sched: Remove remaining dubious usage of "power"
[linux-2.6-block.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 212/* in tsk->exit_state */
ad86622b
ON
213#define EXIT_DEAD 16
214#define EXIT_ZOMBIE 32
abd50b39 215#define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
4a8342d2 216/* in tsk->state again */
af927232 217#define TASK_DEAD 64
f021a3c2 218#define TASK_WAKEKILL 128
e9c84311 219#define TASK_WAKING 256
f2530dc7
TG
220#define TASK_PARKED 512
221#define TASK_STATE_MAX 1024
f021a3c2 222
f2530dc7 223#define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
73342151 224
e1781538
PZ
225extern char ___assert_task_state[1 - 2*!!(
226 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
f021a3c2
MW
227
228/* Convenience macros for the sake of set_task_state */
229#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
230#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
231#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
1da177e4 232
92a1f4bc
MW
233/* Convenience macros for the sake of wake_up */
234#define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
f021a3c2 235#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
92a1f4bc
MW
236
237/* get_task_state() */
238#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
f021a3c2 239 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
74e37200 240 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
92a1f4bc 241
f021a3c2
MW
242#define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
243#define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
92a1f4bc 244#define task_is_stopped_or_traced(task) \
f021a3c2 245 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
92a1f4bc 246#define task_contributes_to_load(task) \
e3c8ca83 247 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
376fede8 248 (task->flags & PF_FROZEN) == 0)
1da177e4
LT
249
250#define __set_task_state(tsk, state_value) \
251 do { (tsk)->state = (state_value); } while (0)
252#define set_task_state(tsk, state_value) \
253 set_mb((tsk)->state, (state_value))
254
498d0c57
AM
255/*
256 * set_current_state() includes a barrier so that the write of current->state
257 * is correctly serialised wrt the caller's subsequent test of whether to
258 * actually sleep:
259 *
260 * set_current_state(TASK_UNINTERRUPTIBLE);
261 * if (do_i_need_to_sleep())
262 * schedule();
263 *
264 * If the caller does not need such serialisation then use __set_current_state()
265 */
1da177e4
LT
266#define __set_current_state(state_value) \
267 do { current->state = (state_value); } while (0)
268#define set_current_state(state_value) \
269 set_mb(current->state, (state_value))
270
271/* Task command name length */
272#define TASK_COMM_LEN 16
273
1da177e4
LT
274#include <linux/spinlock.h>
275
276/*
277 * This serializes "schedule()" and also protects
278 * the run-queue from deletions/modifications (but
279 * _adding_ to the beginning of the run-queue has
280 * a separate lock).
281 */
282extern rwlock_t tasklist_lock;
283extern spinlock_t mmlist_lock;
284
36c8b586 285struct task_struct;
1da177e4 286
db1466b3
PM
287#ifdef CONFIG_PROVE_RCU
288extern int lockdep_tasklist_lock_is_held(void);
289#endif /* #ifdef CONFIG_PROVE_RCU */
290
1da177e4
LT
291extern void sched_init(void);
292extern void sched_init_smp(void);
2d07b255 293extern asmlinkage void schedule_tail(struct task_struct *prev);
36c8b586 294extern void init_idle(struct task_struct *idle, int cpu);
1df21055 295extern void init_idle_bootup_task(struct task_struct *idle);
1da177e4 296
89f19f04 297extern int runqueue_is_locked(int cpu);
017730c1 298
3451d024 299#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
c1cc017c 300extern void nohz_balance_enter_idle(int cpu);
69e1e811 301extern void set_cpu_sd_state_idle(void);
6201b4d6 302extern int get_nohz_timer_target(int pinned);
46cb4b7c 303#else
c1cc017c 304static inline void nohz_balance_enter_idle(int cpu) { }
fdaabd80 305static inline void set_cpu_sd_state_idle(void) { }
6201b4d6
VK
306static inline int get_nohz_timer_target(int pinned)
307{
308 return smp_processor_id();
309}
46cb4b7c 310#endif
1da177e4 311
e59e2ae2 312/*
39bc89fd 313 * Only dump TASK_* tasks. (0 for all tasks)
e59e2ae2
IM
314 */
315extern void show_state_filter(unsigned long state_filter);
316
317static inline void show_state(void)
318{
39bc89fd 319 show_state_filter(0);
e59e2ae2
IM
320}
321
1da177e4
LT
322extern void show_regs(struct pt_regs *);
323
324/*
325 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
326 * task), SP is the stack pointer of the first frame that should be shown in the back
327 * trace (or NULL if the entire call-chain of the task should be shown).
328 */
329extern void show_stack(struct task_struct *task, unsigned long *sp);
330
331void io_schedule(void);
332long io_schedule_timeout(long timeout);
333
334extern void cpu_init (void);
335extern void trap_init(void);
336extern void update_process_times(int user);
337extern void scheduler_tick(void);
338
82a1fcb9
IM
339extern void sched_show_task(struct task_struct *p);
340
19cc36c0 341#ifdef CONFIG_LOCKUP_DETECTOR
8446f1d3 342extern void touch_softlockup_watchdog(void);
d6ad3e28 343extern void touch_softlockup_watchdog_sync(void);
04c9167f 344extern void touch_all_softlockup_watchdogs(void);
332fbdbc
DZ
345extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
346 void __user *buffer,
347 size_t *lenp, loff_t *ppos);
9c44bc03 348extern unsigned int softlockup_panic;
004417a6 349void lockup_detector_init(void);
8446f1d3 350#else
8446f1d3
IM
351static inline void touch_softlockup_watchdog(void)
352{
353}
d6ad3e28
JW
354static inline void touch_softlockup_watchdog_sync(void)
355{
356}
04c9167f
JF
357static inline void touch_all_softlockup_watchdogs(void)
358{
359}
004417a6
PZ
360static inline void lockup_detector_init(void)
361{
362}
8446f1d3
IM
363#endif
364
8b414521
MT
365#ifdef CONFIG_DETECT_HUNG_TASK
366void reset_hung_task_detector(void);
367#else
368static inline void reset_hung_task_detector(void)
369{
370}
371#endif
372
1da177e4
LT
373/* Attach to any functions which should be ignored in wchan output. */
374#define __sched __attribute__((__section__(".sched.text")))
deaf2227
IM
375
376/* Linker adds these: start and end of __sched functions */
377extern char __sched_text_start[], __sched_text_end[];
378
1da177e4
LT
379/* Is this address in the __sched functions? */
380extern int in_sched_functions(unsigned long addr);
381
382#define MAX_SCHEDULE_TIMEOUT LONG_MAX
b3c97528 383extern signed long schedule_timeout(signed long timeout);
64ed93a2 384extern signed long schedule_timeout_interruptible(signed long timeout);
294d5cc2 385extern signed long schedule_timeout_killable(signed long timeout);
64ed93a2 386extern signed long schedule_timeout_uninterruptible(signed long timeout);
1da177e4 387asmlinkage void schedule(void);
c5491ea7 388extern void schedule_preempt_disabled(void);
1da177e4 389
ab516013 390struct nsproxy;
acce292c 391struct user_namespace;
1da177e4 392
efc1a3b1
DH
393#ifdef CONFIG_MMU
394extern void arch_pick_mmap_layout(struct mm_struct *mm);
1da177e4
LT
395extern unsigned long
396arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
397 unsigned long, unsigned long);
398extern unsigned long
399arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
400 unsigned long len, unsigned long pgoff,
401 unsigned long flags);
efc1a3b1
DH
402#else
403static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
404#endif
1da177e4 405
d049f74f
KC
406#define SUID_DUMP_DISABLE 0 /* No setuid dumping */
407#define SUID_DUMP_USER 1 /* Dump as user of process */
408#define SUID_DUMP_ROOT 2 /* Dump as root */
409
6c5d5238 410/* mm flags */
f8af4da3 411
7288e118 412/* for SUID_DUMP_* above */
3cb4a0bb 413#define MMF_DUMPABLE_BITS 2
f8af4da3 414#define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
3cb4a0bb 415
942be387
ON
416extern void set_dumpable(struct mm_struct *mm, int value);
417/*
418 * This returns the actual value of the suid_dumpable flag. For things
419 * that are using this for checking for privilege transitions, it must
420 * test against SUID_DUMP_USER rather than treating it as a boolean
421 * value.
422 */
423static inline int __get_dumpable(unsigned long mm_flags)
424{
425 return mm_flags & MMF_DUMPABLE_MASK;
426}
427
428static inline int get_dumpable(struct mm_struct *mm)
429{
430 return __get_dumpable(mm->flags);
431}
432
3cb4a0bb
KH
433/* coredump filter bits */
434#define MMF_DUMP_ANON_PRIVATE 2
435#define MMF_DUMP_ANON_SHARED 3
436#define MMF_DUMP_MAPPED_PRIVATE 4
437#define MMF_DUMP_MAPPED_SHARED 5
82df3973 438#define MMF_DUMP_ELF_HEADERS 6
e575f111
KM
439#define MMF_DUMP_HUGETLB_PRIVATE 7
440#define MMF_DUMP_HUGETLB_SHARED 8
f8af4da3 441
3cb4a0bb 442#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
e575f111 443#define MMF_DUMP_FILTER_BITS 7
3cb4a0bb
KH
444#define MMF_DUMP_FILTER_MASK \
445 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
446#define MMF_DUMP_FILTER_DEFAULT \
e575f111 447 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
656eb2cd
RM
448 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
449
450#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
451# define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
452#else
453# define MMF_DUMP_MASK_DEFAULT_ELF 0
454#endif
f8af4da3
HD
455 /* leave room for more dump flags */
456#define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
ba76149f 457#define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
bafb282d 458#define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
f8af4da3 459
9f68f672
ON
460#define MMF_HAS_UPROBES 19 /* has uprobes */
461#define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
f8ac4ec9 462
f8af4da3 463#define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
6c5d5238 464
1da177e4
LT
465struct sighand_struct {
466 atomic_t count;
467 struct k_sigaction action[_NSIG];
468 spinlock_t siglock;
b8fceee1 469 wait_queue_head_t signalfd_wqh;
1da177e4
LT
470};
471
0e464814 472struct pacct_struct {
f6ec29a4
KK
473 int ac_flag;
474 long ac_exitcode;
0e464814 475 unsigned long ac_mem;
77787bfb
KK
476 cputime_t ac_utime, ac_stime;
477 unsigned long ac_minflt, ac_majflt;
0e464814
KK
478};
479
42c4ab41
SG
480struct cpu_itimer {
481 cputime_t expires;
482 cputime_t incr;
8356b5f9
SG
483 u32 error;
484 u32 incr_error;
42c4ab41
SG
485};
486
d37f761d
FW
487/**
488 * struct cputime - snaphsot of system and user cputime
489 * @utime: time spent in user mode
490 * @stime: time spent in system mode
491 *
492 * Gathers a generic snapshot of user and system time.
493 */
494struct cputime {
495 cputime_t utime;
496 cputime_t stime;
497};
498
f06febc9
FM
499/**
500 * struct task_cputime - collected CPU time counts
501 * @utime: time spent in user mode, in &cputime_t units
502 * @stime: time spent in kernel mode, in &cputime_t units
503 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
5ce73a4a 504 *
d37f761d
FW
505 * This is an extension of struct cputime that includes the total runtime
506 * spent by the task from the scheduler point of view.
507 *
508 * As a result, this structure groups together three kinds of CPU time
509 * that are tracked for threads and thread groups. Most things considering
f06febc9
FM
510 * CPU time want to group these counts together and treat all three
511 * of them in parallel.
512 */
513struct task_cputime {
514 cputime_t utime;
515 cputime_t stime;
516 unsigned long long sum_exec_runtime;
517};
518/* Alternate field names when used to cache expirations. */
519#define prof_exp stime
520#define virt_exp utime
521#define sched_exp sum_exec_runtime
522
4cd4c1b4
PZ
523#define INIT_CPUTIME \
524 (struct task_cputime) { \
64861634
MS
525 .utime = 0, \
526 .stime = 0, \
4cd4c1b4
PZ
527 .sum_exec_runtime = 0, \
528 }
529
a233f112
PZ
530#ifdef CONFIG_PREEMPT_COUNT
531#define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
532#else
533#define PREEMPT_DISABLED PREEMPT_ENABLED
534#endif
535
c99e6efe
PZ
536/*
537 * Disable preemption until the scheduler is running.
538 * Reset by start_kernel()->sched_init()->init_idle().
d86ee480
PZ
539 *
540 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
541 * before the scheduler is active -- see should_resched().
c99e6efe 542 */
a233f112 543#define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
c99e6efe 544
f06febc9 545/**
4cd4c1b4
PZ
546 * struct thread_group_cputimer - thread group interval timer counts
547 * @cputime: thread group interval timers.
548 * @running: non-zero when there are timers running and
549 * @cputime receives updates.
550 * @lock: lock for fields in this struct.
f06febc9
FM
551 *
552 * This structure contains the version of task_cputime, above, that is
4cd4c1b4 553 * used for thread group CPU timer calculations.
f06febc9 554 */
4cd4c1b4
PZ
555struct thread_group_cputimer {
556 struct task_cputime cputime;
557 int running;
ee30a7b2 558 raw_spinlock_t lock;
f06febc9 559};
f06febc9 560
4714d1d3 561#include <linux/rwsem.h>
5091faa4
MG
562struct autogroup;
563
1da177e4 564/*
e815f0a8 565 * NOTE! "signal_struct" does not have its own
1da177e4
LT
566 * locking, because a shared signal_struct always
567 * implies a shared sighand_struct, so locking
568 * sighand_struct is always a proper superset of
569 * the locking of signal_struct.
570 */
571struct signal_struct {
ea6d290c 572 atomic_t sigcnt;
1da177e4 573 atomic_t live;
b3ac022c 574 int nr_threads;
0c740d0a 575 struct list_head thread_head;
1da177e4
LT
576
577 wait_queue_head_t wait_chldexit; /* for wait4() */
578
579 /* current thread group signal load-balancing target: */
36c8b586 580 struct task_struct *curr_target;
1da177e4
LT
581
582 /* shared signal handling: */
583 struct sigpending shared_pending;
584
585 /* thread group exit support */
586 int group_exit_code;
587 /* overloaded:
588 * - notify group_exit_task when ->count is equal to notify_count
589 * - everyone except group_exit_task is stopped during signal delivery
590 * of fatal signals, group_exit_task processes the signal.
591 */
1da177e4 592 int notify_count;
07dd20e0 593 struct task_struct *group_exit_task;
1da177e4
LT
594
595 /* thread group stop support, overloads group_exit_code too */
596 int group_stop_count;
597 unsigned int flags; /* see SIGNAL_* flags below */
598
ebec18a6
LP
599 /*
600 * PR_SET_CHILD_SUBREAPER marks a process, like a service
601 * manager, to re-parent orphan (double-forking) child processes
602 * to this process instead of 'init'. The service manager is
603 * able to receive SIGCHLD signals and is able to investigate
604 * the process until it calls wait(). All children of this
605 * process will inherit a flag if they should look for a
606 * child_subreaper process at exit.
607 */
608 unsigned int is_child_subreaper:1;
609 unsigned int has_child_subreaper:1;
610
1da177e4 611 /* POSIX.1b Interval Timers */
5ed67f05
PE
612 int posix_timer_id;
613 struct list_head posix_timers;
1da177e4
LT
614
615 /* ITIMER_REAL timer for the process */
2ff678b8 616 struct hrtimer real_timer;
fea9d175 617 struct pid *leader_pid;
2ff678b8 618 ktime_t it_real_incr;
1da177e4 619
42c4ab41
SG
620 /*
621 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
622 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
623 * values are defined to 0 and 1 respectively
624 */
625 struct cpu_itimer it[2];
1da177e4 626
f06febc9 627 /*
4cd4c1b4
PZ
628 * Thread group totals for process CPU timers.
629 * See thread_group_cputimer(), et al, for details.
f06febc9 630 */
4cd4c1b4 631 struct thread_group_cputimer cputimer;
f06febc9
FM
632
633 /* Earliest-expiration cache. */
634 struct task_cputime cputime_expires;
635
636 struct list_head cpu_timers[3];
637
ab521dc0 638 struct pid *tty_old_pgrp;
1ec320af 639
1da177e4
LT
640 /* boolean value for session group leader */
641 int leader;
642
643 struct tty_struct *tty; /* NULL if no tty */
644
5091faa4
MG
645#ifdef CONFIG_SCHED_AUTOGROUP
646 struct autogroup *autogroup;
647#endif
1da177e4
LT
648 /*
649 * Cumulative resource counters for dead threads in the group,
650 * and for reaped dead child processes forked by this group.
651 * Live threads maintain their own counters and add to these
652 * in __exit_signal, except for the group leader.
653 */
32bd671d 654 cputime_t utime, stime, cutime, cstime;
9ac52315
LV
655 cputime_t gtime;
656 cputime_t cgtime;
9fbc42ea 657#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
d37f761d 658 struct cputime prev_cputime;
0cf55e1e 659#endif
1da177e4
LT
660 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
661 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
6eaeeaba 662 unsigned long inblock, oublock, cinblock, coublock;
1f10206c 663 unsigned long maxrss, cmaxrss;
940389b8 664 struct task_io_accounting ioac;
1da177e4 665
32bd671d
PZ
666 /*
667 * Cumulative ns of schedule CPU time fo dead threads in the
668 * group, not including a zombie group leader, (This only differs
669 * from jiffies_to_ns(utime + stime) if sched_clock uses something
670 * other than jiffies.)
671 */
672 unsigned long long sum_sched_runtime;
673
1da177e4
LT
674 /*
675 * We don't bother to synchronize most readers of this at all,
676 * because there is no reader checking a limit that actually needs
677 * to get both rlim_cur and rlim_max atomically, and either one
678 * alone is a single word that can safely be read normally.
679 * getrlimit/setrlimit use task_lock(current->group_leader) to
680 * protect this instead of the siglock, because they really
681 * have no need to disable irqs.
682 */
683 struct rlimit rlim[RLIM_NLIMITS];
684
0e464814
KK
685#ifdef CONFIG_BSD_PROCESS_ACCT
686 struct pacct_struct pacct; /* per-process accounting information */
687#endif
ad4ecbcb 688#ifdef CONFIG_TASKSTATS
ad4ecbcb
SN
689 struct taskstats *stats;
690#endif
522ed776
MT
691#ifdef CONFIG_AUDIT
692 unsigned audit_tty;
46e959ea 693 unsigned audit_tty_log_passwd;
522ed776
MT
694 struct tty_audit_buf *tty_audit_buf;
695#endif
4714d1d3
BB
696#ifdef CONFIG_CGROUPS
697 /*
77e4ef99
TH
698 * group_rwsem prevents new tasks from entering the threadgroup and
699 * member tasks from exiting,a more specifically, setting of
700 * PF_EXITING. fork and exit paths are protected with this rwsem
701 * using threadgroup_change_begin/end(). Users which require
702 * threadgroup to remain stable should use threadgroup_[un]lock()
703 * which also takes care of exec path. Currently, cgroup is the
704 * only user.
4714d1d3 705 */
257058ae 706 struct rw_semaphore group_rwsem;
4714d1d3 707#endif
28b83c51 708
e1e12d2f 709 oom_flags_t oom_flags;
a9c58b90
DR
710 short oom_score_adj; /* OOM kill score adjustment */
711 short oom_score_adj_min; /* OOM kill score adjustment min value.
712 * Only settable by CAP_SYS_RESOURCE. */
9b1bf12d
KM
713
714 struct mutex cred_guard_mutex; /* guard against foreign influences on
715 * credential calculations
716 * (notably. ptrace) */
1da177e4
LT
717};
718
719/*
720 * Bits in flags field of signal_struct.
721 */
722#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
ee77f075
ON
723#define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
724#define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
403bad72 725#define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
e4420551
ON
726/*
727 * Pending notifications to parent.
728 */
729#define SIGNAL_CLD_STOPPED 0x00000010
730#define SIGNAL_CLD_CONTINUED 0x00000020
731#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
1da177e4 732
fae5fa44
ON
733#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
734
ed5d2cac
ON
735/* If true, all threads except ->group_exit_task have pending SIGKILL */
736static inline int signal_group_exit(const struct signal_struct *sig)
737{
738 return (sig->flags & SIGNAL_GROUP_EXIT) ||
739 (sig->group_exit_task != NULL);
740}
741
1da177e4
LT
742/*
743 * Some day this will be a full-fledged user tracking system..
744 */
745struct user_struct {
746 atomic_t __count; /* reference count */
747 atomic_t processes; /* How many processes does this user have? */
748 atomic_t files; /* How many open files does this user have? */
749 atomic_t sigpending; /* How many pending signals does this user have? */
2d9048e2 750#ifdef CONFIG_INOTIFY_USER
0eeca283
RL
751 atomic_t inotify_watches; /* How many inotify watches does this user have? */
752 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
753#endif
4afeff85
EP
754#ifdef CONFIG_FANOTIFY
755 atomic_t fanotify_listeners;
756#endif
7ef9964e 757#ifdef CONFIG_EPOLL
52bd19f7 758 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
7ef9964e 759#endif
970a8645 760#ifdef CONFIG_POSIX_MQUEUE
1da177e4
LT
761 /* protected by mq_lock */
762 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
970a8645 763#endif
1da177e4
LT
764 unsigned long locked_shm; /* How many pages of mlocked shm ? */
765
766#ifdef CONFIG_KEYS
767 struct key *uid_keyring; /* UID specific keyring */
768 struct key *session_keyring; /* UID's default session keyring */
769#endif
770
771 /* Hash table maintenance information */
735de223 772 struct hlist_node uidhash_node;
7b44ab97 773 kuid_t uid;
24e377a8 774
cdd6c482 775#ifdef CONFIG_PERF_EVENTS
789f90fc
PZ
776 atomic_long_t locked_vm;
777#endif
1da177e4
LT
778};
779
eb41d946 780extern int uids_sysfs_init(void);
5cb350ba 781
7b44ab97 782extern struct user_struct *find_user(kuid_t);
1da177e4
LT
783
784extern struct user_struct root_user;
785#define INIT_USER (&root_user)
786
b6dff3ec 787
1da177e4
LT
788struct backing_dev_info;
789struct reclaim_state;
790
52f17b6c 791#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1da177e4
LT
792struct sched_info {
793 /* cumulative counters */
2d72376b 794 unsigned long pcount; /* # of times run on this cpu */
9c2c4802 795 unsigned long long run_delay; /* time spent waiting on a runqueue */
1da177e4
LT
796
797 /* timestamps */
172ba844
BS
798 unsigned long long last_arrival,/* when we last ran on a cpu */
799 last_queued; /* when we were last queued to run */
1da177e4 800};
52f17b6c 801#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
1da177e4 802
ca74e92b
SN
803#ifdef CONFIG_TASK_DELAY_ACCT
804struct task_delay_info {
805 spinlock_t lock;
806 unsigned int flags; /* Private per-task flags */
807
808 /* For each stat XXX, add following, aligned appropriately
809 *
810 * struct timespec XXX_start, XXX_end;
811 * u64 XXX_delay;
812 * u32 XXX_count;
813 *
814 * Atomicity of updates to XXX_delay, XXX_count protected by
815 * single lock above (split into XXX_lock if contention is an issue).
816 */
0ff92245
SN
817
818 /*
819 * XXX_count is incremented on every XXX operation, the delay
820 * associated with the operation is added to XXX_delay.
821 * XXX_delay contains the accumulated delay time in nanoseconds.
822 */
823 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
824 u64 blkio_delay; /* wait for sync block io completion */
825 u64 swapin_delay; /* wait for swapin block io completion */
826 u32 blkio_count; /* total count of the number of sync block */
827 /* io operations performed */
828 u32 swapin_count; /* total count of the number of swapin block */
829 /* io operations performed */
873b4771
KK
830
831 struct timespec freepages_start, freepages_end;
832 u64 freepages_delay; /* wait for memory reclaim */
833 u32 freepages_count; /* total count of memory reclaim */
ca74e92b 834};
52f17b6c
CS
835#endif /* CONFIG_TASK_DELAY_ACCT */
836
837static inline int sched_info_on(void)
838{
839#ifdef CONFIG_SCHEDSTATS
840 return 1;
841#elif defined(CONFIG_TASK_DELAY_ACCT)
842 extern int delayacct_on;
843 return delayacct_on;
844#else
845 return 0;
ca74e92b 846#endif
52f17b6c 847}
ca74e92b 848
d15bcfdb
IM
849enum cpu_idle_type {
850 CPU_IDLE,
851 CPU_NOT_IDLE,
852 CPU_NEWLY_IDLE,
853 CPU_MAX_IDLE_TYPES
1da177e4
LT
854};
855
1399fa78
NR
856/*
857 * Increase resolution of cpu_power calculations
858 */
859#define SCHED_POWER_SHIFT 10
860#define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
1da177e4 861
1399fa78
NR
862/*
863 * sched-domains (multiprocessor balancing) declarations:
864 */
2dd73a4f 865#ifdef CONFIG_SMP
b5d978e0
PZ
866#define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
867#define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
868#define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
869#define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
c88d5910 870#define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
b5d978e0 871#define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
b5d978e0 872#define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
d77b3ed5 873#define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
b5d978e0
PZ
874#define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
875#define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
532cb4c4 876#define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
b5d978e0 877#define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
e3589f6c 878#define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
3a7053b3 879#define SD_NUMA 0x4000 /* cross-node balancing */
5c45bf27 880
143e1e28
VG
881#ifdef CONFIG_SCHED_SMT
882static inline const int cpu_smt_flags(void)
883{
884 return SD_SHARE_CPUPOWER | SD_SHARE_PKG_RESOURCES;
885}
886#endif
887
888#ifdef CONFIG_SCHED_MC
889static inline const int cpu_core_flags(void)
890{
891 return SD_SHARE_PKG_RESOURCES;
892}
893#endif
894
895#ifdef CONFIG_NUMA
896static inline const int cpu_numa_flags(void)
897{
898 return SD_NUMA;
899}
900#endif
901
1d3504fc
HS
902struct sched_domain_attr {
903 int relax_domain_level;
904};
905
906#define SD_ATTR_INIT (struct sched_domain_attr) { \
907 .relax_domain_level = -1, \
908}
909
60495e77
PZ
910extern int sched_domain_level_max;
911
5e6521ea
LZ
912struct sched_group;
913
1da177e4
LT
914struct sched_domain {
915 /* These fields must be setup */
916 struct sched_domain *parent; /* top domain must be null terminated */
1a848870 917 struct sched_domain *child; /* bottom domain must be null terminated */
1da177e4 918 struct sched_group *groups; /* the balancing groups of the domain */
1da177e4
LT
919 unsigned long min_interval; /* Minimum balance interval ms */
920 unsigned long max_interval; /* Maximum balance interval ms */
921 unsigned int busy_factor; /* less balancing by factor if busy */
922 unsigned int imbalance_pct; /* No balance until over watermark */
1da177e4 923 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
7897986b
NP
924 unsigned int busy_idx;
925 unsigned int idle_idx;
926 unsigned int newidle_idx;
927 unsigned int wake_idx;
147cbb4b 928 unsigned int forkexec_idx;
a52bfd73 929 unsigned int smt_gain;
25f55d9d
VG
930
931 int nohz_idle; /* NOHZ IDLE status */
1da177e4 932 int flags; /* See SD_* */
60495e77 933 int level;
1da177e4
LT
934
935 /* Runtime fields. */
936 unsigned long last_balance; /* init to jiffies. units in jiffies */
937 unsigned int balance_interval; /* initialise to 1. units in ms. */
938 unsigned int nr_balance_failed; /* initialise to 0 */
939
f48627e6 940 /* idle_balance() stats */
9bd721c5 941 u64 max_newidle_lb_cost;
f48627e6 942 unsigned long next_decay_max_lb_cost;
2398f2c6 943
1da177e4
LT
944#ifdef CONFIG_SCHEDSTATS
945 /* load_balance() stats */
480b9434
KC
946 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
947 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
948 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
949 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
950 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
951 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
952 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
953 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1da177e4
LT
954
955 /* Active load balancing */
480b9434
KC
956 unsigned int alb_count;
957 unsigned int alb_failed;
958 unsigned int alb_pushed;
1da177e4 959
68767a0a 960 /* SD_BALANCE_EXEC stats */
480b9434
KC
961 unsigned int sbe_count;
962 unsigned int sbe_balanced;
963 unsigned int sbe_pushed;
1da177e4 964
68767a0a 965 /* SD_BALANCE_FORK stats */
480b9434
KC
966 unsigned int sbf_count;
967 unsigned int sbf_balanced;
968 unsigned int sbf_pushed;
68767a0a 969
1da177e4 970 /* try_to_wake_up() stats */
480b9434
KC
971 unsigned int ttwu_wake_remote;
972 unsigned int ttwu_move_affine;
973 unsigned int ttwu_move_balance;
1da177e4 974#endif
a5d8c348
IM
975#ifdef CONFIG_SCHED_DEBUG
976 char *name;
977#endif
dce840a0
PZ
978 union {
979 void *private; /* used during construction */
980 struct rcu_head rcu; /* used during destruction */
981 };
6c99e9ad 982
669c55e9 983 unsigned int span_weight;
4200efd9
IM
984 /*
985 * Span of all CPUs in this domain.
986 *
987 * NOTE: this field is variable length. (Allocated dynamically
988 * by attaching extra space to the end of the structure,
989 * depending on how many CPUs the kernel has booted up with)
4200efd9
IM
990 */
991 unsigned long span[0];
1da177e4
LT
992};
993
758b2cdc
RR
994static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
995{
6c99e9ad 996 return to_cpumask(sd->span);
758b2cdc
RR
997}
998
acc3f5d7 999extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1d3504fc 1000 struct sched_domain_attr *dattr_new);
029190c5 1001
acc3f5d7
RR
1002/* Allocate an array of sched domains, for partition_sched_domains(). */
1003cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1004void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1005
39be3501
PZ
1006bool cpus_share_cache(int this_cpu, int that_cpu);
1007
143e1e28
VG
1008typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1009typedef const int (*sched_domain_flags_f)(void);
1010
1011#define SDTL_OVERLAP 0x01
1012
1013struct sd_data {
1014 struct sched_domain **__percpu sd;
1015 struct sched_group **__percpu sg;
63b2ca30 1016 struct sched_group_capacity **__percpu sgc;
143e1e28
VG
1017};
1018
1019struct sched_domain_topology_level {
1020 sched_domain_mask_f mask;
1021 sched_domain_flags_f sd_flags;
1022 int flags;
1023 int numa_level;
1024 struct sd_data data;
1025#ifdef CONFIG_SCHED_DEBUG
1026 char *name;
1027#endif
1028};
1029
1030extern struct sched_domain_topology_level *sched_domain_topology;
1031
1032extern void set_sched_topology(struct sched_domain_topology_level *tl);
1033
1034#ifdef CONFIG_SCHED_DEBUG
1035# define SD_INIT_NAME(type) .name = #type
1036#else
1037# define SD_INIT_NAME(type)
1038#endif
1039
1b427c15 1040#else /* CONFIG_SMP */
1da177e4 1041
1b427c15 1042struct sched_domain_attr;
d02c7a8c 1043
1b427c15 1044static inline void
acc3f5d7 1045partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1b427c15
IM
1046 struct sched_domain_attr *dattr_new)
1047{
d02c7a8c 1048}
39be3501
PZ
1049
1050static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1051{
1052 return true;
1053}
1054
1b427c15 1055#endif /* !CONFIG_SMP */
1da177e4 1056
47fe38fc 1057
1da177e4 1058struct io_context; /* See blkdev.h */
1da177e4 1059
1da177e4 1060
383f2835 1061#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
36c8b586 1062extern void prefetch_stack(struct task_struct *t);
383f2835
CK
1063#else
1064static inline void prefetch_stack(struct task_struct *t) { }
1065#endif
1da177e4
LT
1066
1067struct audit_context; /* See audit.c */
1068struct mempolicy;
b92ce558 1069struct pipe_inode_info;
4865ecf1 1070struct uts_namespace;
1da177e4 1071
20b8a59f 1072struct load_weight {
9dbdb155
PZ
1073 unsigned long weight;
1074 u32 inv_weight;
20b8a59f
IM
1075};
1076
9d85f21c
PT
1077struct sched_avg {
1078 /*
1079 * These sums represent an infinite geometric series and so are bound
239003ea 1080 * above by 1024/(1-y). Thus we only need a u32 to store them for all
9d85f21c
PT
1081 * choices of y < 1-2^(-32)*1024.
1082 */
1083 u32 runnable_avg_sum, runnable_avg_period;
1084 u64 last_runnable_update;
9ee474f5 1085 s64 decay_count;
2dac754e 1086 unsigned long load_avg_contrib;
9d85f21c
PT
1087};
1088
94c18227 1089#ifdef CONFIG_SCHEDSTATS
41acab88 1090struct sched_statistics {
20b8a59f 1091 u64 wait_start;
94c18227 1092 u64 wait_max;
6d082592
AV
1093 u64 wait_count;
1094 u64 wait_sum;
8f0dfc34
AV
1095 u64 iowait_count;
1096 u64 iowait_sum;
94c18227 1097
20b8a59f 1098 u64 sleep_start;
20b8a59f 1099 u64 sleep_max;
94c18227
IM
1100 s64 sum_sleep_runtime;
1101
1102 u64 block_start;
20b8a59f
IM
1103 u64 block_max;
1104 u64 exec_max;
eba1ed4b 1105 u64 slice_max;
cc367732 1106
cc367732
IM
1107 u64 nr_migrations_cold;
1108 u64 nr_failed_migrations_affine;
1109 u64 nr_failed_migrations_running;
1110 u64 nr_failed_migrations_hot;
1111 u64 nr_forced_migrations;
cc367732
IM
1112
1113 u64 nr_wakeups;
1114 u64 nr_wakeups_sync;
1115 u64 nr_wakeups_migrate;
1116 u64 nr_wakeups_local;
1117 u64 nr_wakeups_remote;
1118 u64 nr_wakeups_affine;
1119 u64 nr_wakeups_affine_attempts;
1120 u64 nr_wakeups_passive;
1121 u64 nr_wakeups_idle;
41acab88
LDM
1122};
1123#endif
1124
1125struct sched_entity {
1126 struct load_weight load; /* for load-balancing */
1127 struct rb_node run_node;
1128 struct list_head group_node;
1129 unsigned int on_rq;
1130
1131 u64 exec_start;
1132 u64 sum_exec_runtime;
1133 u64 vruntime;
1134 u64 prev_sum_exec_runtime;
1135
41acab88
LDM
1136 u64 nr_migrations;
1137
41acab88
LDM
1138#ifdef CONFIG_SCHEDSTATS
1139 struct sched_statistics statistics;
94c18227
IM
1140#endif
1141
20b8a59f 1142#ifdef CONFIG_FAIR_GROUP_SCHED
fed14d45 1143 int depth;
20b8a59f
IM
1144 struct sched_entity *parent;
1145 /* rq on which this entity is (to be) queued: */
1146 struct cfs_rq *cfs_rq;
1147 /* rq "owned" by this entity/group: */
1148 struct cfs_rq *my_q;
1149#endif
8bd75c77 1150
141965c7 1151#ifdef CONFIG_SMP
f4e26b12 1152 /* Per-entity load-tracking */
9d85f21c
PT
1153 struct sched_avg avg;
1154#endif
20b8a59f 1155};
70b97a7f 1156
fa717060
PZ
1157struct sched_rt_entity {
1158 struct list_head run_list;
78f2c7db 1159 unsigned long timeout;
57d2aa00 1160 unsigned long watchdog_stamp;
bee367ed 1161 unsigned int time_slice;
6f505b16 1162
58d6c2d7 1163 struct sched_rt_entity *back;
052f1dc7 1164#ifdef CONFIG_RT_GROUP_SCHED
6f505b16
PZ
1165 struct sched_rt_entity *parent;
1166 /* rq on which this entity is (to be) queued: */
1167 struct rt_rq *rt_rq;
1168 /* rq "owned" by this entity/group: */
1169 struct rt_rq *my_q;
1170#endif
fa717060
PZ
1171};
1172
aab03e05
DF
1173struct sched_dl_entity {
1174 struct rb_node rb_node;
1175
1176 /*
1177 * Original scheduling parameters. Copied here from sched_attr
4027d080 1178 * during sched_setattr(), they will remain the same until
1179 * the next sched_setattr().
aab03e05
DF
1180 */
1181 u64 dl_runtime; /* maximum runtime for each instance */
1182 u64 dl_deadline; /* relative deadline of each instance */
755378a4 1183 u64 dl_period; /* separation of two instances (period) */
332ac17e 1184 u64 dl_bw; /* dl_runtime / dl_deadline */
aab03e05
DF
1185
1186 /*
1187 * Actual scheduling parameters. Initialized with the values above,
1188 * they are continously updated during task execution. Note that
1189 * the remaining runtime could be < 0 in case we are in overrun.
1190 */
1191 s64 runtime; /* remaining runtime for this instance */
1192 u64 deadline; /* absolute deadline for this instance */
1193 unsigned int flags; /* specifying the scheduler behaviour */
1194
1195 /*
1196 * Some bool flags:
1197 *
1198 * @dl_throttled tells if we exhausted the runtime. If so, the
1199 * task has to wait for a replenishment to be performed at the
1200 * next firing of dl_timer.
1201 *
1202 * @dl_new tells if a new instance arrived. If so we must
1203 * start executing it with full runtime and reset its absolute
1204 * deadline;
2d3d891d
DF
1205 *
1206 * @dl_boosted tells if we are boosted due to DI. If so we are
1207 * outside bandwidth enforcement mechanism (but only until we
5bfd126e
JL
1208 * exit the critical section);
1209 *
1210 * @dl_yielded tells if task gave up the cpu before consuming
1211 * all its available runtime during the last job.
aab03e05 1212 */
5bfd126e 1213 int dl_throttled, dl_new, dl_boosted, dl_yielded;
aab03e05
DF
1214
1215 /*
1216 * Bandwidth enforcement timer. Each -deadline task has its
1217 * own bandwidth to be enforced, thus we need one timer per task.
1218 */
1219 struct hrtimer dl_timer;
1220};
8bd75c77 1221
86848966
PM
1222struct rcu_node;
1223
8dc85d54
PZ
1224enum perf_event_task_context {
1225 perf_invalid_context = -1,
1226 perf_hw_context = 0,
89a1e187 1227 perf_sw_context,
8dc85d54
PZ
1228 perf_nr_task_contexts,
1229};
1230
1da177e4
LT
1231struct task_struct {
1232 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
f7e4217b 1233 void *stack;
1da177e4 1234 atomic_t usage;
97dc32cd
WC
1235 unsigned int flags; /* per process flags, defined below */
1236 unsigned int ptrace;
1da177e4 1237
2dd73a4f 1238#ifdef CONFIG_SMP
fa14ff4a 1239 struct llist_node wake_entry;
3ca7a440 1240 int on_cpu;
62470419
MW
1241 struct task_struct *last_wakee;
1242 unsigned long wakee_flips;
1243 unsigned long wakee_flip_decay_ts;
ac66f547
PZ
1244
1245 int wake_cpu;
2dd73a4f 1246#endif
fd2f4419 1247 int on_rq;
50e645a8 1248
b29739f9 1249 int prio, static_prio, normal_prio;
c7aceaba 1250 unsigned int rt_priority;
5522d5d5 1251 const struct sched_class *sched_class;
20b8a59f 1252 struct sched_entity se;
fa717060 1253 struct sched_rt_entity rt;
8323f26c
PZ
1254#ifdef CONFIG_CGROUP_SCHED
1255 struct task_group *sched_task_group;
1256#endif
aab03e05 1257 struct sched_dl_entity dl;
1da177e4 1258
e107be36
AK
1259#ifdef CONFIG_PREEMPT_NOTIFIERS
1260 /* list of struct preempt_notifier: */
1261 struct hlist_head preempt_notifiers;
1262#endif
1263
6c5c9341 1264#ifdef CONFIG_BLK_DEV_IO_TRACE
2056a782 1265 unsigned int btrace_seq;
6c5c9341 1266#endif
1da177e4 1267
97dc32cd 1268 unsigned int policy;
29baa747 1269 int nr_cpus_allowed;
1da177e4 1270 cpumask_t cpus_allowed;
1da177e4 1271
a57eb940 1272#ifdef CONFIG_PREEMPT_RCU
e260be67 1273 int rcu_read_lock_nesting;
f41d911f 1274 char rcu_read_unlock_special;
f41d911f 1275 struct list_head rcu_node_entry;
a57eb940
PM
1276#endif /* #ifdef CONFIG_PREEMPT_RCU */
1277#ifdef CONFIG_TREE_PREEMPT_RCU
1278 struct rcu_node *rcu_blocked_node;
f41d911f 1279#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
24278d14
PM
1280#ifdef CONFIG_RCU_BOOST
1281 struct rt_mutex *rcu_boost_mutex;
1282#endif /* #ifdef CONFIG_RCU_BOOST */
e260be67 1283
52f17b6c 1284#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1da177e4
LT
1285 struct sched_info sched_info;
1286#endif
1287
1288 struct list_head tasks;
806c09a7 1289#ifdef CONFIG_SMP
917b627d 1290 struct plist_node pushable_tasks;
1baca4ce 1291 struct rb_node pushable_dl_tasks;
806c09a7 1292#endif
1da177e4
LT
1293
1294 struct mm_struct *mm, *active_mm;
4471a675
JK
1295#ifdef CONFIG_COMPAT_BRK
1296 unsigned brk_randomized:1;
1297#endif
615d6e87
DB
1298 /* per-thread vma caching */
1299 u32 vmacache_seqnum;
1300 struct vm_area_struct *vmacache[VMACACHE_SIZE];
34e55232
KH
1301#if defined(SPLIT_RSS_COUNTING)
1302 struct task_rss_stat rss_stat;
1303#endif
1da177e4 1304/* task state */
97dc32cd 1305 int exit_state;
1da177e4
LT
1306 int exit_code, exit_signal;
1307 int pdeath_signal; /* The signal sent when the parent dies */
a8f072c1 1308 unsigned int jobctl; /* JOBCTL_*, siglock protected */
9b89f6ba
AE
1309
1310 /* Used for emulating ABI behavior of previous Linux versions */
97dc32cd 1311 unsigned int personality;
9b89f6ba 1312
f9ce1f1c
KT
1313 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1314 * execve */
8f0dfc34
AV
1315 unsigned in_iowait:1;
1316
259e5e6c
AL
1317 /* task may not gain privileges */
1318 unsigned no_new_privs:1;
ca94c442
LP
1319
1320 /* Revert to default priority/policy when forking */
1321 unsigned sched_reset_on_fork:1;
a8e4f2ea 1322 unsigned sched_contributes_to_load:1;
ca94c442 1323
1da177e4
LT
1324 pid_t pid;
1325 pid_t tgid;
0a425405 1326
1314562a 1327#ifdef CONFIG_CC_STACKPROTECTOR
0a425405
AV
1328 /* Canary value for the -fstack-protector gcc feature */
1329 unsigned long stack_canary;
1314562a 1330#endif
4d1d61a6 1331 /*
1da177e4 1332 * pointers to (original) parent process, youngest child, younger sibling,
4d1d61a6 1333 * older sibling, respectively. (p->father can be replaced with
f470021a 1334 * p->real_parent->pid)
1da177e4 1335 */
abd63bc3
KC
1336 struct task_struct __rcu *real_parent; /* real parent process */
1337 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1da177e4 1338 /*
f470021a 1339 * children/sibling forms the list of my natural children
1da177e4
LT
1340 */
1341 struct list_head children; /* list of my children */
1342 struct list_head sibling; /* linkage in my parent's children list */
1343 struct task_struct *group_leader; /* threadgroup leader */
1344
f470021a
RM
1345 /*
1346 * ptraced is the list of tasks this task is using ptrace on.
1347 * This includes both natural children and PTRACE_ATTACH targets.
1348 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1349 */
1350 struct list_head ptraced;
1351 struct list_head ptrace_entry;
1352
1da177e4 1353 /* PID/PID hash table linkage. */
92476d7f 1354 struct pid_link pids[PIDTYPE_MAX];
47e65328 1355 struct list_head thread_group;
0c740d0a 1356 struct list_head thread_node;
1da177e4
LT
1357
1358 struct completion *vfork_done; /* for vfork() */
1359 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1360 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1361
c66f08be 1362 cputime_t utime, stime, utimescaled, stimescaled;
9ac52315 1363 cputime_t gtime;
9fbc42ea 1364#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
d37f761d 1365 struct cputime prev_cputime;
6a61671b
FW
1366#endif
1367#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1368 seqlock_t vtime_seqlock;
1369 unsigned long long vtime_snap;
1370 enum {
1371 VTIME_SLEEPING = 0,
1372 VTIME_USER,
1373 VTIME_SYS,
1374 } vtime_snap_whence;
d99ca3b9 1375#endif
1da177e4 1376 unsigned long nvcsw, nivcsw; /* context switch counts */
924b42d5
TJ
1377 struct timespec start_time; /* monotonic time */
1378 struct timespec real_start_time; /* boot based time */
1da177e4
LT
1379/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1380 unsigned long min_flt, maj_flt;
1381
f06febc9 1382 struct task_cputime cputime_expires;
1da177e4
LT
1383 struct list_head cpu_timers[3];
1384
1385/* process credentials */
1b0ba1c9 1386 const struct cred __rcu *real_cred; /* objective and real subjective task
3b11a1de 1387 * credentials (COW) */
1b0ba1c9 1388 const struct cred __rcu *cred; /* effective (overridable) subjective task
3b11a1de 1389 * credentials (COW) */
36772092
PBG
1390 char comm[TASK_COMM_LEN]; /* executable name excluding path
1391 - access with [gs]et_task_comm (which lock
1392 it with task_lock())
221af7f8 1393 - initialized normally by setup_new_exec */
1da177e4
LT
1394/* file system info */
1395 int link_count, total_link_count;
3d5b6fcc 1396#ifdef CONFIG_SYSVIPC
1da177e4
LT
1397/* ipc stuff */
1398 struct sysv_sem sysvsem;
3d5b6fcc 1399#endif
e162b39a 1400#ifdef CONFIG_DETECT_HUNG_TASK
82a1fcb9 1401/* hung task detection */
82a1fcb9
IM
1402 unsigned long last_switch_count;
1403#endif
1da177e4
LT
1404/* CPU-specific state of this task */
1405 struct thread_struct thread;
1406/* filesystem information */
1407 struct fs_struct *fs;
1408/* open file information */
1409 struct files_struct *files;
1651e14e 1410/* namespaces */
ab516013 1411 struct nsproxy *nsproxy;
1da177e4
LT
1412/* signal handlers */
1413 struct signal_struct *signal;
1414 struct sighand_struct *sighand;
1415
1416 sigset_t blocked, real_blocked;
f3de272b 1417 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1da177e4
LT
1418 struct sigpending pending;
1419
1420 unsigned long sas_ss_sp;
1421 size_t sas_ss_size;
1422 int (*notifier)(void *priv);
1423 void *notifier_data;
1424 sigset_t *notifier_mask;
67d12145 1425 struct callback_head *task_works;
e73f8959 1426
1da177e4 1427 struct audit_context *audit_context;
bfef93a5 1428#ifdef CONFIG_AUDITSYSCALL
e1760bd5 1429 kuid_t loginuid;
4746ec5b 1430 unsigned int sessionid;
bfef93a5 1431#endif
932ecebb 1432 struct seccomp seccomp;
1da177e4
LT
1433
1434/* Thread group tracking */
1435 u32 parent_exec_id;
1436 u32 self_exec_id;
58568d2a
MX
1437/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1438 * mempolicy */
1da177e4 1439 spinlock_t alloc_lock;
1da177e4 1440
b29739f9 1441 /* Protection of the PI data structures: */
1d615482 1442 raw_spinlock_t pi_lock;
b29739f9 1443
23f78d4a
IM
1444#ifdef CONFIG_RT_MUTEXES
1445 /* PI waiters blocked on a rt_mutex held by this task */
fb00aca4
PZ
1446 struct rb_root pi_waiters;
1447 struct rb_node *pi_waiters_leftmost;
23f78d4a
IM
1448 /* Deadlock detection and priority inheritance handling */
1449 struct rt_mutex_waiter *pi_blocked_on;
2d3d891d
DF
1450 /* Top pi_waiters task */
1451 struct task_struct *pi_top_task;
23f78d4a
IM
1452#endif
1453
408894ee
IM
1454#ifdef CONFIG_DEBUG_MUTEXES
1455 /* mutex deadlock detection */
1456 struct mutex_waiter *blocked_on;
1457#endif
de30a2b3
IM
1458#ifdef CONFIG_TRACE_IRQFLAGS
1459 unsigned int irq_events;
de30a2b3 1460 unsigned long hardirq_enable_ip;
de30a2b3 1461 unsigned long hardirq_disable_ip;
fa1452e8 1462 unsigned int hardirq_enable_event;
de30a2b3 1463 unsigned int hardirq_disable_event;
fa1452e8
HS
1464 int hardirqs_enabled;
1465 int hardirq_context;
de30a2b3 1466 unsigned long softirq_disable_ip;
de30a2b3 1467 unsigned long softirq_enable_ip;
fa1452e8 1468 unsigned int softirq_disable_event;
de30a2b3 1469 unsigned int softirq_enable_event;
fa1452e8 1470 int softirqs_enabled;
de30a2b3
IM
1471 int softirq_context;
1472#endif
fbb9ce95 1473#ifdef CONFIG_LOCKDEP
bdb9441e 1474# define MAX_LOCK_DEPTH 48UL
fbb9ce95
IM
1475 u64 curr_chain_key;
1476 int lockdep_depth;
fbb9ce95 1477 unsigned int lockdep_recursion;
c7aceaba 1478 struct held_lock held_locks[MAX_LOCK_DEPTH];
cf40bd16 1479 gfp_t lockdep_reclaim_gfp;
fbb9ce95 1480#endif
408894ee 1481
1da177e4
LT
1482/* journalling filesystem info */
1483 void *journal_info;
1484
d89d8796 1485/* stacked block device info */
bddd87c7 1486 struct bio_list *bio_list;
d89d8796 1487
73c10101
JA
1488#ifdef CONFIG_BLOCK
1489/* stack plugging */
1490 struct blk_plug *plug;
1491#endif
1492
1da177e4
LT
1493/* VM state */
1494 struct reclaim_state *reclaim_state;
1495
1da177e4
LT
1496 struct backing_dev_info *backing_dev_info;
1497
1498 struct io_context *io_context;
1499
1500 unsigned long ptrace_message;
1501 siginfo_t *last_siginfo; /* For ptrace use. */
7c3ab738 1502 struct task_io_accounting ioac;
8f0ab514 1503#if defined(CONFIG_TASK_XACCT)
1da177e4
LT
1504 u64 acct_rss_mem1; /* accumulated rss usage */
1505 u64 acct_vm_mem1; /* accumulated virtual memory usage */
49b5cf34 1506 cputime_t acct_timexpd; /* stime + utime since last update */
1da177e4
LT
1507#endif
1508#ifdef CONFIG_CPUSETS
58568d2a 1509 nodemask_t mems_allowed; /* Protected by alloc_lock */
cc9a6c87 1510 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
825a46af 1511 int cpuset_mem_spread_rotor;
6adef3eb 1512 int cpuset_slab_spread_rotor;
1da177e4 1513#endif
ddbcc7e8 1514#ifdef CONFIG_CGROUPS
817929ec 1515 /* Control Group info protected by css_set_lock */
2c392b8c 1516 struct css_set __rcu *cgroups;
817929ec
PM
1517 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1518 struct list_head cg_list;
ddbcc7e8 1519#endif
42b2dd0a 1520#ifdef CONFIG_FUTEX
0771dfef 1521 struct robust_list_head __user *robust_list;
34f192c6
IM
1522#ifdef CONFIG_COMPAT
1523 struct compat_robust_list_head __user *compat_robust_list;
1524#endif
c87e2837
IM
1525 struct list_head pi_state_list;
1526 struct futex_pi_state *pi_state_cache;
c7aceaba 1527#endif
cdd6c482 1528#ifdef CONFIG_PERF_EVENTS
8dc85d54 1529 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
cdd6c482
IM
1530 struct mutex perf_event_mutex;
1531 struct list_head perf_event_list;
a63eaf34 1532#endif
8f47b187
TG
1533#ifdef CONFIG_DEBUG_PREEMPT
1534 unsigned long preempt_disable_ip;
1535#endif
c7aceaba 1536#ifdef CONFIG_NUMA
58568d2a 1537 struct mempolicy *mempolicy; /* Protected by alloc_lock */
c7aceaba 1538 short il_next;
207205a2 1539 short pref_node_fork;
42b2dd0a 1540#endif
cbee9f88
PZ
1541#ifdef CONFIG_NUMA_BALANCING
1542 int numa_scan_seq;
cbee9f88 1543 unsigned int numa_scan_period;
598f0ec0 1544 unsigned int numa_scan_period_max;
de1c9ce6 1545 int numa_preferred_nid;
6b9a7460 1546 unsigned long numa_migrate_retry;
cbee9f88 1547 u64 node_stamp; /* migration stamp */
7e2703e6
RR
1548 u64 last_task_numa_placement;
1549 u64 last_sum_exec_runtime;
cbee9f88 1550 struct callback_head numa_work;
f809ca9a 1551
8c8a743c
PZ
1552 struct list_head numa_entry;
1553 struct numa_group *numa_group;
1554
745d6147
MG
1555 /*
1556 * Exponential decaying average of faults on a per-node basis.
1557 * Scheduling placement decisions are made based on the these counts.
1558 * The values remain static for the duration of a PTE scan
1559 */
ff1df896 1560 unsigned long *numa_faults_memory;
83e1d2cd 1561 unsigned long total_numa_faults;
745d6147
MG
1562
1563 /*
1564 * numa_faults_buffer records faults per node during the current
ff1df896
RR
1565 * scan window. When the scan completes, the counts in
1566 * numa_faults_memory decay and these values are copied.
745d6147 1567 */
ff1df896 1568 unsigned long *numa_faults_buffer_memory;
745d6147 1569
50ec8a40
RR
1570 /*
1571 * Track the nodes the process was running on when a NUMA hinting
1572 * fault was incurred.
1573 */
1574 unsigned long *numa_faults_cpu;
1575 unsigned long *numa_faults_buffer_cpu;
1576
04bb2f94
RR
1577 /*
1578 * numa_faults_locality tracks if faults recorded during the last
1579 * scan window were remote/local. The task scan period is adapted
1580 * based on the locality of the faults with different weights
1581 * depending on whether they were shared or private faults
1582 */
1583 unsigned long numa_faults_locality[2];
1584
b32e86b4 1585 unsigned long numa_pages_migrated;
cbee9f88
PZ
1586#endif /* CONFIG_NUMA_BALANCING */
1587
e56d0903 1588 struct rcu_head rcu;
b92ce558
JA
1589
1590 /*
1591 * cache last used pipe for splice
1592 */
1593 struct pipe_inode_info *splice_pipe;
5640f768
ED
1594
1595 struct page_frag task_frag;
1596
ca74e92b
SN
1597#ifdef CONFIG_TASK_DELAY_ACCT
1598 struct task_delay_info *delays;
f4f154fd
AM
1599#endif
1600#ifdef CONFIG_FAULT_INJECTION
1601 int make_it_fail;
ca74e92b 1602#endif
9d823e8f
WF
1603 /*
1604 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1605 * balance_dirty_pages() for some dirty throttling pause
1606 */
1607 int nr_dirtied;
1608 int nr_dirtied_pause;
83712358 1609 unsigned long dirty_paused_when; /* start of a write-and-pause period */
9d823e8f 1610
9745512c
AV
1611#ifdef CONFIG_LATENCYTOP
1612 int latency_record_count;
1613 struct latency_record latency_record[LT_SAVECOUNT];
1614#endif
6976675d
AV
1615 /*
1616 * time slack values; these are used to round up poll() and
1617 * select() etc timeout values. These are in nanoseconds.
1618 */
1619 unsigned long timer_slack_ns;
1620 unsigned long default_timer_slack_ns;
f8d570a4 1621
fb52607a 1622#ifdef CONFIG_FUNCTION_GRAPH_TRACER
3ad2f3fb 1623 /* Index of current stored address in ret_stack */
f201ae23
FW
1624 int curr_ret_stack;
1625 /* Stack of return addresses for return function tracing */
1626 struct ftrace_ret_stack *ret_stack;
8aef2d28
SR
1627 /* time stamp for last schedule */
1628 unsigned long long ftrace_timestamp;
f201ae23
FW
1629 /*
1630 * Number of functions that haven't been traced
1631 * because of depth overrun.
1632 */
1633 atomic_t trace_overrun;
380c4b14
FW
1634 /* Pause for the tracing */
1635 atomic_t tracing_graph_pause;
f201ae23 1636#endif
ea4e2bc4
SR
1637#ifdef CONFIG_TRACING
1638 /* state flags for use by tracers */
1639 unsigned long trace;
b1cff0ad 1640 /* bitmask and counter of trace recursion */
261842b7
SR
1641 unsigned long trace_recursion;
1642#endif /* CONFIG_TRACING */
c255a458 1643#ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
569b846d
KH
1644 struct memcg_batch_info {
1645 int do_batch; /* incremented when batch uncharge started */
1646 struct mem_cgroup *memcg; /* target memcg of uncharge */
7ffd4ca7
JW
1647 unsigned long nr_pages; /* uncharged usage */
1648 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
569b846d 1649 } memcg_batch;
0e9d92f2 1650 unsigned int memcg_kmem_skip_account;
519e5247 1651 struct memcg_oom_info {
49426420
JW
1652 struct mem_cgroup *memcg;
1653 gfp_t gfp_mask;
1654 int order;
519e5247
JW
1655 unsigned int may_oom:1;
1656 } memcg_oom;
569b846d 1657#endif
0326f5a9
SD
1658#ifdef CONFIG_UPROBES
1659 struct uprobe_task *utask;
0326f5a9 1660#endif
cafe5635
KO
1661#if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1662 unsigned int sequential_io;
1663 unsigned int sequential_io_avg;
1664#endif
1da177e4
LT
1665};
1666
76e6eee0 1667/* Future-safe accessor for struct task_struct's cpus_allowed. */
a4636818 1668#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
76e6eee0 1669
6688cc05
PZ
1670#define TNF_MIGRATED 0x01
1671#define TNF_NO_GROUP 0x02
dabe1d99 1672#define TNF_SHARED 0x04
04bb2f94 1673#define TNF_FAULT_LOCAL 0x08
6688cc05 1674
cbee9f88 1675#ifdef CONFIG_NUMA_BALANCING
6688cc05 1676extern void task_numa_fault(int last_node, int node, int pages, int flags);
e29cf08b 1677extern pid_t task_numa_group_id(struct task_struct *p);
1a687c2e 1678extern void set_numabalancing_state(bool enabled);
82727018 1679extern void task_numa_free(struct task_struct *p);
10f39042
RR
1680extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1681 int src_nid, int dst_cpu);
cbee9f88 1682#else
ac8e895b 1683static inline void task_numa_fault(int last_node, int node, int pages,
6688cc05 1684 int flags)
cbee9f88
PZ
1685{
1686}
e29cf08b
MG
1687static inline pid_t task_numa_group_id(struct task_struct *p)
1688{
1689 return 0;
1690}
1a687c2e
MG
1691static inline void set_numabalancing_state(bool enabled)
1692{
1693}
82727018
RR
1694static inline void task_numa_free(struct task_struct *p)
1695{
1696}
10f39042
RR
1697static inline bool should_numa_migrate_memory(struct task_struct *p,
1698 struct page *page, int src_nid, int dst_cpu)
1699{
1700 return true;
1701}
cbee9f88
PZ
1702#endif
1703
e868171a 1704static inline struct pid *task_pid(struct task_struct *task)
22c935f4
EB
1705{
1706 return task->pids[PIDTYPE_PID].pid;
1707}
1708
e868171a 1709static inline struct pid *task_tgid(struct task_struct *task)
22c935f4
EB
1710{
1711 return task->group_leader->pids[PIDTYPE_PID].pid;
1712}
1713
6dda81f4
ON
1714/*
1715 * Without tasklist or rcu lock it is not safe to dereference
1716 * the result of task_pgrp/task_session even if task == current,
1717 * we can race with another thread doing sys_setsid/sys_setpgid.
1718 */
e868171a 1719static inline struct pid *task_pgrp(struct task_struct *task)
22c935f4
EB
1720{
1721 return task->group_leader->pids[PIDTYPE_PGID].pid;
1722}
1723
e868171a 1724static inline struct pid *task_session(struct task_struct *task)
22c935f4
EB
1725{
1726 return task->group_leader->pids[PIDTYPE_SID].pid;
1727}
1728
7af57294
PE
1729struct pid_namespace;
1730
1731/*
1732 * the helpers to get the task's different pids as they are seen
1733 * from various namespaces
1734 *
1735 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
44c4e1b2
EB
1736 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1737 * current.
7af57294
PE
1738 * task_xid_nr_ns() : id seen from the ns specified;
1739 *
1740 * set_task_vxid() : assigns a virtual id to a task;
1741 *
7af57294
PE
1742 * see also pid_nr() etc in include/linux/pid.h
1743 */
52ee2dfd
ON
1744pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1745 struct pid_namespace *ns);
7af57294 1746
e868171a 1747static inline pid_t task_pid_nr(struct task_struct *tsk)
7af57294
PE
1748{
1749 return tsk->pid;
1750}
1751
52ee2dfd
ON
1752static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1753 struct pid_namespace *ns)
1754{
1755 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1756}
7af57294
PE
1757
1758static inline pid_t task_pid_vnr(struct task_struct *tsk)
1759{
52ee2dfd 1760 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
7af57294
PE
1761}
1762
1763
e868171a 1764static inline pid_t task_tgid_nr(struct task_struct *tsk)
7af57294
PE
1765{
1766 return tsk->tgid;
1767}
1768
2f2a3a46 1769pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
7af57294
PE
1770
1771static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1772{
1773 return pid_vnr(task_tgid(tsk));
1774}
1775
1776
80e0b6e8 1777static inline int pid_alive(const struct task_struct *p);
ad36d282
RGB
1778static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1779{
1780 pid_t pid = 0;
1781
1782 rcu_read_lock();
1783 if (pid_alive(tsk))
1784 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1785 rcu_read_unlock();
1786
1787 return pid;
1788}
1789
1790static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1791{
1792 return task_ppid_nr_ns(tsk, &init_pid_ns);
1793}
1794
52ee2dfd
ON
1795static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1796 struct pid_namespace *ns)
7af57294 1797{
52ee2dfd 1798 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
7af57294
PE
1799}
1800
7af57294
PE
1801static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1802{
52ee2dfd 1803 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
7af57294
PE
1804}
1805
1806
52ee2dfd
ON
1807static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1808 struct pid_namespace *ns)
7af57294 1809{
52ee2dfd 1810 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
7af57294
PE
1811}
1812
7af57294
PE
1813static inline pid_t task_session_vnr(struct task_struct *tsk)
1814{
52ee2dfd 1815 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
7af57294
PE
1816}
1817
1b0f7ffd
ON
1818/* obsolete, do not use */
1819static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1820{
1821 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1822}
7af57294 1823
1da177e4
LT
1824/**
1825 * pid_alive - check that a task structure is not stale
1826 * @p: Task structure to be checked.
1827 *
1828 * Test if a process is not yet dead (at most zombie state)
1829 * If pid_alive fails, then pointers within the task structure
1830 * can be stale and must not be dereferenced.
e69f6186
YB
1831 *
1832 * Return: 1 if the process is alive. 0 otherwise.
1da177e4 1833 */
ad36d282 1834static inline int pid_alive(const struct task_struct *p)
1da177e4 1835{
92476d7f 1836 return p->pids[PIDTYPE_PID].pid != NULL;
1da177e4
LT
1837}
1838
f400e198 1839/**
b460cbc5 1840 * is_global_init - check if a task structure is init
3260259f
H
1841 * @tsk: Task structure to be checked.
1842 *
1843 * Check if a task structure is the first user space task the kernel created.
e69f6186
YB
1844 *
1845 * Return: 1 if the task structure is init. 0 otherwise.
b460cbc5 1846 */
e868171a 1847static inline int is_global_init(struct task_struct *tsk)
b461cc03
PE
1848{
1849 return tsk->pid == 1;
1850}
b460cbc5 1851
9ec52099
CLG
1852extern struct pid *cad_pid;
1853
1da177e4 1854extern void free_task(struct task_struct *tsk);
1da177e4 1855#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
e56d0903 1856
158d9ebd 1857extern void __put_task_struct(struct task_struct *t);
e56d0903
IM
1858
1859static inline void put_task_struct(struct task_struct *t)
1860{
1861 if (atomic_dec_and_test(&t->usage))
8c7904a0 1862 __put_task_struct(t);
e56d0903 1863}
1da177e4 1864
6a61671b
FW
1865#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1866extern void task_cputime(struct task_struct *t,
1867 cputime_t *utime, cputime_t *stime);
1868extern void task_cputime_scaled(struct task_struct *t,
1869 cputime_t *utimescaled, cputime_t *stimescaled);
1870extern cputime_t task_gtime(struct task_struct *t);
1871#else
6fac4829
FW
1872static inline void task_cputime(struct task_struct *t,
1873 cputime_t *utime, cputime_t *stime)
1874{
1875 if (utime)
1876 *utime = t->utime;
1877 if (stime)
1878 *stime = t->stime;
1879}
1880
1881static inline void task_cputime_scaled(struct task_struct *t,
1882 cputime_t *utimescaled,
1883 cputime_t *stimescaled)
1884{
1885 if (utimescaled)
1886 *utimescaled = t->utimescaled;
1887 if (stimescaled)
1888 *stimescaled = t->stimescaled;
1889}
6a61671b
FW
1890
1891static inline cputime_t task_gtime(struct task_struct *t)
1892{
1893 return t->gtime;
1894}
1895#endif
e80d0a1a
FW
1896extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1897extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
49048622 1898
1da177e4
LT
1899/*
1900 * Per process flags
1901 */
1da177e4 1902#define PF_EXITING 0x00000004 /* getting shut down */
778e9a9c 1903#define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
94886b84 1904#define PF_VCPU 0x00000010 /* I'm a virtual CPU */
21aa9af0 1905#define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1da177e4 1906#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
4db96cf0 1907#define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1da177e4
LT
1908#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1909#define PF_DUMPCORE 0x00000200 /* dumped core */
1910#define PF_SIGNALED 0x00000400 /* killed by a signal */
1911#define PF_MEMALLOC 0x00000800 /* Allocating memory */
72fa5997 1912#define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1da177e4 1913#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
774a1221 1914#define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1da177e4
LT
1915#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1916#define PF_FROZEN 0x00010000 /* frozen for system suspend */
1917#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1918#define PF_KSWAPD 0x00040000 /* I am kswapd */
21caf2fc 1919#define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1da177e4 1920#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
246bb0b1 1921#define PF_KTHREAD 0x00200000 /* I am a kernel thread */
b31dc66a
JA
1922#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1923#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1924#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1925#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
14a40ffc 1926#define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
4db96cf0 1927#define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
61a87122 1928#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
58a69cb4 1929#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2b44c4db 1930#define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1da177e4
LT
1931
1932/*
1933 * Only the _current_ task can read/write to tsk->flags, but other
1934 * tasks can access tsk->flags in readonly mode for example
1935 * with tsk_used_math (like during threaded core dumping).
1936 * There is however an exception to this rule during ptrace
1937 * or during fork: the ptracer task is allowed to write to the
1938 * child->flags of its traced child (same goes for fork, the parent
1939 * can write to the child->flags), because we're guaranteed the
1940 * child is not running and in turn not changing child->flags
1941 * at the same time the parent does it.
1942 */
1943#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1944#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1945#define clear_used_math() clear_stopped_child_used_math(current)
1946#define set_used_math() set_stopped_child_used_math(current)
1947#define conditional_stopped_child_used_math(condition, child) \
1948 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1949#define conditional_used_math(condition) \
1950 conditional_stopped_child_used_math(condition, current)
1951#define copy_to_stopped_child_used_math(child) \
1952 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1953/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1954#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1955#define used_math() tsk_used_math(current)
1956
21caf2fc
ML
1957/* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1958static inline gfp_t memalloc_noio_flags(gfp_t flags)
1959{
1960 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1961 flags &= ~__GFP_IO;
1962 return flags;
1963}
1964
1965static inline unsigned int memalloc_noio_save(void)
1966{
1967 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1968 current->flags |= PF_MEMALLOC_NOIO;
1969 return flags;
1970}
1971
1972static inline void memalloc_noio_restore(unsigned int flags)
1973{
1974 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1975}
1976
e5c1902e 1977/*
a8f072c1 1978 * task->jobctl flags
e5c1902e 1979 */
a8f072c1 1980#define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
e5c1902e 1981
a8f072c1
TH
1982#define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1983#define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1984#define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
73ddff2b 1985#define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
fb1d910c 1986#define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
a8f072c1 1987#define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
544b2c91 1988#define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
a8f072c1
TH
1989
1990#define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1991#define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1992#define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
73ddff2b 1993#define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
fb1d910c 1994#define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
a8f072c1 1995#define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
544b2c91 1996#define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
a8f072c1 1997
fb1d910c 1998#define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
73ddff2b 1999#define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
3759a0d9 2000
7dd3db54
TH
2001extern bool task_set_jobctl_pending(struct task_struct *task,
2002 unsigned int mask);
73ddff2b 2003extern void task_clear_jobctl_trapping(struct task_struct *task);
3759a0d9
TH
2004extern void task_clear_jobctl_pending(struct task_struct *task,
2005 unsigned int mask);
39efa3ef 2006
a57eb940 2007#ifdef CONFIG_PREEMPT_RCU
f41d911f
PM
2008
2009#define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1aa03f11 2010#define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
f41d911f
PM
2011
2012static inline void rcu_copy_process(struct task_struct *p)
2013{
2014 p->rcu_read_lock_nesting = 0;
2015 p->rcu_read_unlock_special = 0;
a57eb940 2016#ifdef CONFIG_TREE_PREEMPT_RCU
dd5d19ba 2017 p->rcu_blocked_node = NULL;
24278d14
PM
2018#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
2019#ifdef CONFIG_RCU_BOOST
2020 p->rcu_boost_mutex = NULL;
2021#endif /* #ifdef CONFIG_RCU_BOOST */
f41d911f
PM
2022 INIT_LIST_HEAD(&p->rcu_node_entry);
2023}
2024
f41d911f
PM
2025#else
2026
2027static inline void rcu_copy_process(struct task_struct *p)
2028{
2029}
2030
2031#endif
2032
907aed48
MG
2033static inline void tsk_restore_flags(struct task_struct *task,
2034 unsigned long orig_flags, unsigned long flags)
2035{
2036 task->flags &= ~flags;
2037 task->flags |= orig_flags & flags;
2038}
2039
1da177e4 2040#ifdef CONFIG_SMP
1e1b6c51
KM
2041extern void do_set_cpus_allowed(struct task_struct *p,
2042 const struct cpumask *new_mask);
2043
cd8ba7cd 2044extern int set_cpus_allowed_ptr(struct task_struct *p,
96f874e2 2045 const struct cpumask *new_mask);
1da177e4 2046#else
1e1b6c51
KM
2047static inline void do_set_cpus_allowed(struct task_struct *p,
2048 const struct cpumask *new_mask)
2049{
2050}
cd8ba7cd 2051static inline int set_cpus_allowed_ptr(struct task_struct *p,
96f874e2 2052 const struct cpumask *new_mask)
1da177e4 2053{
96f874e2 2054 if (!cpumask_test_cpu(0, new_mask))
1da177e4
LT
2055 return -EINVAL;
2056 return 0;
2057}
2058#endif
e0ad9556 2059
3451d024 2060#ifdef CONFIG_NO_HZ_COMMON
5167e8d5
PZ
2061void calc_load_enter_idle(void);
2062void calc_load_exit_idle(void);
2063#else
2064static inline void calc_load_enter_idle(void) { }
2065static inline void calc_load_exit_idle(void) { }
3451d024 2066#endif /* CONFIG_NO_HZ_COMMON */
5167e8d5 2067
e0ad9556 2068#ifndef CONFIG_CPUMASK_OFFSTACK
cd8ba7cd
MT
2069static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2070{
2071 return set_cpus_allowed_ptr(p, &new_mask);
2072}
e0ad9556 2073#endif
1da177e4 2074
b342501c 2075/*
c676329a
PZ
2076 * Do not use outside of architecture code which knows its limitations.
2077 *
2078 * sched_clock() has no promise of monotonicity or bounded drift between
2079 * CPUs, use (which you should not) requires disabling IRQs.
2080 *
2081 * Please use one of the three interfaces below.
b342501c 2082 */
1bbfa6f2 2083extern unsigned long long notrace sched_clock(void);
c676329a 2084/*
489a71b0 2085 * See the comment in kernel/sched/clock.c
c676329a
PZ
2086 */
2087extern u64 cpu_clock(int cpu);
2088extern u64 local_clock(void);
2089extern u64 sched_clock_cpu(int cpu);
2090
e436d800 2091
c1955a3d 2092extern void sched_clock_init(void);
3e51f33f 2093
c1955a3d 2094#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
3e51f33f
PZ
2095static inline void sched_clock_tick(void)
2096{
2097}
2098
2099static inline void sched_clock_idle_sleep_event(void)
2100{
2101}
2102
2103static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2104{
2105}
2106#else
c676329a
PZ
2107/*
2108 * Architectures can set this to 1 if they have specified
2109 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2110 * but then during bootup it turns out that sched_clock()
2111 * is reliable after all:
2112 */
35af99e6
PZ
2113extern int sched_clock_stable(void);
2114extern void set_sched_clock_stable(void);
2115extern void clear_sched_clock_stable(void);
c676329a 2116
3e51f33f
PZ
2117extern void sched_clock_tick(void);
2118extern void sched_clock_idle_sleep_event(void);
2119extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2120#endif
2121
b52bfee4
VP
2122#ifdef CONFIG_IRQ_TIME_ACCOUNTING
2123/*
2124 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2125 * The reason for this explicit opt-in is not to have perf penalty with
2126 * slow sched_clocks.
2127 */
2128extern void enable_sched_clock_irqtime(void);
2129extern void disable_sched_clock_irqtime(void);
2130#else
2131static inline void enable_sched_clock_irqtime(void) {}
2132static inline void disable_sched_clock_irqtime(void) {}
2133#endif
2134
36c8b586 2135extern unsigned long long
41b86e9c 2136task_sched_runtime(struct task_struct *task);
1da177e4
LT
2137
2138/* sched_exec is called by processes performing an exec */
2139#ifdef CONFIG_SMP
2140extern void sched_exec(void);
2141#else
2142#define sched_exec() {}
2143#endif
2144
2aa44d05
IM
2145extern void sched_clock_idle_sleep_event(void);
2146extern void sched_clock_idle_wakeup_event(u64 delta_ns);
bb29ab26 2147
1da177e4
LT
2148#ifdef CONFIG_HOTPLUG_CPU
2149extern void idle_task_exit(void);
2150#else
2151static inline void idle_task_exit(void) {}
2152#endif
2153
3451d024 2154#if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
1c20091e 2155extern void wake_up_nohz_cpu(int cpu);
06d8308c 2156#else
1c20091e 2157static inline void wake_up_nohz_cpu(int cpu) { }
06d8308c
TG
2158#endif
2159
ce831b38
FW
2160#ifdef CONFIG_NO_HZ_FULL
2161extern bool sched_can_stop_tick(void);
265f22a9 2162extern u64 scheduler_tick_max_deferment(void);
ce831b38
FW
2163#else
2164static inline bool sched_can_stop_tick(void) { return false; }
06d8308c
TG
2165#endif
2166
5091faa4 2167#ifdef CONFIG_SCHED_AUTOGROUP
5091faa4
MG
2168extern void sched_autogroup_create_attach(struct task_struct *p);
2169extern void sched_autogroup_detach(struct task_struct *p);
2170extern void sched_autogroup_fork(struct signal_struct *sig);
2171extern void sched_autogroup_exit(struct signal_struct *sig);
2172#ifdef CONFIG_PROC_FS
2173extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2e5b5b3a 2174extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
5091faa4
MG
2175#endif
2176#else
2177static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2178static inline void sched_autogroup_detach(struct task_struct *p) { }
2179static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2180static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2181#endif
2182
fa93384f 2183extern int yield_to(struct task_struct *p, bool preempt);
36c8b586
IM
2184extern void set_user_nice(struct task_struct *p, long nice);
2185extern int task_prio(const struct task_struct *p);
d0ea0268
DY
2186/**
2187 * task_nice - return the nice value of a given task.
2188 * @p: the task in question.
2189 *
2190 * Return: The nice value [ -20 ... 0 ... 19 ].
2191 */
2192static inline int task_nice(const struct task_struct *p)
2193{
2194 return PRIO_TO_NICE((p)->static_prio);
2195}
36c8b586
IM
2196extern int can_nice(const struct task_struct *p, const int nice);
2197extern int task_curr(const struct task_struct *p);
1da177e4 2198extern int idle_cpu(int cpu);
fe7de49f
KM
2199extern int sched_setscheduler(struct task_struct *, int,
2200 const struct sched_param *);
961ccddd 2201extern int sched_setscheduler_nocheck(struct task_struct *, int,
fe7de49f 2202 const struct sched_param *);
d50dde5a
DF
2203extern int sched_setattr(struct task_struct *,
2204 const struct sched_attr *);
36c8b586 2205extern struct task_struct *idle_task(int cpu);
c4f30608
PM
2206/**
2207 * is_idle_task - is the specified task an idle task?
fa757281 2208 * @p: the task in question.
e69f6186
YB
2209 *
2210 * Return: 1 if @p is an idle task. 0 otherwise.
c4f30608 2211 */
7061ca3b 2212static inline bool is_idle_task(const struct task_struct *p)
c4f30608
PM
2213{
2214 return p->pid == 0;
2215}
36c8b586
IM
2216extern struct task_struct *curr_task(int cpu);
2217extern void set_curr_task(int cpu, struct task_struct *p);
1da177e4
LT
2218
2219void yield(void);
2220
2221/*
2222 * The default (Linux) execution domain.
2223 */
2224extern struct exec_domain default_exec_domain;
2225
2226union thread_union {
2227 struct thread_info thread_info;
2228 unsigned long stack[THREAD_SIZE/sizeof(long)];
2229};
2230
2231#ifndef __HAVE_ARCH_KSTACK_END
2232static inline int kstack_end(void *addr)
2233{
2234 /* Reliable end of stack detection:
2235 * Some APM bios versions misalign the stack
2236 */
2237 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2238}
2239#endif
2240
2241extern union thread_union init_thread_union;
2242extern struct task_struct init_task;
2243
2244extern struct mm_struct init_mm;
2245
198fe21b
PE
2246extern struct pid_namespace init_pid_ns;
2247
2248/*
2249 * find a task by one of its numerical ids
2250 *
198fe21b
PE
2251 * find_task_by_pid_ns():
2252 * finds a task by its pid in the specified namespace
228ebcbe
PE
2253 * find_task_by_vpid():
2254 * finds a task by its virtual pid
198fe21b 2255 *
e49859e7 2256 * see also find_vpid() etc in include/linux/pid.h
198fe21b
PE
2257 */
2258
228ebcbe
PE
2259extern struct task_struct *find_task_by_vpid(pid_t nr);
2260extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2261 struct pid_namespace *ns);
198fe21b 2262
1da177e4 2263/* per-UID process charging. */
7b44ab97 2264extern struct user_struct * alloc_uid(kuid_t);
1da177e4
LT
2265static inline struct user_struct *get_uid(struct user_struct *u)
2266{
2267 atomic_inc(&u->__count);
2268 return u;
2269}
2270extern void free_uid(struct user_struct *);
1da177e4
LT
2271
2272#include <asm/current.h>
2273
f0af911a 2274extern void xtime_update(unsigned long ticks);
1da177e4 2275
b3c97528
HH
2276extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2277extern int wake_up_process(struct task_struct *tsk);
3e51e3ed 2278extern void wake_up_new_task(struct task_struct *tsk);
1da177e4
LT
2279#ifdef CONFIG_SMP
2280 extern void kick_process(struct task_struct *tsk);
2281#else
2282 static inline void kick_process(struct task_struct *tsk) { }
2283#endif
aab03e05 2284extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
ad46c2c4 2285extern void sched_dead(struct task_struct *p);
1da177e4 2286
1da177e4
LT
2287extern void proc_caches_init(void);
2288extern void flush_signals(struct task_struct *);
3bcac026 2289extern void __flush_signals(struct task_struct *);
10ab825b 2290extern void ignore_signals(struct task_struct *);
1da177e4
LT
2291extern void flush_signal_handlers(struct task_struct *, int force_default);
2292extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2293
2294static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2295{
2296 unsigned long flags;
2297 int ret;
2298
2299 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2300 ret = dequeue_signal(tsk, mask, info);
2301 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2302
2303 return ret;
53c8f9f1 2304}
1da177e4
LT
2305
2306extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2307 sigset_t *mask);
2308extern void unblock_all_signals(void);
2309extern void release_task(struct task_struct * p);
2310extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1da177e4
LT
2311extern int force_sigsegv(int, struct task_struct *);
2312extern int force_sig_info(int, struct siginfo *, struct task_struct *);
c4b92fc1 2313extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
c4b92fc1 2314extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
d178bc3a
SH
2315extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2316 const struct cred *, u32);
c4b92fc1
EB
2317extern int kill_pgrp(struct pid *pid, int sig, int priv);
2318extern int kill_pid(struct pid *pid, int sig, int priv);
c3de4b38 2319extern int kill_proc_info(int, struct siginfo *, pid_t);
86773473 2320extern __must_check bool do_notify_parent(struct task_struct *, int);
a7f0765e 2321extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
1da177e4 2322extern void force_sig(int, struct task_struct *);
1da177e4 2323extern int send_sig(int, struct task_struct *, int);
09faef11 2324extern int zap_other_threads(struct task_struct *p);
1da177e4
LT
2325extern struct sigqueue *sigqueue_alloc(void);
2326extern void sigqueue_free(struct sigqueue *);
ac5c2153 2327extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
9ac95f2f 2328extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1da177e4 2329
51a7b448
AV
2330static inline void restore_saved_sigmask(void)
2331{
2332 if (test_and_clear_restore_sigmask())
77097ae5 2333 __set_current_blocked(&current->saved_sigmask);
51a7b448
AV
2334}
2335
b7f9a11a
AV
2336static inline sigset_t *sigmask_to_save(void)
2337{
2338 sigset_t *res = &current->blocked;
2339 if (unlikely(test_restore_sigmask()))
2340 res = &current->saved_sigmask;
2341 return res;
2342}
2343
9ec52099
CLG
2344static inline int kill_cad_pid(int sig, int priv)
2345{
2346 return kill_pid(cad_pid, sig, priv);
2347}
2348
1da177e4
LT
2349/* These can be the second arg to send_sig_info/send_group_sig_info. */
2350#define SEND_SIG_NOINFO ((struct siginfo *) 0)
2351#define SEND_SIG_PRIV ((struct siginfo *) 1)
2352#define SEND_SIG_FORCED ((struct siginfo *) 2)
2353
2a855dd0
SAS
2354/*
2355 * True if we are on the alternate signal stack.
2356 */
1da177e4
LT
2357static inline int on_sig_stack(unsigned long sp)
2358{
2a855dd0
SAS
2359#ifdef CONFIG_STACK_GROWSUP
2360 return sp >= current->sas_ss_sp &&
2361 sp - current->sas_ss_sp < current->sas_ss_size;
2362#else
2363 return sp > current->sas_ss_sp &&
2364 sp - current->sas_ss_sp <= current->sas_ss_size;
2365#endif
1da177e4
LT
2366}
2367
2368static inline int sas_ss_flags(unsigned long sp)
2369{
2370 return (current->sas_ss_size == 0 ? SS_DISABLE
2371 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2372}
2373
5a1b98d3
AV
2374static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2375{
2376 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2377#ifdef CONFIG_STACK_GROWSUP
2378 return current->sas_ss_sp;
2379#else
2380 return current->sas_ss_sp + current->sas_ss_size;
2381#endif
2382 return sp;
2383}
2384
1da177e4
LT
2385/*
2386 * Routines for handling mm_structs
2387 */
2388extern struct mm_struct * mm_alloc(void);
2389
2390/* mmdrop drops the mm and the page tables */
b3c97528 2391extern void __mmdrop(struct mm_struct *);
1da177e4
LT
2392static inline void mmdrop(struct mm_struct * mm)
2393{
6fb43d7b 2394 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1da177e4
LT
2395 __mmdrop(mm);
2396}
2397
2398/* mmput gets rid of the mappings and all user-space */
2399extern void mmput(struct mm_struct *);
2400/* Grab a reference to a task's mm, if it is not already going away */
2401extern struct mm_struct *get_task_mm(struct task_struct *task);
8cdb878d
CY
2402/*
2403 * Grab a reference to a task's mm, if it is not already going away
2404 * and ptrace_may_access with the mode parameter passed to it
2405 * succeeds.
2406 */
2407extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
1da177e4
LT
2408/* Remove the current tasks stale references to the old mm_struct */
2409extern void mm_release(struct task_struct *, struct mm_struct *);
2410
6f2c55b8 2411extern int copy_thread(unsigned long, unsigned long, unsigned long,
afa86fc4 2412 struct task_struct *);
1da177e4
LT
2413extern void flush_thread(void);
2414extern void exit_thread(void);
2415
1da177e4 2416extern void exit_files(struct task_struct *);
a7e5328a 2417extern void __cleanup_sighand(struct sighand_struct *);
cbaffba1 2418
1da177e4 2419extern void exit_itimers(struct signal_struct *);
cbaffba1 2420extern void flush_itimer_signals(void);
1da177e4 2421
9402c95f 2422extern void do_group_exit(int);
1da177e4 2423
1da177e4
LT
2424extern int allow_signal(int);
2425extern int disallow_signal(int);
1da177e4 2426
c4ad8f98 2427extern int do_execve(struct filename *,
d7627467 2428 const char __user * const __user *,
da3d4c5f 2429 const char __user * const __user *);
e80d6661 2430extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
36c8b586 2431struct task_struct *fork_idle(int);
2aa3a7f8 2432extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
1da177e4 2433
23aebe16 2434extern void set_task_comm(struct task_struct *tsk, const char *from);
59714d65 2435extern char *get_task_comm(char *to, struct task_struct *tsk);
1da177e4
LT
2436
2437#ifdef CONFIG_SMP
317f3941 2438void scheduler_ipi(void);
85ba2d86 2439extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1da177e4 2440#else
184748cc 2441static inline void scheduler_ipi(void) { }
85ba2d86
RM
2442static inline unsigned long wait_task_inactive(struct task_struct *p,
2443 long match_state)
2444{
2445 return 1;
2446}
1da177e4
LT
2447#endif
2448
05725f7e
JP
2449#define next_task(p) \
2450 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
1da177e4
LT
2451
2452#define for_each_process(p) \
2453 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2454
5bb459bb 2455extern bool current_is_single_threaded(void);
d84f4f99 2456
1da177e4
LT
2457/*
2458 * Careful: do_each_thread/while_each_thread is a double loop so
2459 * 'break' will not work as expected - use goto instead.
2460 */
2461#define do_each_thread(g, t) \
2462 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2463
2464#define while_each_thread(g, t) \
2465 while ((t = next_thread(t)) != g)
2466
0c740d0a
ON
2467#define __for_each_thread(signal, t) \
2468 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2469
2470#define for_each_thread(p, t) \
2471 __for_each_thread((p)->signal, t)
2472
2473/* Careful: this is a double loop, 'break' won't work as expected. */
2474#define for_each_process_thread(p, t) \
2475 for_each_process(p) for_each_thread(p, t)
2476
7e49827c
ON
2477static inline int get_nr_threads(struct task_struct *tsk)
2478{
b3ac022c 2479 return tsk->signal->nr_threads;
7e49827c
ON
2480}
2481
087806b1
ON
2482static inline bool thread_group_leader(struct task_struct *p)
2483{
2484 return p->exit_signal >= 0;
2485}
1da177e4 2486
0804ef4b
EB
2487/* Do to the insanities of de_thread it is possible for a process
2488 * to have the pid of the thread group leader without actually being
2489 * the thread group leader. For iteration through the pids in proc
2490 * all we care about is that we have a task with the appropriate
2491 * pid, we don't actually care if we have the right task.
2492 */
e1403b8e 2493static inline bool has_group_leader_pid(struct task_struct *p)
0804ef4b 2494{
e1403b8e 2495 return task_pid(p) == p->signal->leader_pid;
0804ef4b
EB
2496}
2497
bac0abd6 2498static inline
e1403b8e 2499bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
bac0abd6 2500{
e1403b8e 2501 return p1->signal == p2->signal;
bac0abd6
PE
2502}
2503
36c8b586 2504static inline struct task_struct *next_thread(const struct task_struct *p)
47e65328 2505{
05725f7e
JP
2506 return list_entry_rcu(p->thread_group.next,
2507 struct task_struct, thread_group);
47e65328
ON
2508}
2509
e868171a 2510static inline int thread_group_empty(struct task_struct *p)
1da177e4 2511{
47e65328 2512 return list_empty(&p->thread_group);
1da177e4
LT
2513}
2514
2515#define delay_group_leader(p) \
2516 (thread_group_leader(p) && !thread_group_empty(p))
2517
1da177e4 2518/*
260ea101 2519 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
22e2c507 2520 * subscriptions and synchronises with wait4(). Also used in procfs. Also
ddbcc7e8 2521 * pins the final release of task.io_context. Also protects ->cpuset and
d68b46fe 2522 * ->cgroup.subsys[]. And ->vfork_done.
1da177e4
LT
2523 *
2524 * Nests both inside and outside of read_lock(&tasklist_lock).
2525 * It must not be nested with write_lock_irq(&tasklist_lock),
2526 * neither inside nor outside.
2527 */
2528static inline void task_lock(struct task_struct *p)
2529{
2530 spin_lock(&p->alloc_lock);
2531}
2532
2533static inline void task_unlock(struct task_struct *p)
2534{
2535 spin_unlock(&p->alloc_lock);
2536}
2537
b8ed374e 2538extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
f63ee72e
ON
2539 unsigned long *flags);
2540
9388dc30
AV
2541static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2542 unsigned long *flags)
2543{
2544 struct sighand_struct *ret;
2545
2546 ret = __lock_task_sighand(tsk, flags);
2547 (void)__cond_lock(&tsk->sighand->siglock, ret);
2548 return ret;
2549}
b8ed374e 2550
f63ee72e
ON
2551static inline void unlock_task_sighand(struct task_struct *tsk,
2552 unsigned long *flags)
2553{
2554 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2555}
2556
4714d1d3 2557#ifdef CONFIG_CGROUPS
257058ae 2558static inline void threadgroup_change_begin(struct task_struct *tsk)
4714d1d3 2559{
257058ae 2560 down_read(&tsk->signal->group_rwsem);
4714d1d3 2561}
257058ae 2562static inline void threadgroup_change_end(struct task_struct *tsk)
4714d1d3 2563{
257058ae 2564 up_read(&tsk->signal->group_rwsem);
4714d1d3 2565}
77e4ef99
TH
2566
2567/**
2568 * threadgroup_lock - lock threadgroup
2569 * @tsk: member task of the threadgroup to lock
2570 *
2571 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2572 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
e56fb287
ON
2573 * change ->group_leader/pid. This is useful for cases where the threadgroup
2574 * needs to stay stable across blockable operations.
77e4ef99
TH
2575 *
2576 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2577 * synchronization. While held, no new task will be added to threadgroup
2578 * and no existing live task will have its PF_EXITING set.
2579 *
e56fb287
ON
2580 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2581 * sub-thread becomes a new leader.
77e4ef99 2582 */
257058ae 2583static inline void threadgroup_lock(struct task_struct *tsk)
4714d1d3 2584{
257058ae 2585 down_write(&tsk->signal->group_rwsem);
4714d1d3 2586}
77e4ef99
TH
2587
2588/**
2589 * threadgroup_unlock - unlock threadgroup
2590 * @tsk: member task of the threadgroup to unlock
2591 *
2592 * Reverse threadgroup_lock().
2593 */
257058ae 2594static inline void threadgroup_unlock(struct task_struct *tsk)
4714d1d3 2595{
257058ae 2596 up_write(&tsk->signal->group_rwsem);
4714d1d3
BB
2597}
2598#else
257058ae
TH
2599static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2600static inline void threadgroup_change_end(struct task_struct *tsk) {}
2601static inline void threadgroup_lock(struct task_struct *tsk) {}
2602static inline void threadgroup_unlock(struct task_struct *tsk) {}
4714d1d3
BB
2603#endif
2604
f037360f
AV
2605#ifndef __HAVE_THREAD_FUNCTIONS
2606
f7e4217b
RZ
2607#define task_thread_info(task) ((struct thread_info *)(task)->stack)
2608#define task_stack_page(task) ((task)->stack)
a1261f54 2609
10ebffde
AV
2610static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2611{
2612 *task_thread_info(p) = *task_thread_info(org);
2613 task_thread_info(p)->task = p;
2614}
2615
2616static inline unsigned long *end_of_stack(struct task_struct *p)
2617{
f7e4217b 2618 return (unsigned long *)(task_thread_info(p) + 1);
10ebffde
AV
2619}
2620
f037360f
AV
2621#endif
2622
8b05c7e6
FT
2623static inline int object_is_on_stack(void *obj)
2624{
2625 void *stack = task_stack_page(current);
2626
2627 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2628}
2629
8c9843e5
BH
2630extern void thread_info_cache_init(void);
2631
7c9f8861
ES
2632#ifdef CONFIG_DEBUG_STACK_USAGE
2633static inline unsigned long stack_not_used(struct task_struct *p)
2634{
2635 unsigned long *n = end_of_stack(p);
2636
2637 do { /* Skip over canary */
2638 n++;
2639 } while (!*n);
2640
2641 return (unsigned long)n - (unsigned long)end_of_stack(p);
2642}
2643#endif
2644
1da177e4
LT
2645/* set thread flags in other task's structures
2646 * - see asm/thread_info.h for TIF_xxxx flags available
2647 */
2648static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2649{
a1261f54 2650 set_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
2651}
2652
2653static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2654{
a1261f54 2655 clear_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
2656}
2657
2658static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2659{
a1261f54 2660 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
2661}
2662
2663static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2664{
a1261f54 2665 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
2666}
2667
2668static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2669{
a1261f54 2670 return test_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
2671}
2672
2673static inline void set_tsk_need_resched(struct task_struct *tsk)
2674{
2675 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2676}
2677
2678static inline void clear_tsk_need_resched(struct task_struct *tsk)
2679{
2680 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2681}
2682
8ae121ac
GH
2683static inline int test_tsk_need_resched(struct task_struct *tsk)
2684{
2685 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2686}
2687
690cc3ff
EB
2688static inline int restart_syscall(void)
2689{
2690 set_tsk_thread_flag(current, TIF_SIGPENDING);
2691 return -ERESTARTNOINTR;
2692}
2693
1da177e4
LT
2694static inline int signal_pending(struct task_struct *p)
2695{
2696 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2697}
f776d12d 2698
d9588725
RM
2699static inline int __fatal_signal_pending(struct task_struct *p)
2700{
2701 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2702}
f776d12d
MW
2703
2704static inline int fatal_signal_pending(struct task_struct *p)
2705{
2706 return signal_pending(p) && __fatal_signal_pending(p);
2707}
2708
16882c1e
ON
2709static inline int signal_pending_state(long state, struct task_struct *p)
2710{
2711 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2712 return 0;
2713 if (!signal_pending(p))
2714 return 0;
2715
16882c1e
ON
2716 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2717}
2718
1da177e4
LT
2719/*
2720 * cond_resched() and cond_resched_lock(): latency reduction via
2721 * explicit rescheduling in places that are safe. The return
2722 * value indicates whether a reschedule was done in fact.
2723 * cond_resched_lock() will drop the spinlock before scheduling,
2724 * cond_resched_softirq() will enable bhs before scheduling.
2725 */
c3921ab7 2726extern int _cond_resched(void);
6f80bd98 2727
613afbf8
FW
2728#define cond_resched() ({ \
2729 __might_sleep(__FILE__, __LINE__, 0); \
2730 _cond_resched(); \
2731})
6f80bd98 2732
613afbf8
FW
2733extern int __cond_resched_lock(spinlock_t *lock);
2734
bdd4e85d 2735#ifdef CONFIG_PREEMPT_COUNT
716a4234 2736#define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
02b67cc3 2737#else
716a4234 2738#define PREEMPT_LOCK_OFFSET 0
02b67cc3 2739#endif
716a4234 2740
613afbf8 2741#define cond_resched_lock(lock) ({ \
716a4234 2742 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
613afbf8
FW
2743 __cond_resched_lock(lock); \
2744})
2745
2746extern int __cond_resched_softirq(void);
2747
75e1056f
VP
2748#define cond_resched_softirq() ({ \
2749 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2750 __cond_resched_softirq(); \
613afbf8 2751})
1da177e4 2752
f6f3c437
SH
2753static inline void cond_resched_rcu(void)
2754{
2755#if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2756 rcu_read_unlock();
2757 cond_resched();
2758 rcu_read_lock();
2759#endif
2760}
2761
1da177e4
LT
2762/*
2763 * Does a critical section need to be broken due to another
95c354fe
NP
2764 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2765 * but a general need for low latency)
1da177e4 2766 */
95c354fe 2767static inline int spin_needbreak(spinlock_t *lock)
1da177e4 2768{
95c354fe
NP
2769#ifdef CONFIG_PREEMPT
2770 return spin_is_contended(lock);
2771#else
1da177e4 2772 return 0;
95c354fe 2773#endif
1da177e4
LT
2774}
2775
ee761f62
TG
2776/*
2777 * Idle thread specific functions to determine the need_resched
69dd0f84 2778 * polling state.
ee761f62 2779 */
69dd0f84 2780#ifdef TIF_POLLING_NRFLAG
ee761f62
TG
2781static inline int tsk_is_polling(struct task_struct *p)
2782{
2783 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2784}
ea811747
PZ
2785
2786static inline void __current_set_polling(void)
3a98f871
TG
2787{
2788 set_thread_flag(TIF_POLLING_NRFLAG);
2789}
2790
ea811747
PZ
2791static inline bool __must_check current_set_polling_and_test(void)
2792{
2793 __current_set_polling();
2794
2795 /*
2796 * Polling state must be visible before we test NEED_RESCHED,
2797 * paired by resched_task()
2798 *
2799 * XXX: assumes set/clear bit are identical barrier wise.
2800 */
2801 smp_mb__after_clear_bit();
2802
2803 return unlikely(tif_need_resched());
2804}
2805
2806static inline void __current_clr_polling(void)
3a98f871
TG
2807{
2808 clear_thread_flag(TIF_POLLING_NRFLAG);
2809}
ea811747
PZ
2810
2811static inline bool __must_check current_clr_polling_and_test(void)
2812{
2813 __current_clr_polling();
2814
2815 /*
2816 * Polling state must be visible before we test NEED_RESCHED,
2817 * paired by resched_task()
2818 */
2819 smp_mb__after_clear_bit();
2820
2821 return unlikely(tif_need_resched());
2822}
2823
ee761f62
TG
2824#else
2825static inline int tsk_is_polling(struct task_struct *p) { return 0; }
ea811747
PZ
2826static inline void __current_set_polling(void) { }
2827static inline void __current_clr_polling(void) { }
2828
2829static inline bool __must_check current_set_polling_and_test(void)
2830{
2831 return unlikely(tif_need_resched());
2832}
2833static inline bool __must_check current_clr_polling_and_test(void)
2834{
2835 return unlikely(tif_need_resched());
2836}
ee761f62
TG
2837#endif
2838
8cb75e0c
PZ
2839static inline void current_clr_polling(void)
2840{
2841 __current_clr_polling();
2842
2843 /*
2844 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2845 * Once the bit is cleared, we'll get IPIs with every new
2846 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2847 * fold.
2848 */
2849 smp_mb(); /* paired with resched_task() */
2850
2851 preempt_fold_need_resched();
2852}
2853
75f93fed
PZ
2854static __always_inline bool need_resched(void)
2855{
2856 return unlikely(tif_need_resched());
2857}
2858
f06febc9
FM
2859/*
2860 * Thread group CPU time accounting.
2861 */
4cd4c1b4 2862void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
4da94d49 2863void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
f06febc9 2864
490dea45 2865static inline void thread_group_cputime_init(struct signal_struct *sig)
f06febc9 2866{
ee30a7b2 2867 raw_spin_lock_init(&sig->cputimer.lock);
f06febc9
FM
2868}
2869
7bb44ade
RM
2870/*
2871 * Reevaluate whether the task has signals pending delivery.
2872 * Wake the task if so.
2873 * This is required every time the blocked sigset_t changes.
2874 * callers must hold sighand->siglock.
2875 */
2876extern void recalc_sigpending_and_wake(struct task_struct *t);
1da177e4
LT
2877extern void recalc_sigpending(void);
2878
910ffdb1
ON
2879extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2880
2881static inline void signal_wake_up(struct task_struct *t, bool resume)
2882{
2883 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2884}
2885static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2886{
2887 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2888}
1da177e4
LT
2889
2890/*
2891 * Wrappers for p->thread_info->cpu access. No-op on UP.
2892 */
2893#ifdef CONFIG_SMP
2894
2895static inline unsigned int task_cpu(const struct task_struct *p)
2896{
a1261f54 2897 return task_thread_info(p)->cpu;
1da177e4
LT
2898}
2899
b32e86b4
IM
2900static inline int task_node(const struct task_struct *p)
2901{
2902 return cpu_to_node(task_cpu(p));
2903}
2904
c65cc870 2905extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
1da177e4
LT
2906
2907#else
2908
2909static inline unsigned int task_cpu(const struct task_struct *p)
2910{
2911 return 0;
2912}
2913
2914static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2915{
2916}
2917
2918#endif /* CONFIG_SMP */
2919
96f874e2
RR
2920extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2921extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
5c45bf27 2922
7c941438 2923#ifdef CONFIG_CGROUP_SCHED
07e06b01 2924extern struct task_group root_task_group;
8323f26c 2925#endif /* CONFIG_CGROUP_SCHED */
9b5b7751 2926
54e99124
DG
2927extern int task_can_switch_user(struct user_struct *up,
2928 struct task_struct *tsk);
2929
4b98d11b
AD
2930#ifdef CONFIG_TASK_XACCT
2931static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2932{
940389b8 2933 tsk->ioac.rchar += amt;
4b98d11b
AD
2934}
2935
2936static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2937{
940389b8 2938 tsk->ioac.wchar += amt;
4b98d11b
AD
2939}
2940
2941static inline void inc_syscr(struct task_struct *tsk)
2942{
940389b8 2943 tsk->ioac.syscr++;
4b98d11b
AD
2944}
2945
2946static inline void inc_syscw(struct task_struct *tsk)
2947{
940389b8 2948 tsk->ioac.syscw++;
4b98d11b
AD
2949}
2950#else
2951static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2952{
2953}
2954
2955static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2956{
2957}
2958
2959static inline void inc_syscr(struct task_struct *tsk)
2960{
2961}
2962
2963static inline void inc_syscw(struct task_struct *tsk)
2964{
2965}
2966#endif
2967
82455257
DH
2968#ifndef TASK_SIZE_OF
2969#define TASK_SIZE_OF(tsk) TASK_SIZE
2970#endif
2971
cf475ad2
BS
2972#ifdef CONFIG_MM_OWNER
2973extern void mm_update_next_owner(struct mm_struct *mm);
2974extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2975#else
2976static inline void mm_update_next_owner(struct mm_struct *mm)
2977{
2978}
2979
2980static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2981{
2982}
2983#endif /* CONFIG_MM_OWNER */
2984
3e10e716
JS
2985static inline unsigned long task_rlimit(const struct task_struct *tsk,
2986 unsigned int limit)
2987{
2988 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2989}
2990
2991static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2992 unsigned int limit)
2993{
2994 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2995}
2996
2997static inline unsigned long rlimit(unsigned int limit)
2998{
2999 return task_rlimit(current, limit);
3000}
3001
3002static inline unsigned long rlimit_max(unsigned int limit)
3003{
3004 return task_rlimit_max(current, limit);
3005}
3006
1da177e4 3007#endif