_idle(void) { }
3451d024 1992#endif /* CONFIG_NO_HZ_COMMON */
5167e8d5 1993
e0ad9556 1994#ifndef CONFIG_CPUMASK_OFFSTACK
cd8ba7cd
MT
1995static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1996{
1997 return set_cpus_allowed_ptr(p, &new_mask);
1998}
e0ad9556 1999#endif
1da177e4 2000
b342501c 2001/*
c676329a
PZ
2002 * Do not use outside of architecture code which knows its limitations.
2003 *
2004 * sched_clock() has no promise of monotonicity or bounded drift between
2005 * CPUs, use (which you should not) requires disabling IRQs.
2006 *
2007 * Please use one of the three interfaces below.
b342501c 2008 */
1bbfa6f2 2009extern unsigned long long notrace sched_clock(void);
c676329a 2010/*
489a71b0 2011 * See the comment in kernel/sched/clock.c
c676329a
PZ
2012 */
2013extern u64 cpu_clock(int cpu);
2014extern u64 local_clock(void);
2015extern u64 sched_clock_cpu(int cpu);
2016
e436d800 2017
c1955a3d 2018extern void sched_clock_init(void);
3e51f33f 2019
c1955a3d 2020#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
3e51f33f
PZ
2021static inline void sched_clock_tick(void)
2022{
2023}
2024
2025static inline void sched_clock_idle_sleep_event(void)
2026{
2027}
2028
2029static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2030{
2031}
2032#else
c676329a
PZ
2033/*
2034 * Architectures can set this to 1 if they have specified
2035 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2036 * but then during bootup it turns out that sched_clock()
2037 * is reliable after all:
2038 */
35af99e6
PZ
2039extern int sched_clock_stable(void);
2040extern void set_sched_clock_stable(void);
2041extern void clear_sched_clock_stable(void);
c676329a 2042
3e51f33f
PZ
2043extern void sched_clock_tick(void);
2044extern void sched_clock_idle_sleep_event(void);
2045extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2046#endif
2047
b52bfee4
VP
2048#ifdef CONFIG_IRQ_TIME_ACCOUNTING
2049/*
2050 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2051 * The reason for this explicit opt-in is not to have perf penalty with
2052 * slow sched_clocks.
2053 */
2054extern void enable_sched_clock_irqtime(void);
2055extern void disable_sched_clock_irqtime(void);
2056#else
2057static inline void enable_sched_clock_irqtime(void) {}
2058static inline void disable_sched_clock_irqtime(void) {}
2059#endif
2060
36c8b586 2061extern unsigned long long
41b86e9c 2062task_sched_runtime(struct task_struct *task);
1da177e4
LT
2063
2064/* sched_exec is called by processes performing an exec */
2065#ifdef CONFIG_SMP
2066extern void sched_exec(void);
2067#else
2068#define sched_exec() {}
2069#endif
2070
2aa44d05
IM
2071extern void sched_clock_idle_sleep_event(void);
2072extern void sched_clock_idle_wakeup_event(u64 delta_ns);
bb29ab26 2073
1da177e4
LT
2074#ifdef CONFIG_HOTPLUG_CPU
2075extern void idle_task_exit(void);
2076#else
2077static inline void idle_task_exit(void) {}
2078#endif
2079
3451d024 2080#if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
1c20091e 2081extern void wake_up_nohz_cpu(int cpu);
06d8308c 2082#else
1c20091e 2083static inline void wake_up_nohz_cpu(int cpu) { }
06d8308c
TG
2084#endif
2085
ce831b38
FW
2086#ifdef CONFIG_NO_HZ_FULL
2087extern bool sched_can_stop_tick(void);
265f22a9 2088extern u64 scheduler_tick_max_deferment(void);
ce831b38
FW
2089#else
2090static inline bool sched_can_stop_tick(void) { return false; }
06d8308c
TG
2091#endif
2092
5091faa4 2093#ifdef CONFIG_SCHED_AUTOGROUP
5091faa4
MG
2094extern void sched_autogroup_create_attach(struct task_struct *p);
2095extern void sched_autogroup_detach(struct task_struct *p);
2096extern void sched_autogroup_fork(struct signal_struct *sig);
2097extern void sched_autogroup_exit(struct signal_struct *sig);
2098#ifdef CONFIG_PROC_FS
2099extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2e5b5b3a 2100extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
5091faa4
MG
2101#endif
2102#else
2103static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2104static inline void sched_autogroup_detach(struct task_struct *p) { }
2105static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2106static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2107#endif
2108
d95f4122 2109extern bool yield_to(struct task_struct *p, bool preempt);
36c8b586
IM
2110extern void set_user_nice(struct task_struct *p, long nice);
2111extern int task_prio(const struct task_struct *p);
d0ea0268
DY
2112/**
2113 * task_nice - return the nice value of a given task.
2114 * @p: the task in question.
2115 *
2116 * Return: The nice value [ -20 ... 0 ... 19 ].
2117 */
2118static inline int task_nice(const struct task_struct *p)
2119{
2120 return PRIO_TO_NICE((p)->static_prio);
2121}
36c8b586
IM
2122extern int can_nice(const struct task_struct *p, const int nice);
2123extern int task_curr(const struct task_struct *p);
1da177e4 2124extern int idle_cpu(int cpu);
fe7de49f
KM
2125extern int sched_setscheduler(struct task_struct *, int,
2126 const struct sched_param *);
961ccddd 2127extern int sched_setscheduler_nocheck(struct task_struct *, int,
fe7de49f 2128 const struct sched_param *);
d50dde5a
DF
2129extern int sched_setattr(struct task_struct *,
2130 const struct sched_attr *);
36c8b586 2131extern struct task_struct *idle_task(int cpu);
c4f30608
PM
2132/**
2133 * is_idle_task - is the specified task an idle task?
fa757281 2134 * @p: the task in question.
e69f6186
YB
2135 *
2136 * Return: 1 if @p is an idle task. 0 otherwise.
c4f30608 2137 */
7061ca3b 2138static inline bool is_idle_task(const struct task_struct *p)
c4f30608
PM
2139{
2140 return p->pid == 0;
2141}
36c8b586
IM
2142extern struct task_struct *curr_task(int cpu);
2143extern void set_curr_task(int cpu, struct task_struct *p);
1da177e4
LT
2144
2145void yield(void);
2146
2147/*
2148 * The default (Linux) execution domain.
2149 */
2150extern struct exec_domain default_exec_domain;
2151
2152union thread_union {
2153 struct thread_info thread_info;
2154 unsigned long stack[THREAD_SIZE/sizeof(long)];
2155};
2156
2157#ifndef __HAVE_ARCH_KSTACK_END
2158static inline int kstack_end(void *addr)
2159{
2160 /* Reliable end of stack detection:
2161 * Some APM bios versions misalign the stack
2162 */
2163 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2164}
2165#endif
2166
2167extern union thread_union init_thread_union;
2168extern struct task_struct init_task;
2169
2170extern struct mm_struct init_mm;
2171
198fe21b
PE
2172extern struct pid_namespace init_pid_ns;
2173
2174/*
2175 * find a task by one of its numerical ids
2176 *
198fe21b
PE
2177 * find_task_by_pid_ns():
2178 * finds a task by its pid in the specified namespace
228ebcbe
PE
2179 * find_task_by_vpid():
2180 * finds a task by its virtual pid
198fe21b 2181 *
e49859e7 2182 * see also find_vpid() etc in include/linux/pid.h
198fe21b
PE
2183 */
2184
228ebcbe
PE
2185extern struct task_struct *find_task_by_vpid(pid_t nr);
2186extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2187 struct pid_namespace *ns);
198fe21b 2188
1da177e4 2189/* per-UID process charging. */
7b44ab97 2190extern struct user_struct * alloc_uid(kuid_t);
1da177e4
LT
2191static inline struct user_struct *get_uid(struct user_struct *u)
2192{
2193 atomic_inc(&u->__count);
2194 return u;
2195}
2196extern void free_uid(struct user_struct *);
1da177e4
LT
2197
2198#include <asm/current.h>
2199
f0af911a 2200extern void xtime_update(unsigned long ticks);
1da177e4 2201
b3c97528
HH
2202extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2203extern int wake_up_process(struct task_struct *tsk);
3e51e3ed 2204extern void wake_up_new_task(struct task_struct *tsk);
1da177e4
LT
2205#ifdef CONFIG_SMP
2206 extern void kick_process(struct task_struct *tsk);
2207#else
2208 static inline void kick_process(struct task_struct *tsk) { }
2209#endif
aab03e05 2210extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
ad46c2c4 2211extern void sched_dead(struct task_struct *p);
1da177e4 2212
1da177e4
LT
2213extern void proc_caches_init(void);
2214extern void flush_signals(struct task_struct *);
3bcac026 2215extern void __flush_signals(struct task_struct *);
10ab825b 2216extern void ignore_signals(struct task_struct *);
1da177e4
LT
2217extern void flush_signal_handlers(struct task_struct *, int force_default);
2218extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2219
2220static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2221{
2222 unsigned long flags;
2223 int ret;
2224
2225 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2226 ret = dequeue_signal(tsk, mask, info);
2227 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2228
2229 return ret;
53c8f9f1 2230}
1da177e4
LT
2231
2232extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2233 sigset_t *mask);
2234extern void unblock_all_signals(void);
2235extern void release_task(struct task_struct * p);
2236extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1da177e4
LT
2237extern int force_sigsegv(int, struct task_struct *);
2238extern int force_sig_info(int, struct siginfo *, struct task_struct *);
c4b92fc1 2239extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
c4b92fc1 2240extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
d178bc3a
SH
2241extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2242 const struct cred *, u32);
c4b92fc1
EB
2243extern int kill_pgrp(struct pid *pid, int sig, int priv);
2244extern int kill_pid(struct pid *pid, int sig, int priv);
c3de4b38 2245extern int kill_proc_info(int, struct siginfo *, pid_t);
86773473 2246extern __must_check bool do_notify_parent(struct task_struct *, int);
a7f0765e 2247extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
1da177e4 2248extern void force_sig(int, struct task_struct *);
1da177e4 2249extern int send_sig(int, struct task_struct *, int);
09faef11 2250extern int zap_other_threads(struct task_struct *p);
1da177e4
LT
2251extern struct sigqueue *sigqueue_alloc(void);
2252extern void sigqueue_free(struct sigqueue *);
ac5c2153 2253extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
9ac95f2f 2254extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1da177e4 2255
51a7b448
AV
2256static inline void restore_saved_sigmask(void)
2257{
2258 if (test_and_clear_restore_sigmask())
77097ae5 2259 __set_current_blocked(&current->saved_sigmask);
51a7b448
AV
2260}
2261
b7f9a11a
AV
2262static inline sigset_t *sigmask_to_save(void)
2263{
2264 sigset_t *res = &current->blocked;
2265 if (unlikely(test_restore_sigmask()))
2266 res = &current->saved_sigmask;
2267 return res;
2268}
2269
9ec52099
CLG
2270static inline int kill_cad_pid(int sig, int priv)
2271{
2272 return kill_pid(cad_pid, sig, priv);
2273}
2274
1da177e4
LT
2275/* These can be the second arg to send_sig_info/send_group_sig_info. */
2276#define SEND_SIG_NOINFO ((struct siginfo *) 0)
2277#define SEND_SIG_PRIV ((struct siginfo *) 1)
2278#define SEND_SIG_FORCED ((struct siginfo *) 2)
2279
2a855dd0
SAS
2280/*
2281 * True if we are on the alternate signal stack.
2282 */
1da177e4
LT
2283static inline int on_sig_stack(unsigned long sp)
2284{
2a855dd0
SAS
2285#ifdef CONFIG_STACK_GROWSUP
2286 return sp >= current->sas_ss_sp &&
2287 sp - current->sas_ss_sp < current->sas_ss_size;
2288#else
2289 return sp > current->sas_ss_sp &&
2290 sp - current->sas_ss_sp <= current->sas_ss_size;
2291#endif
1da177e4
LT
2292}
2293
2294static inline int sas_ss_flags(unsigned long sp)
2295{
2296 return (current->sas_ss_size == 0 ? SS_DISABLE
2297 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2298}
2299
5a1b98d3
AV
2300static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2301{
2302 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2303#ifdef CONFIG_STACK_GROWSUP
2304 return current->sas_ss_sp;
2305#else
2306 return current->sas_ss_sp + current->sas_ss_size;
2307#endif
2308 return sp;
2309}
2310
1da177e4
LT
2311/*
2312 * Routines for handling mm_structs
2313 */
2314extern struct mm_struct * mm_alloc(void);
2315
2316/* mmdrop drops the mm and the page tables */
b3c97528 2317extern void __mmdrop(struct mm_struct *);
1da177e4
LT
2318static inline void mmdrop(struct mm_struct * mm)
2319{
6fb43d7b 2320 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1da177e4
LT
2321 __mmdrop(mm);
2322}
2323
2324/* mmput gets rid of the mappings and all user-space */
2325extern void mmput(struct mm_struct *);
2326/* Grab a reference to a task's mm, if it is not already going away */
2327extern struct mm_struct *get_task_mm(struct task_struct *task);
8cdb878d
CY
2328/*
2329 * Grab a reference to a task's mm, if it is not already going away
2330 * and ptrace_may_access with the mode parameter passed to it
2331 * succeeds.
2332 */
2333extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
1da177e4
LT
2334/* Remove the current tasks stale references to the old mm_struct */
2335extern void mm_release(struct task_struct *, struct mm_struct *);
2336
6f2c55b8 2337extern int copy_thread(unsigned long, unsigned long, unsigned long,
afa86fc4 2338 struct task_struct *);
1da177e4
LT
2339extern void flush_thread(void);
2340extern void exit_thread(void);
2341
1da177e4 2342extern void exit_files(struct task_struct *);
a7e5328a 2343extern void __cleanup_sighand(struct sighand_struct *);
cbaffba1 2344
1da177e4 2345extern void exit_itimers(struct signal_struct *);
cbaffba1 2346extern void flush_itimer_signals(void);
1da177e4 2347
9402c95f 2348extern void do_group_exit(int);
1da177e4 2349
1da177e4
LT
2350extern int allow_signal(int);
2351extern int disallow_signal(int);
1da177e4 2352
c4ad8f98 2353extern int do_execve(struct filename *,
d7627467 2354 const char __user * const __user *,
da3d4c5f 2355 const char __user * const __user *);
e80d6661 2356extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
36c8b586 2357struct task_struct *fork_idle(int);
2aa3a7f8 2358extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
1da177e4 2359
23aebe16 2360extern void set_task_comm(struct task_struct *tsk, const char *from);
59714d65 2361extern char *get_task_comm(char *to, struct task_struct *tsk);
1da177e4
LT
2362
2363#ifdef CONFIG_SMP
317f3941 2364void scheduler_ipi(void);
85ba2d86 2365extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1da177e4 2366#else
184748cc 2367static inline void scheduler_ipi(void) { }
85ba2d86
RM
2368static inline unsigned long wait_task_inactive(struct task_struct *p,
2369 long match_state)
2370{
2371 return 1;
2372}
1da177e4
LT
2373#endif
2374
05725f7e
JP
2375#define next_task(p) \
2376 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
1da177e4
LT
2377
2378#define for_each_process(p) \
2379 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2380
5bb459bb 2381extern bool current_is_single_threaded(void);
d84f4f99 2382
1da177e4
LT
2383/*
2384 * Careful: do_each_thread/while_each_thread is a double loop so
2385 * 'break' will not work as expected - use goto instead.
2386 */
2387#define do_each_thread(g, t) \
2388 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2389
2390#define while_each_thread(g, t) \
2391 while ((t = next_thread(t)) != g)
2392
0c740d0a
ON
2393#define __for_each_thread(signal, t) \
2394 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2395
2396#define for_each_thread(p, t) \
2397 __for_each_thread((p)->signal, t)
2398
2399/* Careful: this is a double loop, 'break' won't work as expected. */
2400#define for_each_process_thread(p, t) \
2401 for_each_process(p) for_each_thread(p, t)
2402
7e49827c
ON
2403static inline int get_nr_threads(struct task_struct *tsk)
2404{
b3ac022c 2405 return tsk->signal->nr_threads;
7e49827c
ON
2406}
2407
087806b1
ON
2408static inline bool thread_group_leader(struct task_struct *p)
2409{
2410 return p->exit_signal >= 0;
2411}
1da177e4 2412
0804ef4b
EB
2413/* Do to the insanities of de_thread it is possible for a process
2414 * to have the pid of the thread group leader without actually being
2415 * the thread group leader. For iteration through the pids in proc
2416 * all we care about is that we have a task with the appropriate
2417 * pid, we don't actually care if we have the right task.
2418 */
e1403b8e 2419static inline bool has_group_leader_pid(struct task_struct *p)
0804ef4b 2420{
e1403b8e 2421 return task_pid(p) == p->signal->leader_pid;
0804ef4b
EB
2422}
2423
bac0abd6 2424static inline
e1403b8e 2425bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
bac0abd6 2426{
e1403b8e 2427 return p1->signal == p2->signal;
bac0abd6
PE
2428}
2429
36c8b586 2430static inline struct task_struct *next_thread(const struct task_struct *p)
47e65328 2431{
05725f7e
JP
2432 return list_entry_rcu(p->thread_group.next,
2433 struct task_struct, thread_group);
47e65328
ON
2434}
2435
e868171a 2436static inline int thread_group_empty(struct task_struct *p)
1da177e4 2437{
47e65328 2438 return list_empty(&p->thread_group);
1da177e4
LT
2439}
2440
2441#define delay_group_leader(p) \
2442 (thread_group_leader(p) && !thread_group_empty(p))
2443
1da177e4 2444/*
260ea101 2445 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
22e2c507 2446 * subscriptions and synchronises with wait4(). Also used in procfs. Also
ddbcc7e8 2447 * pins the final release of task.io_context. Also protects ->cpuset and
d68b46fe 2448 * ->cgroup.subsys[]. And ->vfork_done.
1da177e4
LT
2449 *
2450 * Nests both inside and outside of read_lock(&tasklist_lock).
2451 * It must not be nested with write_lock_irq(&tasklist_lock),
2452 * neither inside nor outside.
2453 */
2454static inline void task_lock(struct task_struct *p)
2455{
2456 spin_lock(&p->alloc_lock);
2457}
2458
2459static inline void task_unlock(struct task_struct *p)
2460{
2461 spin_unlock(&p->alloc_lock);
2462}
2463
b8ed374e 2464extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
f63ee72e
ON
2465 unsigned long *flags);
2466
9388dc30
AV
2467static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2468 unsigned long *flags)
2469{
2470 struct sighand_struct *ret;
2471
2472 ret = __lock_task_sighand(tsk, flags);
2473 (void)__cond_lock(&tsk->sighand->siglock, ret);
2474 return ret;
2475}
b8ed374e 2476
f63ee72e
ON
2477static inline void unlock_task_sighand(struct task_struct *tsk,
2478 unsigned long *flags)
2479{
2480 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2481}
2482
4714d1d3 2483#ifdef CONFIG_CGROUPS
257058ae 2484static inline void threadgroup_change_begin(struct task_struct *tsk)
4714d1d3 2485{
257058ae 2486 down_read(&tsk->signal->group_rwsem);
4714d1d3 2487}
257058ae 2488static inline void threadgroup_change_end(struct task_struct *tsk)
4714d1d3 2489{
257058ae 2490 up_read(&tsk->signal->group_rwsem);
4714d1d3 2491}
77e4ef99
TH
2492
2493/**
2494 * threadgroup_lock - lock threadgroup
2495 * @tsk: member task of the threadgroup to lock
2496 *
2497 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2498 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
e56fb287
ON
2499 * change ->group_leader/pid. This is useful for cases where the threadgroup
2500 * needs to stay stable across blockable operations.
77e4ef99
TH
2501 *
2502 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2503 * synchronization. While held, no new task will be added to threadgroup
2504 * and no existing live task will have its PF_EXITING set.
2505 *
e56fb287
ON
2506 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2507 * sub-thread becomes a new leader.
77e4ef99 2508 */
257058ae 2509static inline void threadgroup_lock(struct task_struct *tsk)
4714d1d3 2510{
257058ae 2511 down_write(&tsk->signal->group_rwsem);
4714d1d3 2512}
77e4ef99
TH
2513
2514/**
2515 * threadgroup_unlock - unlock threadgroup
2516 * @tsk: member task of the threadgroup to unlock
2517 *
2518 * Reverse threadgroup_lock().
2519 */
257058ae 2520static inline void threadgroup_unlock(struct task_struct *tsk)
4714d1d3 2521{
257058ae 2522 up_write(&tsk->signal->group_rwsem);
4714d1d3
BB
2523}
2524#else
257058ae
TH
2525static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2526static inline void threadgroup_change_end(struct task_struct *tsk) {}
2527static inline void threadgroup_lock(struct task_struct *tsk) {}
2528static inline void threadgroup_unlock(struct task_struct *tsk) {}
4714d1d3
BB
2529#endif
2530
f037360f
AV
2531#ifndef __HAVE_THREAD_FUNCTIONS
2532
f7e4217b
RZ
2533#define task_thread_info(task) ((struct thread_info *)(task)->stack)
2534#define task_stack_page(task) ((task)->stack)
a1261f54 2535
10ebffde
AV
2536static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2537{
2538 *task_thread_info(p) = *task_thread_info(org);
2539 task_thread_info(p)->task = p;
2540}
2541
2542static inline unsigned long *end_of_stack(struct task_struct *p)
2543{
f7e4217b 2544 return (unsigned long *)(task_thread_info(p) + 1);
10ebffde
AV
2545}
2546
f037360f
AV
2547#endif
2548
8b05c7e6
FT
2549static inline int object_is_on_stack(void *obj)
2550{
2551 void *stack = task_stack_page(current);
2552
2553 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2554}
2555
8c9843e5
BH
2556extern void thread_info_cache_init(void);
2557
7c9f8861
ES
2558#ifdef CONFIG_DEBUG_STACK_USAGE
2559static inline unsigned long stack_not_used(struct task_struct *p)
2560{
2561 unsigned long *n = end_of_stack(p);
2562
2563 do { /* Skip over canary */
2564 n++;
2565 } while (!*n);
2566
2567 return (unsigned long)n - (unsigned long)end_of_stack(p);
2568}
2569#endif
2570
1da177e4
LT
2571/* set thread flags in other task's structures
2572 * - see asm/thread_info.h for TIF_xxxx flags available
2573 */
2574static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2575{
a1261f54 2576 set_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
2577}
2578
2579static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2580{
a1261f54 2581 clear_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
2582}
2583
2584static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2585{
a1261f54 2586 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
2587}
2588
2589static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2590{
a1261f54 2591 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
2592}
2593
2594static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2595{
a1261f54 2596 return test_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
2597}
2598
2599static inline void set_tsk_need_resched(struct task_struct *tsk)
2600{
2601 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2602}
2603
2604static inline void clear_tsk_need_resched(struct task_struct *tsk)
2605{
2606 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2607}
2608
8ae121ac
GH
2609static inline int test_tsk_need_resched(struct task_struct *tsk)
2610{
2611 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2612}
2613
690cc3ff
EB
2614static inline int restart_syscall(void)
2615{
2616 set_tsk_thread_flag(current, TIF_SIGPENDING);
2617 return -ERESTARTNOINTR;
2618}
2619
1da177e4
LT
2620static inline int signal_pending(struct task_struct *p)
2621{
2622 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2623}
f776d12d 2624
d9588725
RM
2625static inline int __fatal_signal_pending(struct task_struct *p)
2626{
2627 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2628}
f776d12d
MW
2629
2630static inline int fatal_signal_pending(struct task_struct *p)
2631{
2632 return signal_pending(p) && __fatal_signal_pending(p);
2633}
2634
16882c1e
ON
2635static inline int signal_pending_state(long state, struct task_struct *p)
2636{
2637 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2638 return 0;
2639 if (!signal_pending(p))
2640 return 0;
2641
16882c1e
ON
2642 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2643}
2644
1da177e4
LT
2645/*
2646 * cond_resched() and cond_resched_lock(): latency reduction via
2647 * explicit rescheduling in places that are safe. The return
2648 * value indicates whether a reschedule was done in fact.
2649 * cond_resched_lock() will drop the spinlock before scheduling,
2650 * cond_resched_softirq() will enable bhs before scheduling.
2651 */
c3921ab7 2652extern int _cond_resched(void);
6f80bd98 2653
613afbf8
FW
2654#define cond_resched() ({ \
2655 __might_sleep(__FILE__, __LINE__, 0); \
2656 _cond_resched(); \
2657})
6f80bd98 2658
613afbf8
FW
2659extern int __cond_resched_lock(spinlock_t *lock);
2660
bdd4e85d 2661#ifdef CONFIG_PREEMPT_COUNT
716a4234 2662#define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
02b67cc3 2663#else
716a4234 2664#define PREEMPT_LOCK_OFFSET 0
02b67cc3 2665#endif
716a4234 2666
613afbf8 2667#define cond_resched_lock(lock) ({ \
716a4234 2668 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
613afbf8
FW
2669 __cond_resched_lock(lock); \
2670})
2671
2672extern int __cond_resched_softirq(void);
2673
75e1056f
VP
2674#define cond_resched_softirq() ({ \
2675 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2676 __cond_resched_softirq(); \
613afbf8 2677})
1da177e4 2678
f6f3c437
SH
2679static inline void cond_resched_rcu(void)
2680{
2681#if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2682 rcu_read_unlock();
2683 cond_resched();
2684 rcu_read_lock();
2685#endif
2686}
2687
1da177e4
LT
2688/*
2689 * Does a critical section need to be broken due to another
95c354fe
NP
2690 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2691 * but a general need for low latency)
1da177e4 2692 */
95c354fe 2693static inline int spin_needbreak(spinlock_t *lock)
1da177e4 2694{
95c354fe
NP
2695#ifdef CONFIG_PREEMPT
2696 return spin_is_contended(lock);
2697#else
1da177e4 2698 return 0;
95c354fe 2699#endif
1da177e4
LT
2700}
2701
ee761f62
TG
2702/*
2703 * Idle thread specific functions to determine the need_resched
2704 * polling state. We have two versions, one based on TS_POLLING in
2705 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2706 * thread_info.flags
2707 */
2708#ifdef TS_POLLING
2709static inline int tsk_is_polling(struct task_struct *p)
2710{
2711 return task_thread_info(p)->status & TS_POLLING;
2712}
ea811747 2713static inline void __current_set_polling(void)
3a98f871
TG
2714{
2715 current_thread_info()->status |= TS_POLLING;
2716}
2717
ea811747
PZ
2718static inline bool __must_check current_set_polling_and_test(void)
2719{
2720 __current_set_polling();
2721
2722 /*
2723 * Polling state must be visible before we test NEED_RESCHED,
2724 * paired by resched_task()
2725 */
2726 smp_mb();
2727
2728 return unlikely(tif_need_resched());
2729}
2730
2731static inline void __current_clr_polling(void)
3a98f871
TG
2732{
2733 current_thread_info()->status &= ~TS_POLLING;
ea811747
PZ
2734}
2735
2736static inline bool __must_check current_clr_polling_and_test(void)
2737{
2738 __current_clr_polling();
2739
2740 /*
2741 * Polling state must be visible before we test NEED_RESCHED,
2742 * paired by resched_task()
2743 */
2744 smp_mb();
2745
2746 return unlikely(tif_need_resched());
3a98f871 2747}
ee761f62
TG
2748#elif defined(TIF_POLLING_NRFLAG)
2749static inline int tsk_is_polling(struct task_struct *p)
2750{
2751 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2752}
ea811747
PZ
2753
2754static inline void __current_set_polling(void)
3a98f871
TG
2755{
2756 set_thread_flag(TIF_POLLING_NRFLAG);
2757}
2758
ea811747
PZ
2759static inline bool __must_check current_set_polling_and_test(void)
2760{
2761 __current_set_polling();
2762
2763 /*
2764 * Polling state must be visible before we test NEED_RESCHED,
2765 * paired by resched_task()
2766 *
2767 * XXX: assumes set/clear bit are identical barrier wise.
2768 */
2769 smp_mb__after_clear_bit();
2770
2771 return unlikely(tif_need_resched());
2772}
2773
2774static inline void __current_clr_polling(void)
3a98f871
TG
2775{
2776 clear_thread_flag(TIF_POLLING_NRFLAG);
2777}
ea811747
PZ
2778
2779static inline bool __must_check current_clr_polling_and_test(void)
2780{
2781 __current_clr_polling();
2782
2783 /*
2784 * Polling state must be visible before we test NEED_RESCHED,
2785 * paired by resched_task()
2786 */
2787 smp_mb__after_clear_bit();
2788
2789 return unlikely(tif_need_resched());
2790}
2791
ee761f62
TG
2792#else
2793static inline int tsk_is_polling(struct task_struct *p) { return 0; }
ea811747
PZ
2794static inline void __current_set_polling(void) { }
2795static inline void __current_clr_polling(void) { }
2796
2797static inline bool __must_check current_set_polling_and_test(void)
2798{
2799 return unlikely(tif_need_resched());
2800}
2801static inline bool __must_check current_clr_polling_and_test(void)
2802{
2803 return unlikely(tif_need_resched());
2804}
ee761f62
TG
2805#endif
2806
8cb75e0c
PZ
2807static inline void current_clr_polling(void)
2808{
2809 __current_clr_polling();
2810
2811 /*
2812 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2813 * Once the bit is cleared, we'll get IPIs with every new
2814 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2815 * fold.
2816 */
2817 smp_mb(); /* paired with resched_task() */
2818
2819 preempt_fold_need_resched();
2820}
2821
75f93fed
PZ
2822static __always_inline bool need_resched(void)
2823{
2824 return unlikely(tif_need_resched());
2825}
2826
f06febc9
FM
2827/*
2828 * Thread group CPU time accounting.
2829 */
4cd4c1b4 2830void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
4da94d49 2831void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
f06febc9 2832
490dea45 2833static inline void thread_group_cputime_init(struct signal_struct *sig)
f06febc9 2834{
ee30a7b2 2835 raw_spin_lock_init(&sig->cputimer.lock);
f06febc9
FM
2836}
2837
7bb44ade
RM
2838/*
2839 * Reevaluate whether the task has signals pending delivery.
2840 * Wake the task if so.
2841 * This is required every time the blocked sigset_t changes.
2842 * callers must hold sighand->siglock.
2843 */
2844extern void recalc_sigpending_and_wake(struct task_struct *t);
1da177e4
LT
2845extern void recalc_sigpending(void);
2846
910ffdb1
ON
2847extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2848
2849static inline void signal_wake_up(struct task_struct *t, bool resume)
2850{
2851 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2852}
2853static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2854{
2855 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2856}
1da177e4
LT
2857
2858/*
2859 * Wrappers for p->thread_info->cpu access. No-op on UP.
2860 */
2861#ifdef CONFIG_SMP
2862
2863static inline unsigned int task_cpu(const struct task_struct *p)
2864{
a1261f54 2865 return task_thread_info(p)->cpu;
1da177e4
LT
2866}
2867
b32e86b4
IM
2868static inline int task_node(const struct task_struct *p)
2869{
2870 return cpu_to_node(task_cpu(p));
2871}
2872
c65cc870 2873extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
1da177e4
LT
2874
2875#else
2876
2877static inline unsigned int task_cpu(const struct task_struct *p)
2878{
2879 return 0;
2880}
2881
2882static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2883{
2884}
2885
2886#endif /* CONFIG_SMP */
2887
96f874e2
RR
2888extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2889extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
5c45bf27 2890
7c941438 2891#ifdef CONFIG_CGROUP_SCHED
07e06b01 2892extern struct task_group root_task_group;
8323f26c 2893#endif /* CONFIG_CGROUP_SCHED */
9b5b7751 2894
54e99124
DG
2895extern int task_can_switch_user(struct user_struct *up,
2896 struct task_struct *tsk);
2897
4b98d11b
AD
2898#ifdef CONFIG_TASK_XACCT
2899static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2900{
940389b8 2901 tsk->ioac.rchar += amt;
4b98d11b
AD
2902}
2903
2904static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2905{
940389b8 2906 tsk->ioac.wchar += amt;
4b98d11b
AD
2907}
2908
2909static inline void inc_syscr(struct task_struct *tsk)
2910{
940389b8 2911 tsk->ioac.syscr++;
4b98d11b
AD
2912}
2913
2914static inline void inc_syscw(struct task_struct *tsk)
2915{
940389b8 2916 tsk->ioac.syscw++;
4b98d11b
AD
2917}
2918#else
2919static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2920{
2921}
2922
2923static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2924{
2925}
2926
2927static inline void inc_syscr(struct task_struct *tsk)
2928{
2929}
2930
2931static inline void inc_syscw(struct task_struct *tsk)
2932{
2933}
2934#endif
2935
82455257
DH
2936#ifndef TASK_SIZE_OF
2937#define TASK_SIZE_OF(tsk) TASK_SIZE
2938#endif
2939
cf475ad2
BS
2940#ifdef CONFIG_MM_OWNER
2941extern void mm_update_next_owner(struct mm_struct *mm);
2942extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2943#else
2944static inline void mm_update_next_owner(struct mm_struct *mm)
2945{
2946}
2947
2948static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2949{
2950}
2951#endif /* CONFIG_MM_OWNER */
2952
3e10e716
JS
2953static inline unsigned long task_rlimit(const struct task_struct *tsk,
2954 unsigned int limit)
2955{
2956 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2957}
2958
2959static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2960 unsigned int limit)
2961{
2962 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2963}
2964
2965static inline unsigned long rlimit(unsigned int limit)
2966{
2967 return task_rlimit(current, limit);
2968}
2969
2970static inline unsigned long rlimit_max(unsigned int limit)
2971{
2972 return task_rlimit_max(current, limit);
2973}
2974
1da177e4 2975#endif
This page took 1.694071 seconds and 5 git commands to generate.