init/main.c: Give init_task a canary
[deliverable/linux.git] / include / linux / sched.h
CommitLineData
1da177e4
LT
1#ifndef _LINUX_SCHED_H
2#define _LINUX_SCHED_H
3
607ca46e 4#include <uapi/linux/sched.h>
b7b3c76a 5
5c228079
DY
6#include <linux/sched/prio.h>
7
b7b3c76a
DW
8
9struct sched_param {
10 int sched_priority;
11};
12
1da177e4
LT
13#include <asm/param.h> /* for HZ */
14
1da177e4
LT
15#include <linux/capability.h>
16#include <linux/threads.h>
17#include <linux/kernel.h>
18#include <linux/types.h>
19#include <linux/timex.h>
20#include <linux/jiffies.h>
fb00aca4 21#include <linux/plist.h>
1da177e4
LT
22#include <linux/rbtree.h>
23#include <linux/thread_info.h>
24#include <linux/cpumask.h>
25#include <linux/errno.h>
26#include <linux/nodemask.h>
c92ff1bd 27#include <linux/mm_types.h>
00d1a39e 28#include <linux/preempt_mask.h>
1da177e4 29
1da177e4
LT
30#include <asm/page.h>
31#include <asm/ptrace.h>
bfc3f028 32#include <linux/cputime.h>
1da177e4
LT
33
34#include <linux/smp.h>
35#include <linux/sem.h>
ab602f79 36#include <linux/shm.h>
1da177e4 37#include <linux/signal.h>
1da177e4
LT
38#include <linux/compiler.h>
39#include <linux/completion.h>
40#include <linux/pid.h>
41#include <linux/percpu.h>
42#include <linux/topology.h>
3e26c149 43#include <linux/proportions.h>
1da177e4 44#include <linux/seccomp.h>
e56d0903 45#include <linux/rcupdate.h>
05725f7e 46#include <linux/rculist.h>
23f78d4a 47#include <linux/rtmutex.h>
1da177e4 48
a3b6714e
DW
49#include <linux/time.h>
50#include <linux/param.h>
51#include <linux/resource.h>
52#include <linux/timer.h>
53#include <linux/hrtimer.h>
7c3ab738 54#include <linux/task_io_accounting.h>
9745512c 55#include <linux/latencytop.h>
9e2b2dc4 56#include <linux/cred.h>
fa14ff4a 57#include <linux/llist.h>
7b44ab97 58#include <linux/uidgid.h>
21caf2fc 59#include <linux/gfp.h>
d4311ff1 60#include <linux/magic.h>
a3b6714e
DW
61
62#include <asm/processor.h>
36d57ac4 63
d50dde5a
DF
64#define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
65
66/*
67 * Extended scheduling parameters data structure.
68 *
69 * This is needed because the original struct sched_param can not be
70 * altered without introducing ABI issues with legacy applications
71 * (e.g., in sched_getparam()).
72 *
73 * However, the possibility of specifying more than just a priority for
74 * the tasks may be useful for a wide variety of application fields, e.g.,
75 * multimedia, streaming, automation and control, and many others.
76 *
77 * This variant (sched_attr) is meant at describing a so-called
78 * sporadic time-constrained task. In such model a task is specified by:
79 * - the activation period or minimum instance inter-arrival time;
80 * - the maximum (or average, depending on the actual scheduling
81 * discipline) computation time of all instances, a.k.a. runtime;
82 * - the deadline (relative to the actual activation time) of each
83 * instance.
84 * Very briefly, a periodic (sporadic) task asks for the execution of
85 * some specific computation --which is typically called an instance--
86 * (at most) every period. Moreover, each instance typically lasts no more
87 * than the runtime and must be completed by time instant t equal to
88 * the instance activation time + the deadline.
89 *
90 * This is reflected by the actual fields of the sched_attr structure:
91 *
92 * @size size of the structure, for fwd/bwd compat.
93 *
94 * @sched_policy task's scheduling policy
95 * @sched_flags for customizing the scheduler behaviour
96 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
97 * @sched_priority task's static priority (SCHED_FIFO/RR)
98 * @sched_deadline representative of the task's deadline
99 * @sched_runtime representative of the task's runtime
100 * @sched_period representative of the task's period
101 *
102 * Given this task model, there are a multiplicity of scheduling algorithms
103 * and policies, that can be used to ensure all the tasks will make their
104 * timing constraints.
aab03e05
DF
105 *
106 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
107 * only user of this new interface. More information about the algorithm
108 * available in the scheduling class file or in Documentation/.
d50dde5a
DF
109 */
110struct sched_attr {
111 u32 size;
112
113 u32 sched_policy;
114 u64 sched_flags;
115
116 /* SCHED_NORMAL, SCHED_BATCH */
117 s32 sched_nice;
118
119 /* SCHED_FIFO, SCHED_RR */
120 u32 sched_priority;
121
122 /* SCHED_DEADLINE */
123 u64 sched_runtime;
124 u64 sched_deadline;
125 u64 sched_period;
126};
127
1da177e4 128struct exec_domain;
c87e2837 129struct futex_pi_state;
286100a6 130struct robust_list_head;
bddd87c7 131struct bio_list;
5ad4e53b 132struct fs_struct;
cdd6c482 133struct perf_event_context;
73c10101 134struct blk_plug;
c4ad8f98 135struct filename;
1da177e4 136
615d6e87
DB
137#define VMACACHE_BITS 2
138#define VMACACHE_SIZE (1U << VMACACHE_BITS)
139#define VMACACHE_MASK (VMACACHE_SIZE - 1)
140
1da177e4
LT
141/*
142 * These are the constant used to fake the fixed-point load-average
143 * counting. Some notes:
144 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
145 * a load-average precision of 10 bits integer + 11 bits fractional
146 * - if you want to count load-averages more often, you need more
147 * precision, or rounding will get you. With 2-second counting freq,
148 * the EXP_n values would be 1981, 2034 and 2043 if still using only
149 * 11 bit fractions.
150 */
151extern unsigned long avenrun[]; /* Load averages */
2d02494f 152extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
1da177e4
LT
153
154#define FSHIFT 11 /* nr of bits of precision */
155#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
0c2043ab 156#define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
1da177e4
LT
157#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
158#define EXP_5 2014 /* 1/exp(5sec/5min) */
159#define EXP_15 2037 /* 1/exp(5sec/15min) */
160
161#define CALC_LOAD(load,exp,n) \
162 load *= exp; \
163 load += n*(FIXED_1-exp); \
164 load >>= FSHIFT;
165
166extern unsigned long total_forks;
167extern int nr_threads;
1da177e4
LT
168DECLARE_PER_CPU(unsigned long, process_counts);
169extern int nr_processes(void);
170extern unsigned long nr_running(void);
1da177e4 171extern unsigned long nr_iowait(void);
8c215bd3 172extern unsigned long nr_iowait_cpu(int cpu);
372ba8cb 173extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
69d25870 174
0f004f5a 175extern void calc_global_load(unsigned long ticks);
5aaa0b7a 176extern void update_cpu_load_nohz(void);
1da177e4 177
7e49fcce
SR
178extern unsigned long get_parent_ip(unsigned long addr);
179
b637a328
PM
180extern void dump_cpu_task(int cpu);
181
43ae34cb
IM
182struct seq_file;
183struct cfs_rq;
4cf86d77 184struct task_group;
43ae34cb
IM
185#ifdef CONFIG_SCHED_DEBUG
186extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
187extern void proc_sched_set_task(struct task_struct *p);
188extern void
5cef9eca 189print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
43ae34cb 190#endif
1da177e4 191
4a8342d2
LT
192/*
193 * Task state bitmask. NOTE! These bits are also
194 * encoded in fs/proc/array.c: get_task_state().
195 *
196 * We have two separate sets of flags: task->state
197 * is about runnability, while task->exit_state are
198 * about the task exiting. Confusing, but this way
199 * modifying one set can't modify the other one by
200 * mistake.
201 */
1da177e4
LT
202#define TASK_RUNNING 0
203#define TASK_INTERRUPTIBLE 1
204#define TASK_UNINTERRUPTIBLE 2
f021a3c2
MW
205#define __TASK_STOPPED 4
206#define __TASK_TRACED 8
4a8342d2 207/* in tsk->exit_state */
ad86622b
ON
208#define EXIT_DEAD 16
209#define EXIT_ZOMBIE 32
abd50b39 210#define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
4a8342d2 211/* in tsk->state again */
af927232 212#define TASK_DEAD 64
f021a3c2 213#define TASK_WAKEKILL 128
e9c84311 214#define TASK_WAKING 256
f2530dc7
TG
215#define TASK_PARKED 512
216#define TASK_STATE_MAX 1024
f021a3c2 217
ad0f614e 218#define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWP"
73342151 219
e1781538
PZ
220extern char ___assert_task_state[1 - 2*!!(
221 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
f021a3c2
MW
222
223/* Convenience macros for the sake of set_task_state */
224#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
225#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
226#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
1da177e4 227
92a1f4bc
MW
228/* Convenience macros for the sake of wake_up */
229#define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
f021a3c2 230#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
92a1f4bc
MW
231
232/* get_task_state() */
233#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
f021a3c2 234 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
74e37200 235 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
92a1f4bc 236
f021a3c2
MW
237#define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
238#define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
92a1f4bc 239#define task_is_stopped_or_traced(task) \
f021a3c2 240 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
92a1f4bc 241#define task_contributes_to_load(task) \
e3c8ca83 242 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
376fede8 243 (task->flags & PF_FROZEN) == 0)
1da177e4
LT
244
245#define __set_task_state(tsk, state_value) \
246 do { (tsk)->state = (state_value); } while (0)
247#define set_task_state(tsk, state_value) \
248 set_mb((tsk)->state, (state_value))
249
498d0c57
AM
250/*
251 * set_current_state() includes a barrier so that the write of current->state
252 * is correctly serialised wrt the caller's subsequent test of whether to
253 * actually sleep:
254 *
255 * set_current_state(TASK_UNINTERRUPTIBLE);
256 * if (do_i_need_to_sleep())
257 * schedule();
258 *
259 * If the caller does not need such serialisation then use __set_current_state()
260 */
1da177e4
LT
261#define __set_current_state(state_value) \
262 do { current->state = (state_value); } while (0)
263#define set_current_state(state_value) \
264 set_mb(current->state, (state_value))
265
266/* Task command name length */
267#define TASK_COMM_LEN 16
268
1da177e4
LT
269#include <linux/spinlock.h>
270
271/*
272 * This serializes "schedule()" and also protects
273 * the run-queue from deletions/modifications (but
274 * _adding_ to the beginning of the run-queue has
275 * a separate lock).
276 */
277extern rwlock_t tasklist_lock;
278extern spinlock_t mmlist_lock;
279
36c8b586 280struct task_struct;
1da177e4 281
db1466b3
PM
282#ifdef CONFIG_PROVE_RCU
283extern int lockdep_tasklist_lock_is_held(void);
284#endif /* #ifdef CONFIG_PROVE_RCU */
285
1da177e4
LT
286extern void sched_init(void);
287extern void sched_init_smp(void);
2d07b255 288extern asmlinkage void schedule_tail(struct task_struct *prev);
36c8b586 289extern void init_idle(struct task_struct *idle, int cpu);
1df21055 290extern void init_idle_bootup_task(struct task_struct *idle);
1da177e4 291
89f19f04 292extern int runqueue_is_locked(int cpu);
017730c1 293
3451d024 294#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
c1cc017c 295extern void nohz_balance_enter_idle(int cpu);
69e1e811 296extern void set_cpu_sd_state_idle(void);
6201b4d6 297extern int get_nohz_timer_target(int pinned);
46cb4b7c 298#else
c1cc017c 299static inline void nohz_balance_enter_idle(int cpu) { }
fdaabd80 300static inline void set_cpu_sd_state_idle(void) { }
6201b4d6
VK
301static inline int get_nohz_timer_target(int pinned)
302{
303 return smp_processor_id();
304}
46cb4b7c 305#endif
1da177e4 306
e59e2ae2 307/*
39bc89fd 308 * Only dump TASK_* tasks. (0 for all tasks)
e59e2ae2
IM
309 */
310extern void show_state_filter(unsigned long state_filter);
311
312static inline void show_state(void)
313{
39bc89fd 314 show_state_filter(0);
e59e2ae2
IM
315}
316
1da177e4
LT
317extern void show_regs(struct pt_regs *);
318
319/*
320 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
321 * task), SP is the stack pointer of the first frame that should be shown in the back
322 * trace (or NULL if the entire call-chain of the task should be shown).
323 */
324extern void show_stack(struct task_struct *task, unsigned long *sp);
325
326void io_schedule(void);
327long io_schedule_timeout(long timeout);
328
329extern void cpu_init (void);
330extern void trap_init(void);
331extern void update_process_times(int user);
332extern void scheduler_tick(void);
333
82a1fcb9
IM
334extern void sched_show_task(struct task_struct *p);
335
19cc36c0 336#ifdef CONFIG_LOCKUP_DETECTOR
8446f1d3 337extern void touch_softlockup_watchdog(void);
d6ad3e28 338extern void touch_softlockup_watchdog_sync(void);
04c9167f 339extern void touch_all_softlockup_watchdogs(void);
332fbdbc
DZ
340extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
341 void __user *buffer,
342 size_t *lenp, loff_t *ppos);
9c44bc03 343extern unsigned int softlockup_panic;
004417a6 344void lockup_detector_init(void);
8446f1d3 345#else
8446f1d3
IM
346static inline void touch_softlockup_watchdog(void)
347{
348}
d6ad3e28
JW
349static inline void touch_softlockup_watchdog_sync(void)
350{
351}
04c9167f
JF
352static inline void touch_all_softlockup_watchdogs(void)
353{
354}
004417a6
PZ
355static inline void lockup_detector_init(void)
356{
357}
8446f1d3
IM
358#endif
359
8b414521
MT
360#ifdef CONFIG_DETECT_HUNG_TASK
361void reset_hung_task_detector(void);
362#else
363static inline void reset_hung_task_detector(void)
364{
365}
366#endif
367
1da177e4
LT
368/* Attach to any functions which should be ignored in wchan output. */
369#define __sched __attribute__((__section__(".sched.text")))
deaf2227
IM
370
371/* Linker adds these: start and end of __sched functions */
372extern char __sched_text_start[], __sched_text_end[];
373
1da177e4
LT
374/* Is this address in the __sched functions? */
375extern int in_sched_functions(unsigned long addr);
376
377#define MAX_SCHEDULE_TIMEOUT LONG_MAX
b3c97528 378extern signed long schedule_timeout(signed long timeout);
64ed93a2 379extern signed long schedule_timeout_interruptible(signed long timeout);
294d5cc2 380extern signed long schedule_timeout_killable(signed long timeout);
64ed93a2 381extern signed long schedule_timeout_uninterruptible(signed long timeout);
1da177e4 382asmlinkage void schedule(void);
c5491ea7 383extern void schedule_preempt_disabled(void);
1da177e4 384
ab516013 385struct nsproxy;
acce292c 386struct user_namespace;
1da177e4 387
efc1a3b1
DH
388#ifdef CONFIG_MMU
389extern void arch_pick_mmap_layout(struct mm_struct *mm);
1da177e4
LT
390extern unsigned long
391arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
392 unsigned long, unsigned long);
393extern unsigned long
394arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
395 unsigned long len, unsigned long pgoff,
396 unsigned long flags);
efc1a3b1
DH
397#else
398static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
399#endif
1da177e4 400
d049f74f
KC
401#define SUID_DUMP_DISABLE 0 /* No setuid dumping */
402#define SUID_DUMP_USER 1 /* Dump as user of process */
403#define SUID_DUMP_ROOT 2 /* Dump as root */
404
6c5d5238 405/* mm flags */
f8af4da3 406
7288e118 407/* for SUID_DUMP_* above */
3cb4a0bb 408#define MMF_DUMPABLE_BITS 2
f8af4da3 409#define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
3cb4a0bb 410
942be387
ON
411extern void set_dumpable(struct mm_struct *mm, int value);
412/*
413 * This returns the actual value of the suid_dumpable flag. For things
414 * that are using this for checking for privilege transitions, it must
415 * test against SUID_DUMP_USER rather than treating it as a boolean
416 * value.
417 */
418static inline int __get_dumpable(unsigned long mm_flags)
419{
420 return mm_flags & MMF_DUMPABLE_MASK;
421}
422
423static inline int get_dumpable(struct mm_struct *mm)
424{
425 return __get_dumpable(mm->flags);
426}
427
3cb4a0bb
KH
428/* coredump filter bits */
429#define MMF_DUMP_ANON_PRIVATE 2
430#define MMF_DUMP_ANON_SHARED 3
431#define MMF_DUMP_MAPPED_PRIVATE 4
432#define MMF_DUMP_MAPPED_SHARED 5
82df3973 433#define MMF_DUMP_ELF_HEADERS 6
e575f111
KM
434#define MMF_DUMP_HUGETLB_PRIVATE 7
435#define MMF_DUMP_HUGETLB_SHARED 8
f8af4da3 436
3cb4a0bb 437#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
e575f111 438#define MMF_DUMP_FILTER_BITS 7
3cb4a0bb
KH
439#define MMF_DUMP_FILTER_MASK \
440 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
441#define MMF_DUMP_FILTER_DEFAULT \
e575f111 442 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
656eb2cd
RM
443 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
444
445#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
446# define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
447#else
448# define MMF_DUMP_MASK_DEFAULT_ELF 0
449#endif
f8af4da3
HD
450 /* leave room for more dump flags */
451#define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
ba76149f 452#define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
bafb282d 453#define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
f8af4da3 454
9f68f672
ON
455#define MMF_HAS_UPROBES 19 /* has uprobes */
456#define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
f8ac4ec9 457
f8af4da3 458#define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
6c5d5238 459
1da177e4
LT
460struct sighand_struct {
461 atomic_t count;
462 struct k_sigaction action[_NSIG];
463 spinlock_t siglock;
b8fceee1 464 wait_queue_head_t signalfd_wqh;
1da177e4
LT
465};
466
0e464814 467struct pacct_struct {
f6ec29a4
KK
468 int ac_flag;
469 long ac_exitcode;
0e464814 470 unsigned long ac_mem;
77787bfb
KK
471 cputime_t ac_utime, ac_stime;
472 unsigned long ac_minflt, ac_majflt;
0e464814
KK
473};
474
42c4ab41
SG
475struct cpu_itimer {
476 cputime_t expires;
477 cputime_t incr;
8356b5f9
SG
478 u32 error;
479 u32 incr_error;
42c4ab41
SG
480};
481
d37f761d
FW
482/**
483 * struct cputime - snaphsot of system and user cputime
484 * @utime: time spent in user mode
485 * @stime: time spent in system mode
486 *
487 * Gathers a generic snapshot of user and system time.
488 */
489struct cputime {
490 cputime_t utime;
491 cputime_t stime;
492};
493
f06febc9
FM
494/**
495 * struct task_cputime - collected CPU time counts
496 * @utime: time spent in user mode, in &cputime_t units
497 * @stime: time spent in kernel mode, in &cputime_t units
498 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
5ce73a4a 499 *
d37f761d
FW
500 * This is an extension of struct cputime that includes the total runtime
501 * spent by the task from the scheduler point of view.
502 *
503 * As a result, this structure groups together three kinds of CPU time
504 * that are tracked for threads and thread groups. Most things considering
f06febc9
FM
505 * CPU time want to group these counts together and treat all three
506 * of them in parallel.
507 */
508struct task_cputime {
509 cputime_t utime;
510 cputime_t stime;
511 unsigned long long sum_exec_runtime;
512};
513/* Alternate field names when used to cache expirations. */
514#define prof_exp stime
515#define virt_exp utime
516#define sched_exp sum_exec_runtime
517
4cd4c1b4
PZ
518#define INIT_CPUTIME \
519 (struct task_cputime) { \
64861634
MS
520 .utime = 0, \
521 .stime = 0, \
4cd4c1b4
PZ
522 .sum_exec_runtime = 0, \
523 }
524
a233f112
PZ
525#ifdef CONFIG_PREEMPT_COUNT
526#define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
527#else
528#define PREEMPT_DISABLED PREEMPT_ENABLED
529#endif
530
c99e6efe
PZ
531/*
532 * Disable preemption until the scheduler is running.
533 * Reset by start_kernel()->sched_init()->init_idle().
d86ee480
PZ
534 *
535 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
536 * before the scheduler is active -- see should_resched().
c99e6efe 537 */
a233f112 538#define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
c99e6efe 539
f06febc9 540/**
4cd4c1b4
PZ
541 * struct thread_group_cputimer - thread group interval timer counts
542 * @cputime: thread group interval timers.
543 * @running: non-zero when there are timers running and
544 * @cputime receives updates.
545 * @lock: lock for fields in this struct.
f06febc9
FM
546 *
547 * This structure contains the version of task_cputime, above, that is
4cd4c1b4 548 * used for thread group CPU timer calculations.
f06febc9 549 */
4cd4c1b4
PZ
550struct thread_group_cputimer {
551 struct task_cputime cputime;
552 int running;
ee30a7b2 553 raw_spinlock_t lock;
f06febc9 554};
f06febc9 555
4714d1d3 556#include <linux/rwsem.h>
5091faa4
MG
557struct autogroup;
558
1da177e4 559/*
e815f0a8 560 * NOTE! "signal_struct" does not have its own
1da177e4
LT
561 * locking, because a shared signal_struct always
562 * implies a shared sighand_struct, so locking
563 * sighand_struct is always a proper superset of
564 * the locking of signal_struct.
565 */
566struct signal_struct {
ea6d290c 567 atomic_t sigcnt;
1da177e4 568 atomic_t live;
b3ac022c 569 int nr_threads;
0c740d0a 570 struct list_head thread_head;
1da177e4
LT
571
572 wait_queue_head_t wait_chldexit; /* for wait4() */
573
574 /* current thread group signal load-balancing target: */
36c8b586 575 struct task_struct *curr_target;
1da177e4
LT
576
577 /* shared signal handling: */
578 struct sigpending shared_pending;
579
580 /* thread group exit support */
581 int group_exit_code;
582 /* overloaded:
583 * - notify group_exit_task when ->count is equal to notify_count
584 * - everyone except group_exit_task is stopped during signal delivery
585 * of fatal signals, group_exit_task processes the signal.
586 */
1da177e4 587 int notify_count;
07dd20e0 588 struct task_struct *group_exit_task;
1da177e4
LT
589
590 /* thread group stop support, overloads group_exit_code too */
591 int group_stop_count;
592 unsigned int flags; /* see SIGNAL_* flags below */
593
ebec18a6
LP
594 /*
595 * PR_SET_CHILD_SUBREAPER marks a process, like a service
596 * manager, to re-parent orphan (double-forking) child processes
597 * to this process instead of 'init'. The service manager is
598 * able to receive SIGCHLD signals and is able to investigate
599 * the process until it calls wait(). All children of this
600 * process will inherit a flag if they should look for a
601 * child_subreaper process at exit.
602 */
603 unsigned int is_child_subreaper:1;
604 unsigned int has_child_subreaper:1;
605
1da177e4 606 /* POSIX.1b Interval Timers */
5ed67f05
PE
607 int posix_timer_id;
608 struct list_head posix_timers;
1da177e4
LT
609
610 /* ITIMER_REAL timer for the process */
2ff678b8 611 struct hrtimer real_timer;
fea9d175 612 struct pid *leader_pid;
2ff678b8 613 ktime_t it_real_incr;
1da177e4 614
42c4ab41
SG
615 /*
616 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
617 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
618 * values are defined to 0 and 1 respectively
619 */
620 struct cpu_itimer it[2];
1da177e4 621
f06febc9 622 /*
4cd4c1b4
PZ
623 * Thread group totals for process CPU timers.
624 * See thread_group_cputimer(), et al, for details.
f06febc9 625 */
4cd4c1b4 626 struct thread_group_cputimer cputimer;
f06febc9
FM
627
628 /* Earliest-expiration cache. */
629 struct task_cputime cputime_expires;
630
631 struct list_head cpu_timers[3];
632
ab521dc0 633 struct pid *tty_old_pgrp;
1ec320af 634
1da177e4
LT
635 /* boolean value for session group leader */
636 int leader;
637
638 struct tty_struct *tty; /* NULL if no tty */
639
5091faa4
MG
640#ifdef CONFIG_SCHED_AUTOGROUP
641 struct autogroup *autogroup;
642#endif
1da177e4
LT
643 /*
644 * Cumulative resource counters for dead threads in the group,
645 * and for reaped dead child processes forked by this group.
646 * Live threads maintain their own counters and add to these
647 * in __exit_signal, except for the group leader.
648 */
e78c3496 649 seqlock_t stats_lock;
32bd671d 650 cputime_t utime, stime, cutime, cstime;
9ac52315
LV
651 cputime_t gtime;
652 cputime_t cgtime;
9fbc42ea 653#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
d37f761d 654 struct cputime prev_cputime;
0cf55e1e 655#endif
1da177e4
LT
656 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
657 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
6eaeeaba 658 unsigned long inblock, oublock, cinblock, coublock;
1f10206c 659 unsigned long maxrss, cmaxrss;
940389b8 660 struct task_io_accounting ioac;
1da177e4 661
32bd671d
PZ
662 /*
663 * Cumulative ns of schedule CPU time fo dead threads in the
664 * group, not including a zombie group leader, (This only differs
665 * from jiffies_to_ns(utime + stime) if sched_clock uses something
666 * other than jiffies.)
667 */
668 unsigned long long sum_sched_runtime;
669
1da177e4
LT
670 /*
671 * We don't bother to synchronize most readers of this at all,
672 * because there is no reader checking a limit that actually needs
673 * to get both rlim_cur and rlim_max atomically, and either one
674 * alone is a single word that can safely be read normally.
675 * getrlimit/setrlimit use task_lock(current->group_leader) to
676 * protect this instead of the siglock, because they really
677 * have no need to disable irqs.
678 */
679 struct rlimit rlim[RLIM_NLIMITS];
680
0e464814
KK
681#ifdef CONFIG_BSD_PROCESS_ACCT
682 struct pacct_struct pacct; /* per-process accounting information */
683#endif
ad4ecbcb 684#ifdef CONFIG_TASKSTATS
ad4ecbcb
SN
685 struct taskstats *stats;
686#endif
522ed776
MT
687#ifdef CONFIG_AUDIT
688 unsigned audit_tty;
46e959ea 689 unsigned audit_tty_log_passwd;
522ed776
MT
690 struct tty_audit_buf *tty_audit_buf;
691#endif
4714d1d3
BB
692#ifdef CONFIG_CGROUPS
693 /*
77e4ef99
TH
694 * group_rwsem prevents new tasks from entering the threadgroup and
695 * member tasks from exiting,a more specifically, setting of
696 * PF_EXITING. fork and exit paths are protected with this rwsem
697 * using threadgroup_change_begin/end(). Users which require
698 * threadgroup to remain stable should use threadgroup_[un]lock()
699 * which also takes care of exec path. Currently, cgroup is the
700 * only user.
4714d1d3 701 */
257058ae 702 struct rw_semaphore group_rwsem;
4714d1d3 703#endif
28b83c51 704
e1e12d2f 705 oom_flags_t oom_flags;
a9c58b90
DR
706 short oom_score_adj; /* OOM kill score adjustment */
707 short oom_score_adj_min; /* OOM kill score adjustment min value.
708 * Only settable by CAP_SYS_RESOURCE. */
9b1bf12d
KM
709
710 struct mutex cred_guard_mutex; /* guard against foreign influences on
711 * credential calculations
712 * (notably. ptrace) */
1da177e4
LT
713};
714
715/*
716 * Bits in flags field of signal_struct.
717 */
718#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
ee77f075
ON
719#define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
720#define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
403bad72 721#define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
e4420551
ON
722/*
723 * Pending notifications to parent.
724 */
725#define SIGNAL_CLD_STOPPED 0x00000010
726#define SIGNAL_CLD_CONTINUED 0x00000020
727#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
1da177e4 728
fae5fa44
ON
729#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
730
ed5d2cac
ON
731/* If true, all threads except ->group_exit_task have pending SIGKILL */
732static inline int signal_group_exit(const struct signal_struct *sig)
733{
734 return (sig->flags & SIGNAL_GROUP_EXIT) ||
735 (sig->group_exit_task != NULL);
736}
737
1da177e4
LT
738/*
739 * Some day this will be a full-fledged user tracking system..
740 */
741struct user_struct {
742 atomic_t __count; /* reference count */
743 atomic_t processes; /* How many processes does this user have? */
1da177e4 744 atomic_t sigpending; /* How many pending signals does this user have? */
2d9048e2 745#ifdef CONFIG_INOTIFY_USER
0eeca283
RL
746 atomic_t inotify_watches; /* How many inotify watches does this user have? */
747 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
748#endif
4afeff85
EP
749#ifdef CONFIG_FANOTIFY
750 atomic_t fanotify_listeners;
751#endif
7ef9964e 752#ifdef CONFIG_EPOLL
52bd19f7 753 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
7ef9964e 754#endif
970a8645 755#ifdef CONFIG_POSIX_MQUEUE
1da177e4
LT
756 /* protected by mq_lock */
757 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
970a8645 758#endif
1da177e4
LT
759 unsigned long locked_shm; /* How many pages of mlocked shm ? */
760
761#ifdef CONFIG_KEYS
762 struct key *uid_keyring; /* UID specific keyring */
763 struct key *session_keyring; /* UID's default session keyring */
764#endif
765
766 /* Hash table maintenance information */
735de223 767 struct hlist_node uidhash_node;
7b44ab97 768 kuid_t uid;
24e377a8 769
cdd6c482 770#ifdef CONFIG_PERF_EVENTS
789f90fc
PZ
771 atomic_long_t locked_vm;
772#endif
1da177e4
LT
773};
774
eb41d946 775extern int uids_sysfs_init(void);
5cb350ba 776
7b44ab97 777extern struct user_struct *find_user(kuid_t);
1da177e4
LT
778
779extern struct user_struct root_user;
780#define INIT_USER (&root_user)
781
b6dff3ec 782
1da177e4
LT
783struct backing_dev_info;
784struct reclaim_state;
785
52f17b6c 786#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1da177e4
LT
787struct sched_info {
788 /* cumulative counters */
2d72376b 789 unsigned long pcount; /* # of times run on this cpu */
9c2c4802 790 unsigned long long run_delay; /* time spent waiting on a runqueue */
1da177e4
LT
791
792 /* timestamps */
172ba844
BS
793 unsigned long long last_arrival,/* when we last ran on a cpu */
794 last_queued; /* when we were last queued to run */
1da177e4 795};
52f17b6c 796#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
1da177e4 797
ca74e92b
SN
798#ifdef CONFIG_TASK_DELAY_ACCT
799struct task_delay_info {
800 spinlock_t lock;
801 unsigned int flags; /* Private per-task flags */
802
803 /* For each stat XXX, add following, aligned appropriately
804 *
805 * struct timespec XXX_start, XXX_end;
806 * u64 XXX_delay;
807 * u32 XXX_count;
808 *
809 * Atomicity of updates to XXX_delay, XXX_count protected by
810 * single lock above (split into XXX_lock if contention is an issue).
811 */
0ff92245
SN
812
813 /*
814 * XXX_count is incremented on every XXX operation, the delay
815 * associated with the operation is added to XXX_delay.
816 * XXX_delay contains the accumulated delay time in nanoseconds.
817 */
9667a23d 818 u64 blkio_start; /* Shared by blkio, swapin */
0ff92245
SN
819 u64 blkio_delay; /* wait for sync block io completion */
820 u64 swapin_delay; /* wait for swapin block io completion */
821 u32 blkio_count; /* total count of the number of sync block */
822 /* io operations performed */
823 u32 swapin_count; /* total count of the number of swapin block */
824 /* io operations performed */
873b4771 825
9667a23d 826 u64 freepages_start;
873b4771
KK
827 u64 freepages_delay; /* wait for memory reclaim */
828 u32 freepages_count; /* total count of memory reclaim */
ca74e92b 829};
52f17b6c
CS
830#endif /* CONFIG_TASK_DELAY_ACCT */
831
832static inline int sched_info_on(void)
833{
834#ifdef CONFIG_SCHEDSTATS
835 return 1;
836#elif defined(CONFIG_TASK_DELAY_ACCT)
837 extern int delayacct_on;
838 return delayacct_on;
839#else
840 return 0;
ca74e92b 841#endif
52f17b6c 842}
ca74e92b 843
d15bcfdb
IM
844enum cpu_idle_type {
845 CPU_IDLE,
846 CPU_NOT_IDLE,
847 CPU_NEWLY_IDLE,
848 CPU_MAX_IDLE_TYPES
1da177e4
LT
849};
850
1399fa78 851/*
ca8ce3d0 852 * Increase resolution of cpu_capacity calculations
1399fa78 853 */
ca8ce3d0
NP
854#define SCHED_CAPACITY_SHIFT 10
855#define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
1da177e4 856
1399fa78
NR
857/*
858 * sched-domains (multiprocessor balancing) declarations:
859 */
2dd73a4f 860#ifdef CONFIG_SMP
b5d978e0
PZ
861#define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
862#define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
863#define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
864#define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
c88d5910 865#define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
b5d978e0 866#define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
5d4dfddd 867#define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
d77b3ed5 868#define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
b5d978e0
PZ
869#define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
870#define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
532cb4c4 871#define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
b5d978e0 872#define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
e3589f6c 873#define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
3a7053b3 874#define SD_NUMA 0x4000 /* cross-node balancing */
5c45bf27 875
143e1e28 876#ifdef CONFIG_SCHED_SMT
b6220ad6 877static inline int cpu_smt_flags(void)
143e1e28 878{
5d4dfddd 879 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
143e1e28
VG
880}
881#endif
882
883#ifdef CONFIG_SCHED_MC
b6220ad6 884static inline int cpu_core_flags(void)
143e1e28
VG
885{
886 return SD_SHARE_PKG_RESOURCES;
887}
888#endif
889
890#ifdef CONFIG_NUMA
b6220ad6 891static inline int cpu_numa_flags(void)
143e1e28
VG
892{
893 return SD_NUMA;
894}
895#endif
532cb4c4 896
1d3504fc
HS
897struct sched_domain_attr {
898 int relax_domain_level;
899};
900
901#define SD_ATTR_INIT (struct sched_domain_attr) { \
902 .relax_domain_level = -1, \
903}
904
60495e77
PZ
905extern int sched_domain_level_max;
906
5e6521ea
LZ
907struct sched_group;
908
1da177e4
LT
909struct sched_domain {
910 /* These fields must be setup */
911 struct sched_domain *parent; /* top domain must be null terminated */
1a848870 912 struct sched_domain *child; /* bottom domain must be null terminated */
1da177e4 913 struct sched_group *groups; /* the balancing groups of the domain */
1da177e4
LT
914 unsigned long min_interval; /* Minimum balance interval ms */
915 unsigned long max_interval; /* Maximum balance interval ms */
916 unsigned int busy_factor; /* less balancing by factor if busy */
917 unsigned int imbalance_pct; /* No balance until over watermark */
1da177e4 918 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
7897986b
NP
919 unsigned int busy_idx;
920 unsigned int idle_idx;
921 unsigned int newidle_idx;
922 unsigned int wake_idx;
147cbb4b 923 unsigned int forkexec_idx;
a52bfd73 924 unsigned int smt_gain;
25f55d9d
VG
925
926 int nohz_idle; /* NOHZ IDLE status */
1da177e4 927 int flags; /* See SD_* */
60495e77 928 int level;
1da177e4
LT
929
930 /* Runtime fields. */
931 unsigned long last_balance; /* init to jiffies. units in jiffies */
932 unsigned int balance_interval; /* initialise to 1. units in ms. */
933 unsigned int nr_balance_failed; /* initialise to 0 */
934
f48627e6 935 /* idle_balance() stats */
9bd721c5 936 u64 max_newidle_lb_cost;
f48627e6 937 unsigned long next_decay_max_lb_cost;
2398f2c6 938
1da177e4
LT
939#ifdef CONFIG_SCHEDSTATS
940 /* load_balance() stats */
480b9434
KC
941 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
942 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
943 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
944 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
945 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
946 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
947 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
948 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1da177e4
LT
949
950 /* Active load balancing */
480b9434
KC
951 unsigned int alb_count;
952 unsigned int alb_failed;
953 unsigned int alb_pushed;
1da177e4 954
68767a0a 955 /* SD_BALANCE_EXEC stats */
480b9434
KC
956 unsigned int sbe_count;
957 unsigned int sbe_balanced;
958 unsigned int sbe_pushed;
1da177e4 959
68767a0a 960 /* SD_BALANCE_FORK stats */
480b9434
KC
961 unsigned int sbf_count;
962 unsigned int sbf_balanced;
963 unsigned int sbf_pushed;
68767a0a 964
1da177e4 965 /* try_to_wake_up() stats */
480b9434
KC
966 unsigned int ttwu_wake_remote;
967 unsigned int ttwu_move_affine;
968 unsigned int ttwu_move_balance;
1da177e4 969#endif
a5d8c348
IM
970#ifdef CONFIG_SCHED_DEBUG
971 char *name;
972#endif
dce840a0
PZ
973 union {
974 void *private; /* used during construction */
975 struct rcu_head rcu; /* used during destruction */
976 };
6c99e9ad 977
669c55e9 978 unsigned int span_weight;
4200efd9
IM
979 /*
980 * Span of all CPUs in this domain.
981 *
982 * NOTE: this field is variable length. (Allocated dynamically
983 * by attaching extra space to the end of the structure,
984 * depending on how many CPUs the kernel has booted up with)
4200efd9
IM
985 */
986 unsigned long span[0];
1da177e4
LT
987};
988
758b2cdc
RR
989static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
990{
6c99e9ad 991 return to_cpumask(sd->span);
758b2cdc
RR
992}
993
acc3f5d7 994extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1d3504fc 995 struct sched_domain_attr *dattr_new);
029190c5 996
acc3f5d7
RR
997/* Allocate an array of sched domains, for partition_sched_domains(). */
998cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
999void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1000
39be3501
PZ
1001bool cpus_share_cache(int this_cpu, int that_cpu);
1002
143e1e28 1003typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
b6220ad6 1004typedef int (*sched_domain_flags_f)(void);
143e1e28
VG
1005
1006#define SDTL_OVERLAP 0x01
1007
1008struct sd_data {
1009 struct sched_domain **__percpu sd;
1010 struct sched_group **__percpu sg;
63b2ca30 1011 struct sched_group_capacity **__percpu sgc;
143e1e28
VG
1012};
1013
1014struct sched_domain_topology_level {
1015 sched_domain_mask_f mask;
1016 sched_domain_flags_f sd_flags;
1017 int flags;
1018 int numa_level;
1019 struct sd_data data;
1020#ifdef CONFIG_SCHED_DEBUG
1021 char *name;
1022#endif
1023};
1024
1025extern struct sched_domain_topology_level *sched_domain_topology;
1026
1027extern void set_sched_topology(struct sched_domain_topology_level *tl);
f6be8af1 1028extern void wake_up_if_idle(int cpu);
143e1e28
VG
1029
1030#ifdef CONFIG_SCHED_DEBUG
1031# define SD_INIT_NAME(type) .name = #type
1032#else
1033# define SD_INIT_NAME(type)
1034#endif
1035
1b427c15 1036#else /* CONFIG_SMP */
1da177e4 1037
1b427c15 1038struct sched_domain_attr;
d02c7a8c 1039
1b427c15 1040static inline void
acc3f5d7 1041partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1b427c15
IM
1042 struct sched_domain_attr *dattr_new)
1043{
d02c7a8c 1044}
39be3501
PZ
1045
1046static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1047{
1048 return true;
1049}
1050
1b427c15 1051#endif /* !CONFIG_SMP */
1da177e4 1052
47fe38fc 1053
1da177e4 1054struct io_context; /* See blkdev.h */
1da177e4 1055
1da177e4 1056
383f2835 1057#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
36c8b586 1058extern void prefetch_stack(struct task_struct *t);
383f2835
CK
1059#else
1060static inline void prefetch_stack(struct task_struct *t) { }
1061#endif
1da177e4
LT
1062
1063struct audit_context; /* See audit.c */
1064struct mempolicy;
b92ce558 1065struct pipe_inode_info;
4865ecf1 1066struct uts_namespace;
1da177e4 1067
20b8a59f 1068struct load_weight {
9dbdb155
PZ
1069 unsigned long weight;
1070 u32 inv_weight;
20b8a59f
IM
1071};
1072
9d85f21c
PT
1073struct sched_avg {
1074 /*
1075 * These sums represent an infinite geometric series and so are bound
239003ea 1076 * above by 1024/(1-y). Thus we only need a u32 to store them for all
9d85f21c
PT
1077 * choices of y < 1-2^(-32)*1024.
1078 */
1079 u32 runnable_avg_sum, runnable_avg_period;
1080 u64 last_runnable_update;
9ee474f5 1081 s64 decay_count;
2dac754e 1082 unsigned long load_avg_contrib;
9d85f21c
PT
1083};
1084
94c18227 1085#ifdef CONFIG_SCHEDSTATS
41acab88 1086struct sched_statistics {
20b8a59f 1087 u64 wait_start;
94c18227 1088 u64 wait_max;
6d082592
AV
1089 u64 wait_count;
1090 u64 wait_sum;
8f0dfc34
AV
1091 u64 iowait_count;
1092 u64 iowait_sum;
94c18227 1093
20b8a59f 1094 u64 sleep_start;
20b8a59f 1095 u64 sleep_max;
94c18227
IM
1096 s64 sum_sleep_runtime;
1097
1098 u64 block_start;
20b8a59f
IM
1099 u64 block_max;
1100 u64 exec_max;
eba1ed4b 1101 u64 slice_max;
cc367732 1102
cc367732
IM
1103 u64 nr_migrations_cold;
1104 u64 nr_failed_migrations_affine;
1105 u64 nr_failed_migrations_running;
1106 u64 nr_failed_migrations_hot;
1107 u64 nr_forced_migrations;
cc367732
IM
1108
1109 u64 nr_wakeups;
1110 u64 nr_wakeups_sync;
1111 u64 nr_wakeups_migrate;
1112 u64 nr_wakeups_local;
1113 u64 nr_wakeups_remote;
1114 u64 nr_wakeups_affine;
1115 u64 nr_wakeups_affine_attempts;
1116 u64 nr_wakeups_passive;
1117 u64 nr_wakeups_idle;
41acab88
LDM
1118};
1119#endif
1120
1121struct sched_entity {
1122 struct load_weight load; /* for load-balancing */
1123 struct rb_node run_node;
1124 struct list_head group_node;
1125 unsigned int on_rq;
1126
1127 u64 exec_start;
1128 u64 sum_exec_runtime;
1129 u64 vruntime;
1130 u64 prev_sum_exec_runtime;
1131
41acab88
LDM
1132 u64 nr_migrations;
1133
41acab88
LDM
1134#ifdef CONFIG_SCHEDSTATS
1135 struct sched_statistics statistics;
94c18227
IM
1136#endif
1137
20b8a59f 1138#ifdef CONFIG_FAIR_GROUP_SCHED
fed14d45 1139 int depth;
20b8a59f
IM
1140 struct sched_entity *parent;
1141 /* rq on which this entity is (to be) queued: */
1142 struct cfs_rq *cfs_rq;
1143 /* rq "owned" by this entity/group: */
1144 struct cfs_rq *my_q;
1145#endif
8bd75c77 1146
141965c7 1147#ifdef CONFIG_SMP
f4e26b12 1148 /* Per-entity load-tracking */
9d85f21c
PT
1149 struct sched_avg avg;
1150#endif
20b8a59f 1151};
70b97a7f 1152
fa717060
PZ
1153struct sched_rt_entity {
1154 struct list_head run_list;
78f2c7db 1155 unsigned long timeout;
57d2aa00 1156 unsigned long watchdog_stamp;
bee367ed 1157 unsigned int time_slice;
6f505b16 1158
58d6c2d7 1159 struct sched_rt_entity *back;
052f1dc7 1160#ifdef CONFIG_RT_GROUP_SCHED
6f505b16
PZ
1161 struct sched_rt_entity *parent;
1162 /* rq on which this entity is (to be) queued: */
1163 struct rt_rq *rt_rq;
1164 /* rq "owned" by this entity/group: */
1165 struct rt_rq *my_q;
1166#endif
fa717060
PZ
1167};
1168
aab03e05
DF
1169struct sched_dl_entity {
1170 struct rb_node rb_node;
1171
1172 /*
1173 * Original scheduling parameters. Copied here from sched_attr
4027d080 1174 * during sched_setattr(), they will remain the same until
1175 * the next sched_setattr().
aab03e05
DF
1176 */
1177 u64 dl_runtime; /* maximum runtime for each instance */
1178 u64 dl_deadline; /* relative deadline of each instance */
755378a4 1179 u64 dl_period; /* separation of two instances (period) */
332ac17e 1180 u64 dl_bw; /* dl_runtime / dl_deadline */
aab03e05
DF
1181
1182 /*
1183 * Actual scheduling parameters. Initialized with the values above,
1184 * they are continously updated during task execution. Note that
1185 * the remaining runtime could be < 0 in case we are in overrun.
1186 */
1187 s64 runtime; /* remaining runtime for this instance */
1188 u64 deadline; /* absolute deadline for this instance */
1189 unsigned int flags; /* specifying the scheduler behaviour */
1190
1191 /*
1192 * Some bool flags:
1193 *
1194 * @dl_throttled tells if we exhausted the runtime. If so, the
1195 * task has to wait for a replenishment to be performed at the
1196 * next firing of dl_timer.
1197 *
1198 * @dl_new tells if a new instance arrived. If so we must
1199 * start executing it with full runtime and reset its absolute
1200 * deadline;
2d3d891d
DF
1201 *
1202 * @dl_boosted tells if we are boosted due to DI. If so we are
1203 * outside bandwidth enforcement mechanism (but only until we
5bfd126e
JL
1204 * exit the critical section);
1205 *
1206 * @dl_yielded tells if task gave up the cpu before consuming
1207 * all its available runtime during the last job.
aab03e05 1208 */
5bfd126e 1209 int dl_throttled, dl_new, dl_boosted, dl_yielded;
aab03e05
DF
1210
1211 /*
1212 * Bandwidth enforcement timer. Each -deadline task has its
1213 * own bandwidth to be enforced, thus we need one timer per task.
1214 */
1215 struct hrtimer dl_timer;
1216};
8bd75c77 1217
86848966
PM
1218struct rcu_node;
1219
8dc85d54
PZ
1220enum perf_event_task_context {
1221 perf_invalid_context = -1,
1222 perf_hw_context = 0,
89a1e187 1223 perf_sw_context,
8dc85d54
PZ
1224 perf_nr_task_contexts,
1225};
1226
1da177e4
LT
1227struct task_struct {
1228 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
f7e4217b 1229 void *stack;
1da177e4 1230 atomic_t usage;
97dc32cd
WC
1231 unsigned int flags; /* per process flags, defined below */
1232 unsigned int ptrace;
1da177e4 1233
2dd73a4f 1234#ifdef CONFIG_SMP
fa14ff4a 1235 struct llist_node wake_entry;
3ca7a440 1236 int on_cpu;
62470419
MW
1237 struct task_struct *last_wakee;
1238 unsigned long wakee_flips;
1239 unsigned long wakee_flip_decay_ts;
ac66f547
PZ
1240
1241 int wake_cpu;
2dd73a4f 1242#endif
fd2f4419 1243 int on_rq;
50e645a8 1244
b29739f9 1245 int prio, static_prio, normal_prio;
c7aceaba 1246 unsigned int rt_priority;
5522d5d5 1247 const struct sched_class *sched_class;
20b8a59f 1248 struct sched_entity se;
fa717060 1249 struct sched_rt_entity rt;
8323f26c
PZ
1250#ifdef CONFIG_CGROUP_SCHED
1251 struct task_group *sched_task_group;
1252#endif
aab03e05 1253 struct sched_dl_entity dl;
1da177e4 1254
e107be36
AK
1255#ifdef CONFIG_PREEMPT_NOTIFIERS
1256 /* list of struct preempt_notifier: */
1257 struct hlist_head preempt_notifiers;
1258#endif
1259
6c5c9341 1260#ifdef CONFIG_BLK_DEV_IO_TRACE
2056a782 1261 unsigned int btrace_seq;
6c5c9341 1262#endif
1da177e4 1263
97dc32cd 1264 unsigned int policy;
29baa747 1265 int nr_cpus_allowed;
1da177e4 1266 cpumask_t cpus_allowed;
1da177e4 1267
a57eb940 1268#ifdef CONFIG_PREEMPT_RCU
e260be67 1269 int rcu_read_lock_nesting;
f41d911f 1270 char rcu_read_unlock_special;
f41d911f 1271 struct list_head rcu_node_entry;
a57eb940
PM
1272#endif /* #ifdef CONFIG_PREEMPT_RCU */
1273#ifdef CONFIG_TREE_PREEMPT_RCU
1274 struct rcu_node *rcu_blocked_node;
f41d911f 1275#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
e260be67 1276
52f17b6c 1277#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1da177e4
LT
1278 struct sched_info sched_info;
1279#endif
1280
1281 struct list_head tasks;
806c09a7 1282#ifdef CONFIG_SMP
917b627d 1283 struct plist_node pushable_tasks;
1baca4ce 1284 struct rb_node pushable_dl_tasks;
806c09a7 1285#endif
1da177e4
LT
1286
1287 struct mm_struct *mm, *active_mm;
4471a675
JK
1288#ifdef CONFIG_COMPAT_BRK
1289 unsigned brk_randomized:1;
1290#endif
615d6e87
DB
1291 /* per-thread vma caching */
1292 u32 vmacache_seqnum;
1293 struct vm_area_struct *vmacache[VMACACHE_SIZE];
34e55232
KH
1294#if defined(SPLIT_RSS_COUNTING)
1295 struct task_rss_stat rss_stat;
1296#endif
1da177e4 1297/* task state */
97dc32cd 1298 int exit_state;
1da177e4
LT
1299 int exit_code, exit_signal;
1300 int pdeath_signal; /* The signal sent when the parent dies */
a8f072c1 1301 unsigned int jobctl; /* JOBCTL_*, siglock protected */
9b89f6ba
AE
1302
1303 /* Used for emulating ABI behavior of previous Linux versions */
97dc32cd 1304 unsigned int personality;
9b89f6ba 1305
f9ce1f1c
KT
1306 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1307 * execve */
8f0dfc34
AV
1308 unsigned in_iowait:1;
1309
ca94c442
LP
1310 /* Revert to default priority/policy when forking */
1311 unsigned sched_reset_on_fork:1;
a8e4f2ea 1312 unsigned sched_contributes_to_load:1;
ca94c442 1313
1d4457f9
KC
1314 unsigned long atomic_flags; /* Flags needing atomic access. */
1315
1da177e4
LT
1316 pid_t pid;
1317 pid_t tgid;
0a425405 1318
1314562a 1319#ifdef CONFIG_CC_STACKPROTECTOR
0a425405
AV
1320 /* Canary value for the -fstack-protector gcc feature */
1321 unsigned long stack_canary;
1314562a 1322#endif
4d1d61a6 1323 /*
1da177e4 1324 * pointers to (original) parent process, youngest child, younger sibling,
4d1d61a6 1325 * older sibling, respectively. (p->father can be replaced with
f470021a 1326 * p->real_parent->pid)
1da177e4 1327 */
abd63bc3
KC
1328 struct task_struct __rcu *real_parent; /* real parent process */
1329 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1da177e4 1330 /*
f470021a 1331 * children/sibling forms the list of my natural children
1da177e4
LT
1332 */
1333 struct list_head children; /* list of my children */
1334 struct list_head sibling; /* linkage in my parent's children list */
1335 struct task_struct *group_leader; /* threadgroup leader */
1336
f470021a
RM
1337 /*
1338 * ptraced is the list of tasks this task is using ptrace on.
1339 * This includes both natural children and PTRACE_ATTACH targets.
1340 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1341 */
1342 struct list_head ptraced;
1343 struct list_head ptrace_entry;
1344
1da177e4 1345 /* PID/PID hash table linkage. */
92476d7f 1346 struct pid_link pids[PIDTYPE_MAX];
47e65328 1347 struct list_head thread_group;
0c740d0a 1348 struct list_head thread_node;
1da177e4
LT
1349
1350 struct completion *vfork_done; /* for vfork() */
1351 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1352 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1353
c66f08be 1354 cputime_t utime, stime, utimescaled, stimescaled;
9ac52315 1355 cputime_t gtime;
9fbc42ea 1356#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
d37f761d 1357 struct cputime prev_cputime;
6a61671b
FW
1358#endif
1359#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1360 seqlock_t vtime_seqlock;
1361 unsigned long long vtime_snap;
1362 enum {
1363 VTIME_SLEEPING = 0,
1364 VTIME_USER,
1365 VTIME_SYS,
1366 } vtime_snap_whence;
d99ca3b9 1367#endif
1da177e4 1368 unsigned long nvcsw, nivcsw; /* context switch counts */
ccbf62d8 1369 u64 start_time; /* monotonic time in nsec */
57e0be04 1370 u64 real_start_time; /* boot based time in nsec */
1da177e4
LT
1371/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1372 unsigned long min_flt, maj_flt;
1373
f06febc9 1374 struct task_cputime cputime_expires;
1da177e4
LT
1375 struct list_head cpu_timers[3];
1376
1377/* process credentials */
1b0ba1c9 1378 const struct cred __rcu *real_cred; /* objective and real subjective task
3b11a1de 1379 * credentials (COW) */
1b0ba1c9 1380 const struct cred __rcu *cred; /* effective (overridable) subjective task
3b11a1de 1381 * credentials (COW) */
36772092
PBG
1382 char comm[TASK_COMM_LEN]; /* executable name excluding path
1383 - access with [gs]et_task_comm (which lock
1384 it with task_lock())
221af7f8 1385 - initialized normally by setup_new_exec */
1da177e4
LT
1386/* file system info */
1387 int link_count, total_link_count;
3d5b6fcc 1388#ifdef CONFIG_SYSVIPC
1da177e4
LT
1389/* ipc stuff */
1390 struct sysv_sem sysvsem;
ab602f79 1391 struct sysv_shm sysvshm;
3d5b6fcc 1392#endif
e162b39a 1393#ifdef CONFIG_DETECT_HUNG_TASK
82a1fcb9 1394/* hung task detection */
82a1fcb9
IM
1395 unsigned long last_switch_count;
1396#endif
1da177e4
LT
1397/* CPU-specific state of this task */
1398 struct thread_struct thread;
1399/* filesystem information */
1400 struct fs_struct *fs;
1401/* open file information */
1402 struct files_struct *files;
1651e14e 1403/* namespaces */
ab516013 1404 struct nsproxy *nsproxy;
1da177e4
LT
1405/* signal handlers */
1406 struct signal_struct *signal;
1407 struct sighand_struct *sighand;
1408
1409 sigset_t blocked, real_blocked;
f3de272b 1410 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1da177e4
LT
1411 struct sigpending pending;
1412
1413 unsigned long sas_ss_sp;
1414 size_t sas_ss_size;
1415 int (*notifier)(void *priv);
1416 void *notifier_data;
1417 sigset_t *notifier_mask;
67d12145 1418 struct callback_head *task_works;
e73f8959 1419
1da177e4 1420 struct audit_context *audit_context;
bfef93a5 1421#ifdef CONFIG_AUDITSYSCALL
e1760bd5 1422 kuid_t loginuid;
4746ec5b 1423 unsigned int sessionid;
bfef93a5 1424#endif
932ecebb 1425 struct seccomp seccomp;
1da177e4
LT
1426
1427/* Thread group tracking */
1428 u32 parent_exec_id;
1429 u32 self_exec_id;
58568d2a
MX
1430/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1431 * mempolicy */
1da177e4 1432 spinlock_t alloc_lock;
1da177e4 1433
b29739f9 1434 /* Protection of the PI data structures: */
1d615482 1435 raw_spinlock_t pi_lock;
b29739f9 1436
23f78d4a
IM
1437#ifdef CONFIG_RT_MUTEXES
1438 /* PI waiters blocked on a rt_mutex held by this task */
fb00aca4
PZ
1439 struct rb_root pi_waiters;
1440 struct rb_node *pi_waiters_leftmost;
23f78d4a
IM
1441 /* Deadlock detection and priority inheritance handling */
1442 struct rt_mutex_waiter *pi_blocked_on;
23f78d4a
IM
1443#endif
1444
408894ee
IM
1445#ifdef CONFIG_DEBUG_MUTEXES
1446 /* mutex deadlock detection */
1447 struct mutex_waiter *blocked_on;
1448#endif
de30a2b3
IM
1449#ifdef CONFIG_TRACE_IRQFLAGS
1450 unsigned int irq_events;
de30a2b3 1451 unsigned long hardirq_enable_ip;
de30a2b3 1452 unsigned long hardirq_disable_ip;
fa1452e8 1453 unsigned int hardirq_enable_event;
de30a2b3 1454 unsigned int hardirq_disable_event;
fa1452e8
HS
1455 int hardirqs_enabled;
1456 int hardirq_context;
de30a2b3 1457 unsigned long softirq_disable_ip;
de30a2b3 1458 unsigned long softirq_enable_ip;
fa1452e8 1459 unsigned int softirq_disable_event;
de30a2b3 1460 unsigned int softirq_enable_event;
fa1452e8 1461 int softirqs_enabled;
de30a2b3
IM
1462 int softirq_context;
1463#endif
fbb9ce95 1464#ifdef CONFIG_LOCKDEP
bdb9441e 1465# define MAX_LOCK_DEPTH 48UL
fbb9ce95
IM
1466 u64 curr_chain_key;
1467 int lockdep_depth;
fbb9ce95 1468 unsigned int lockdep_recursion;
c7aceaba 1469 struct held_lock held_locks[MAX_LOCK_DEPTH];
cf40bd16 1470 gfp_t lockdep_reclaim_gfp;
fbb9ce95 1471#endif
408894ee 1472
1da177e4
LT
1473/* journalling filesystem info */
1474 void *journal_info;
1475
d89d8796 1476/* stacked block device info */
bddd87c7 1477 struct bio_list *bio_list;
d89d8796 1478
73c10101
JA
1479#ifdef CONFIG_BLOCK
1480/* stack plugging */
1481 struct blk_plug *plug;
1482#endif
1483
1da177e4
LT
1484/* VM state */
1485 struct reclaim_state *reclaim_state;
1486
1da177e4
LT
1487 struct backing_dev_info *backing_dev_info;
1488
1489 struct io_context *io_context;
1490
1491 unsigned long ptrace_message;
1492 siginfo_t *last_siginfo; /* For ptrace use. */
7c3ab738 1493 struct task_io_accounting ioac;
8f0ab514 1494#if defined(CONFIG_TASK_XACCT)
1da177e4
LT
1495 u64 acct_rss_mem1; /* accumulated rss usage */
1496 u64 acct_vm_mem1; /* accumulated virtual memory usage */
49b5cf34 1497 cputime_t acct_timexpd; /* stime + utime since last update */
1da177e4
LT
1498#endif
1499#ifdef CONFIG_CPUSETS
58568d2a 1500 nodemask_t mems_allowed; /* Protected by alloc_lock */
cc9a6c87 1501 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
825a46af 1502 int cpuset_mem_spread_rotor;
6adef3eb 1503 int cpuset_slab_spread_rotor;
1da177e4 1504#endif
ddbcc7e8 1505#ifdef CONFIG_CGROUPS
817929ec 1506 /* Control Group info protected by css_set_lock */
2c392b8c 1507 struct css_set __rcu *cgroups;
817929ec
PM
1508 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1509 struct list_head cg_list;
ddbcc7e8 1510#endif
42b2dd0a 1511#ifdef CONFIG_FUTEX
0771dfef 1512 struct robust_list_head __user *robust_list;
34f192c6
IM
1513#ifdef CONFIG_COMPAT
1514 struct compat_robust_list_head __user *compat_robust_list;
1515#endif
c87e2837
IM
1516 struct list_head pi_state_list;
1517 struct futex_pi_state *pi_state_cache;
c7aceaba 1518#endif
cdd6c482 1519#ifdef CONFIG_PERF_EVENTS
8dc85d54 1520 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
cdd6c482
IM
1521 struct mutex perf_event_mutex;
1522 struct list_head perf_event_list;
a63eaf34 1523#endif
8f47b187
TG
1524#ifdef CONFIG_DEBUG_PREEMPT
1525 unsigned long preempt_disable_ip;
1526#endif
c7aceaba 1527#ifdef CONFIG_NUMA
58568d2a 1528 struct mempolicy *mempolicy; /* Protected by alloc_lock */
c7aceaba 1529 short il_next;
207205a2 1530 short pref_node_fork;
42b2dd0a 1531#endif
cbee9f88
PZ
1532#ifdef CONFIG_NUMA_BALANCING
1533 int numa_scan_seq;
cbee9f88 1534 unsigned int numa_scan_period;
598f0ec0 1535 unsigned int numa_scan_period_max;
de1c9ce6 1536 int numa_preferred_nid;
6b9a7460 1537 unsigned long numa_migrate_retry;
cbee9f88 1538 u64 node_stamp; /* migration stamp */
7e2703e6
RR
1539 u64 last_task_numa_placement;
1540 u64 last_sum_exec_runtime;
cbee9f88 1541 struct callback_head numa_work;
f809ca9a 1542
8c8a743c
PZ
1543 struct list_head numa_entry;
1544 struct numa_group *numa_group;
1545
745d6147
MG
1546 /*
1547 * Exponential decaying average of faults on a per-node basis.
1548 * Scheduling placement decisions are made based on the these counts.
1549 * The values remain static for the duration of a PTE scan
1550 */
ff1df896 1551 unsigned long *numa_faults_memory;
83e1d2cd 1552 unsigned long total_numa_faults;
745d6147
MG
1553
1554 /*
1555 * numa_faults_buffer records faults per node during the current
ff1df896
RR
1556 * scan window. When the scan completes, the counts in
1557 * numa_faults_memory decay and these values are copied.
745d6147 1558 */
ff1df896 1559 unsigned long *numa_faults_buffer_memory;
745d6147 1560
50ec8a40
RR
1561 /*
1562 * Track the nodes the process was running on when a NUMA hinting
1563 * fault was incurred.
1564 */
1565 unsigned long *numa_faults_cpu;
1566 unsigned long *numa_faults_buffer_cpu;
1567
04bb2f94
RR
1568 /*
1569 * numa_faults_locality tracks if faults recorded during the last
1570 * scan window were remote/local. The task scan period is adapted
1571 * based on the locality of the faults with different weights
1572 * depending on whether they were shared or private faults
1573 */
1574 unsigned long numa_faults_locality[2];
1575
b32e86b4 1576 unsigned long numa_pages_migrated;
cbee9f88
PZ
1577#endif /* CONFIG_NUMA_BALANCING */
1578
e56d0903 1579 struct rcu_head rcu;
b92ce558
JA
1580
1581 /*
1582 * cache last used pipe for splice
1583 */
1584 struct pipe_inode_info *splice_pipe;
5640f768
ED
1585
1586 struct page_frag task_frag;
1587
ca74e92b
SN
1588#ifdef CONFIG_TASK_DELAY_ACCT
1589 struct task_delay_info *delays;
f4f154fd
AM
1590#endif
1591#ifdef CONFIG_FAULT_INJECTION
1592 int make_it_fail;
ca74e92b 1593#endif
9d823e8f
WF
1594 /*
1595 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1596 * balance_dirty_pages() for some dirty throttling pause
1597 */
1598 int nr_dirtied;
1599 int nr_dirtied_pause;
83712358 1600 unsigned long dirty_paused_when; /* start of a write-and-pause period */
9d823e8f 1601
9745512c
AV
1602#ifdef CONFIG_LATENCYTOP
1603 int latency_record_count;
1604 struct latency_record latency_record[LT_SAVECOUNT];
1605#endif
6976675d
AV
1606 /*
1607 * time slack values; these are used to round up poll() and
1608 * select() etc timeout values. These are in nanoseconds.
1609 */
1610 unsigned long timer_slack_ns;
1611 unsigned long default_timer_slack_ns;
f8d570a4 1612
fb52607a 1613#ifdef CONFIG_FUNCTION_GRAPH_TRACER
3ad2f3fb 1614 /* Index of current stored address in ret_stack */
f201ae23
FW
1615 int curr_ret_stack;
1616 /* Stack of return addresses for return function tracing */
1617 struct ftrace_ret_stack *ret_stack;
8aef2d28
SR
1618 /* time stamp for last schedule */
1619 unsigned long long ftrace_timestamp;
f201ae23
FW
1620 /*
1621 * Number of functions that haven't been traced
1622 * because of depth overrun.
1623 */
1624 atomic_t trace_overrun;
380c4b14
FW
1625 /* Pause for the tracing */
1626 atomic_t tracing_graph_pause;
f201ae23 1627#endif
ea4e2bc4
SR
1628#ifdef CONFIG_TRACING
1629 /* state flags for use by tracers */
1630 unsigned long trace;
b1cff0ad 1631 /* bitmask and counter of trace recursion */
261842b7
SR
1632 unsigned long trace_recursion;
1633#endif /* CONFIG_TRACING */
c255a458 1634#ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
0e9d92f2 1635 unsigned int memcg_kmem_skip_account;
519e5247 1636 struct memcg_oom_info {
49426420
JW
1637 struct mem_cgroup *memcg;
1638 gfp_t gfp_mask;
1639 int order;
519e5247
JW
1640 unsigned int may_oom:1;
1641 } memcg_oom;
569b846d 1642#endif
0326f5a9
SD
1643#ifdef CONFIG_UPROBES
1644 struct uprobe_task *utask;
0326f5a9 1645#endif
cafe5635
KO
1646#if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1647 unsigned int sequential_io;
1648 unsigned int sequential_io_avg;
1649#endif
1da177e4
LT
1650};
1651
76e6eee0 1652/* Future-safe accessor for struct task_struct's cpus_allowed. */
a4636818 1653#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
76e6eee0 1654
6688cc05
PZ
1655#define TNF_MIGRATED 0x01
1656#define TNF_NO_GROUP 0x02
dabe1d99 1657#define TNF_SHARED 0x04
04bb2f94 1658#define TNF_FAULT_LOCAL 0x08
6688cc05 1659
cbee9f88 1660#ifdef CONFIG_NUMA_BALANCING
6688cc05 1661extern void task_numa_fault(int last_node, int node, int pages, int flags);
e29cf08b 1662extern pid_t task_numa_group_id(struct task_struct *p);
1a687c2e 1663extern void set_numabalancing_state(bool enabled);
82727018 1664extern void task_numa_free(struct task_struct *p);
10f39042
RR
1665extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1666 int src_nid, int dst_cpu);
cbee9f88 1667#else
ac8e895b 1668static inline void task_numa_fault(int last_node, int node, int pages,
6688cc05 1669 int flags)
cbee9f88
PZ
1670{
1671}
e29cf08b
MG
1672static inline pid_t task_numa_group_id(struct task_struct *p)
1673{
1674 return 0;
1675}
1a687c2e
MG
1676static inline void set_numabalancing_state(bool enabled)
1677{
1678}
82727018
RR
1679static inline void task_numa_free(struct task_struct *p)
1680{
1681}
10f39042
RR
1682static inline bool should_numa_migrate_memory(struct task_struct *p,
1683 struct page *page, int src_nid, int dst_cpu)
1684{
1685 return true;
1686}
cbee9f88
PZ
1687#endif
1688
e868171a 1689static inline struct pid *task_pid(struct task_struct *task)
22c935f4
EB
1690{
1691 return task->pids[PIDTYPE_PID].pid;
1692}
1693
e868171a 1694static inline struct pid *task_tgid(struct task_struct *task)
22c935f4
EB
1695{
1696 return task->group_leader->pids[PIDTYPE_PID].pid;
1697}
1698
6dda81f4
ON
1699/*
1700 * Without tasklist or rcu lock it is not safe to dereference
1701 * the result of task_pgrp/task_session even if task == current,
1702 * we can race with another thread doing sys_setsid/sys_setpgid.
1703 */
e868171a 1704static inline struct pid *task_pgrp(struct task_struct *task)
22c935f4
EB
1705{
1706 return task->group_leader->pids[PIDTYPE_PGID].pid;
1707}
1708
e868171a 1709static inline struct pid *task_session(struct task_struct *task)
22c935f4
EB
1710{
1711 return task->group_leader->pids[PIDTYPE_SID].pid;
1712}
1713
7af57294
PE
1714struct pid_namespace;
1715
1716/*
1717 * the helpers to get the task's different pids as they are seen
1718 * from various namespaces
1719 *
1720 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
44c4e1b2
EB
1721 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1722 * current.
7af57294
PE
1723 * task_xid_nr_ns() : id seen from the ns specified;
1724 *
1725 * set_task_vxid() : assigns a virtual id to a task;
1726 *
7af57294
PE
1727 * see also pid_nr() etc in include/linux/pid.h
1728 */
52ee2dfd
ON
1729pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1730 struct pid_namespace *ns);
7af57294 1731
e868171a 1732static inline pid_t task_pid_nr(struct task_struct *tsk)
7af57294
PE
1733{
1734 return tsk->pid;
1735}
1736
52ee2dfd
ON
1737static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1738 struct pid_namespace *ns)
1739{
1740 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1741}
7af57294
PE
1742
1743static inline pid_t task_pid_vnr(struct task_struct *tsk)
1744{
52ee2dfd 1745 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
7af57294
PE
1746}
1747
1748
e868171a 1749static inline pid_t task_tgid_nr(struct task_struct *tsk)
7af57294
PE
1750{
1751 return tsk->tgid;
1752}
1753
2f2a3a46 1754pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
7af57294
PE
1755
1756static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1757{
1758 return pid_vnr(task_tgid(tsk));
1759}
1760
1761
80e0b6e8 1762static inline int pid_alive(const struct task_struct *p);
ad36d282
RGB
1763static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1764{
1765 pid_t pid = 0;
1766
1767 rcu_read_lock();
1768 if (pid_alive(tsk))
1769 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1770 rcu_read_unlock();
1771
1772 return pid;
1773}
1774
1775static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1776{
1777 return task_ppid_nr_ns(tsk, &init_pid_ns);
1778}
1779
52ee2dfd
ON
1780static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1781 struct pid_namespace *ns)
7af57294 1782{
52ee2dfd 1783 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
7af57294
PE
1784}
1785
7af57294
PE
1786static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1787{
52ee2dfd 1788 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
7af57294
PE
1789}
1790
1791
52ee2dfd
ON
1792static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1793 struct pid_namespace *ns)
7af57294 1794{
52ee2dfd 1795 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
7af57294
PE
1796}
1797
7af57294
PE
1798static inline pid_t task_session_vnr(struct task_struct *tsk)
1799{
52ee2dfd 1800 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
7af57294
PE
1801}
1802
1b0f7ffd
ON
1803/* obsolete, do not use */
1804static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1805{
1806 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1807}
7af57294 1808
1da177e4
LT
1809/**
1810 * pid_alive - check that a task structure is not stale
1811 * @p: Task structure to be checked.
1812 *
1813 * Test if a process is not yet dead (at most zombie state)
1814 * If pid_alive fails, then pointers within the task structure
1815 * can be stale and must not be dereferenced.
e69f6186
YB
1816 *
1817 * Return: 1 if the process is alive. 0 otherwise.
1da177e4 1818 */
ad36d282 1819static inline int pid_alive(const struct task_struct *p)
1da177e4 1820{
92476d7f 1821 return p->pids[PIDTYPE_PID].pid != NULL;
1da177e4
LT
1822}
1823
f400e198 1824/**
b460cbc5 1825 * is_global_init - check if a task structure is init
3260259f
H
1826 * @tsk: Task structure to be checked.
1827 *
1828 * Check if a task structure is the first user space task the kernel created.
e69f6186
YB
1829 *
1830 * Return: 1 if the task structure is init. 0 otherwise.
b460cbc5 1831 */
e868171a 1832static inline int is_global_init(struct task_struct *tsk)
b461cc03
PE
1833{
1834 return tsk->pid == 1;
1835}
b460cbc5 1836
9ec52099
CLG
1837extern struct pid *cad_pid;
1838
1da177e4 1839extern void free_task(struct task_struct *tsk);
1da177e4 1840#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
e56d0903 1841
158d9ebd 1842extern void __put_task_struct(struct task_struct *t);
e56d0903
IM
1843
1844static inline void put_task_struct(struct task_struct *t)
1845{
1846 if (atomic_dec_and_test(&t->usage))
8c7904a0 1847 __put_task_struct(t);
e56d0903 1848}
1da177e4 1849
6a61671b
FW
1850#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1851extern void task_cputime(struct task_struct *t,
1852 cputime_t *utime, cputime_t *stime);
1853extern void task_cputime_scaled(struct task_struct *t,
1854 cputime_t *utimescaled, cputime_t *stimescaled);
1855extern cputime_t task_gtime(struct task_struct *t);
1856#else
6fac4829
FW
1857static inline void task_cputime(struct task_struct *t,
1858 cputime_t *utime, cputime_t *stime)
1859{
1860 if (utime)
1861 *utime = t->utime;
1862 if (stime)
1863 *stime = t->stime;
1864}
1865
1866static inline void task_cputime_scaled(struct task_struct *t,
1867 cputime_t *utimescaled,
1868 cputime_t *stimescaled)
1869{
1870 if (utimescaled)
1871 *utimescaled = t->utimescaled;
1872 if (stimescaled)
1873 *stimescaled = t->stimescaled;
1874}
6a61671b
FW
1875
1876static inline cputime_t task_gtime(struct task_struct *t)
1877{
1878 return t->gtime;
1879}
1880#endif
e80d0a1a
FW
1881extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1882extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
49048622 1883
1da177e4
LT
1884/*
1885 * Per process flags
1886 */
1da177e4 1887#define PF_EXITING 0x00000004 /* getting shut down */
778e9a9c 1888#define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
94886b84 1889#define PF_VCPU 0x00000010 /* I'm a virtual CPU */
21aa9af0 1890#define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1da177e4 1891#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
4db96cf0 1892#define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1da177e4
LT
1893#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1894#define PF_DUMPCORE 0x00000200 /* dumped core */
1895#define PF_SIGNALED 0x00000400 /* killed by a signal */
1896#define PF_MEMALLOC 0x00000800 /* Allocating memory */
72fa5997 1897#define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1da177e4 1898#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
774a1221 1899#define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1da177e4
LT
1900#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1901#define PF_FROZEN 0x00010000 /* frozen for system suspend */
1902#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1903#define PF_KSWAPD 0x00040000 /* I am kswapd */
21caf2fc 1904#define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1da177e4 1905#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
246bb0b1 1906#define PF_KTHREAD 0x00200000 /* I am a kernel thread */
b31dc66a
JA
1907#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1908#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1909#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1910#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
14a40ffc 1911#define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
4db96cf0 1912#define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
61a87122 1913#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
58a69cb4 1914#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2b44c4db 1915#define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1da177e4
LT
1916
1917/*
1918 * Only the _current_ task can read/write to tsk->flags, but other
1919 * tasks can access tsk->flags in readonly mode for example
1920 * with tsk_used_math (like during threaded core dumping).
1921 * There is however an exception to this rule during ptrace
1922 * or during fork: the ptracer task is allowed to write to the
1923 * child->flags of its traced child (same goes for fork, the parent
1924 * can write to the child->flags), because we're guaranteed the
1925 * child is not running and in turn not changing child->flags
1926 * at the same time the parent does it.
1927 */
1928#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1929#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1930#define clear_used_math() clear_stopped_child_used_math(current)
1931#define set_used_math() set_stopped_child_used_math(current)
1932#define conditional_stopped_child_used_math(condition, child) \
1933 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1934#define conditional_used_math(condition) \
1935 conditional_stopped_child_used_math(condition, current)
1936#define copy_to_stopped_child_used_math(child) \
1937 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1938/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1939#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1940#define used_math() tsk_used_math(current)
1941
21caf2fc
ML
1942/* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1943static inline gfp_t memalloc_noio_flags(gfp_t flags)
1944{
1945 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1946 flags &= ~__GFP_IO;
1947 return flags;
1948}
1949
1950static inline unsigned int memalloc_noio_save(void)
1951{
1952 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1953 current->flags |= PF_MEMALLOC_NOIO;
1954 return flags;
1955}
1956
1957static inline void memalloc_noio_restore(unsigned int flags)
1958{
1959 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1960}
1961
1d4457f9
KC
1962/* Per-process atomic flags. */
1963#define PFA_NO_NEW_PRIVS 0x00000001 /* May not gain new privileges. */
1964
1965static inline bool task_no_new_privs(struct task_struct *p)
1966{
1967 return test_bit(PFA_NO_NEW_PRIVS, &p->atomic_flags);
1968}
1969
1970static inline void task_set_no_new_privs(struct task_struct *p)
1971{
1972 set_bit(PFA_NO_NEW_PRIVS, &p->atomic_flags);
1973}
1974
e5c1902e 1975/*
a8f072c1 1976 * task->jobctl flags
e5c1902e 1977 */
a8f072c1 1978#define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
e5c1902e 1979
a8f072c1
TH
1980#define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1981#define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1982#define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
73ddff2b 1983#define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
fb1d910c 1984#define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
a8f072c1 1985#define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
544b2c91 1986#define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
a8f072c1
TH
1987
1988#define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1989#define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1990#define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
73ddff2b 1991#define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
fb1d910c 1992#define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
a8f072c1 1993#define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
544b2c91 1994#define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
a8f072c1 1995
fb1d910c 1996#define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
73ddff2b 1997#define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
3759a0d9 1998
7dd3db54
TH
1999extern bool task_set_jobctl_pending(struct task_struct *task,
2000 unsigned int mask);
73ddff2b 2001extern void task_clear_jobctl_trapping(struct task_struct *task);
3759a0d9
TH
2002extern void task_clear_jobctl_pending(struct task_struct *task,
2003 unsigned int mask);
39efa3ef 2004
a57eb940 2005#ifdef CONFIG_PREEMPT_RCU
f41d911f
PM
2006
2007#define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1aa03f11 2008#define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
f41d911f
PM
2009
2010static inline void rcu_copy_process(struct task_struct *p)
2011{
2012 p->rcu_read_lock_nesting = 0;
2013 p->rcu_read_unlock_special = 0;
a57eb940 2014#ifdef CONFIG_TREE_PREEMPT_RCU
dd5d19ba 2015 p->rcu_blocked_node = NULL;
24278d14 2016#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
f41d911f
PM
2017 INIT_LIST_HEAD(&p->rcu_node_entry);
2018}
2019
f41d911f
PM
2020#else
2021
2022static inline void rcu_copy_process(struct task_struct *p)
2023{
2024}
2025
2026#endif
2027
907aed48
MG
2028static inline void tsk_restore_flags(struct task_struct *task,
2029 unsigned long orig_flags, unsigned long flags)
2030{
2031 task->flags &= ~flags;
2032 task->flags |= orig_flags & flags;
2033}
2034
1da177e4 2035#ifdef CONFIG_SMP
1e1b6c51
KM
2036extern void do_set_cpus_allowed(struct task_struct *p,
2037 const struct cpumask *new_mask);
2038
cd8ba7cd 2039extern int set_cpus_allowed_ptr(struct task_struct *p,
96f874e2 2040 const struct cpumask *new_mask);
1da177e4 2041#else
1e1b6c51
KM
2042static inline void do_set_cpus_allowed(struct task_struct *p,
2043 const struct cpumask *new_mask)
2044{
2045}
cd8ba7cd 2046static inline int set_cpus_allowed_ptr(struct task_struct *p,
96f874e2 2047 const struct cpumask *new_mask)
1da177e4 2048{
96f874e2 2049 if (!cpumask_test_cpu(0, new_mask))
1da177e4
LT
2050 return -EINVAL;
2051 return 0;
2052}
2053#endif
e0ad9556 2054
3451d024 2055#ifdef CONFIG_NO_HZ_COMMON
5167e8d5
PZ
2056void calc_load_enter_idle(void);
2057void calc_load_exit_idle(void);
2058#else
2059static inline void calc_load_enter_idle(void) { }
2060static inline void calc_load_exit_idle(void) { }
3451d024 2061#endif /* CONFIG_NO_HZ_COMMON */
5167e8d5 2062
e0ad9556 2063#ifndef CONFIG_CPUMASK_OFFSTACK
cd8ba7cd
MT
2064static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2065{
2066 return set_cpus_allowed_ptr(p, &new_mask);
2067}
e0ad9556 2068#endif
1da177e4 2069
b342501c 2070/*
c676329a
PZ
2071 * Do not use outside of architecture code which knows its limitations.
2072 *
2073 * sched_clock() has no promise of monotonicity or bounded drift between
2074 * CPUs, use (which you should not) requires disabling IRQs.
2075 *
2076 * Please use one of the three interfaces below.
b342501c 2077 */
1bbfa6f2 2078extern unsigned long long notrace sched_clock(void);
c676329a 2079/*
489a71b0 2080 * See the comment in kernel/sched/clock.c
c676329a
PZ
2081 */
2082extern u64 cpu_clock(int cpu);
2083extern u64 local_clock(void);
2084extern u64 sched_clock_cpu(int cpu);
2085
e436d800 2086
c1955a3d 2087extern void sched_clock_init(void);
3e51f33f 2088
c1955a3d 2089#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
3e51f33f
PZ
2090static inline void sched_clock_tick(void)
2091{
2092}
2093
2094static inline void sched_clock_idle_sleep_event(void)
2095{
2096}
2097
2098static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2099{
2100}
2101#else
c676329a
PZ
2102/*
2103 * Architectures can set this to 1 if they have specified
2104 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2105 * but then during bootup it turns out that sched_clock()
2106 * is reliable after all:
2107 */
35af99e6
PZ
2108extern int sched_clock_stable(void);
2109extern void set_sched_clock_stable(void);
2110extern void clear_sched_clock_stable(void);
c676329a 2111
3e51f33f
PZ
2112extern void sched_clock_tick(void);
2113extern void sched_clock_idle_sleep_event(void);
2114extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2115#endif
2116
b52bfee4
VP
2117#ifdef CONFIG_IRQ_TIME_ACCOUNTING
2118/*
2119 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2120 * The reason for this explicit opt-in is not to have perf penalty with
2121 * slow sched_clocks.
2122 */
2123extern void enable_sched_clock_irqtime(void);
2124extern void disable_sched_clock_irqtime(void);
2125#else
2126static inline void enable_sched_clock_irqtime(void) {}
2127static inline void disable_sched_clock_irqtime(void) {}
2128#endif
2129
36c8b586 2130extern unsigned long long
41b86e9c 2131task_sched_runtime(struct task_struct *task);
1da177e4
LT
2132
2133/* sched_exec is called by processes performing an exec */
2134#ifdef CONFIG_SMP
2135extern void sched_exec(void);
2136#else
2137#define sched_exec() {}
2138#endif
2139
2aa44d05
IM
2140extern void sched_clock_idle_sleep_event(void);
2141extern void sched_clock_idle_wakeup_event(u64 delta_ns);
bb29ab26 2142
1da177e4
LT
2143#ifdef CONFIG_HOTPLUG_CPU
2144extern void idle_task_exit(void);
2145#else
2146static inline void idle_task_exit(void) {}
2147#endif
2148
3451d024 2149#if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
1c20091e 2150extern void wake_up_nohz_cpu(int cpu);
06d8308c 2151#else
1c20091e 2152static inline void wake_up_nohz_cpu(int cpu) { }
06d8308c
TG
2153#endif
2154
ce831b38
FW
2155#ifdef CONFIG_NO_HZ_FULL
2156extern bool sched_can_stop_tick(void);
265f22a9 2157extern u64 scheduler_tick_max_deferment(void);
ce831b38
FW
2158#else
2159static inline bool sched_can_stop_tick(void) { return false; }
06d8308c
TG
2160#endif
2161
5091faa4 2162#ifdef CONFIG_SCHED_AUTOGROUP
5091faa4
MG
2163extern void sched_autogroup_create_attach(struct task_struct *p);
2164extern void sched_autogroup_detach(struct task_struct *p);
2165extern void sched_autogroup_fork(struct signal_struct *sig);
2166extern void sched_autogroup_exit(struct signal_struct *sig);
2167#ifdef CONFIG_PROC_FS
2168extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2e5b5b3a 2169extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
5091faa4
MG
2170#endif
2171#else
2172static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2173static inline void sched_autogroup_detach(struct task_struct *p) { }
2174static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2175static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2176#endif
2177
fa93384f 2178extern int yield_to(struct task_struct *p, bool preempt);
36c8b586
IM
2179extern void set_user_nice(struct task_struct *p, long nice);
2180extern int task_prio(const struct task_struct *p);
d0ea0268
DY
2181/**
2182 * task_nice - return the nice value of a given task.
2183 * @p: the task in question.
2184 *
2185 * Return: The nice value [ -20 ... 0 ... 19 ].
2186 */
2187static inline int task_nice(const struct task_struct *p)
2188{
2189 return PRIO_TO_NICE((p)->static_prio);
2190}
36c8b586
IM
2191extern int can_nice(const struct task_struct *p, const int nice);
2192extern int task_curr(const struct task_struct *p);
1da177e4 2193extern int idle_cpu(int cpu);
fe7de49f
KM
2194extern int sched_setscheduler(struct task_struct *, int,
2195 const struct sched_param *);
961ccddd 2196extern int sched_setscheduler_nocheck(struct task_struct *, int,
fe7de49f 2197 const struct sched_param *);
d50dde5a
DF
2198extern int sched_setattr(struct task_struct *,
2199 const struct sched_attr *);
36c8b586 2200extern struct task_struct *idle_task(int cpu);
c4f30608
PM
2201/**
2202 * is_idle_task - is the specified task an idle task?
fa757281 2203 * @p: the task in question.
e69f6186
YB
2204 *
2205 * Return: 1 if @p is an idle task. 0 otherwise.
c4f30608 2206 */
7061ca3b 2207static inline bool is_idle_task(const struct task_struct *p)
c4f30608
PM
2208{
2209 return p->pid == 0;
2210}
36c8b586
IM
2211extern struct task_struct *curr_task(int cpu);
2212extern void set_curr_task(int cpu, struct task_struct *p);
1da177e4
LT
2213
2214void yield(void);
2215
2216/*
2217 * The default (Linux) execution domain.
2218 */
2219extern struct exec_domain default_exec_domain;
2220
2221union thread_union {
2222 struct thread_info thread_info;
2223 unsigned long stack[THREAD_SIZE/sizeof(long)];
2224};
2225
2226#ifndef __HAVE_ARCH_KSTACK_END
2227static inline int kstack_end(void *addr)
2228{
2229 /* Reliable end of stack detection:
2230 * Some APM bios versions misalign the stack
2231 */
2232 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2233}
2234#endif
2235
2236extern union thread_union init_thread_union;
2237extern struct task_struct init_task;
2238
2239extern struct mm_struct init_mm;
2240
198fe21b
PE
2241extern struct pid_namespace init_pid_ns;
2242
2243/*
2244 * find a task by one of its numerical ids
2245 *
198fe21b
PE
2246 * find_task_by_pid_ns():
2247 * finds a task by its pid in the specified namespace
228ebcbe
PE
2248 * find_task_by_vpid():
2249 * finds a task by its virtual pid
198fe21b 2250 *
e49859e7 2251 * see also find_vpid() etc in include/linux/pid.h
198fe21b
PE
2252 */
2253
228ebcbe
PE
2254extern struct task_struct *find_task_by_vpid(pid_t nr);
2255extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2256 struct pid_namespace *ns);
198fe21b 2257
1da177e4 2258/* per-UID process charging. */
7b44ab97 2259extern struct user_struct * alloc_uid(kuid_t);
1da177e4
LT
2260static inline struct user_struct *get_uid(struct user_struct *u)
2261{
2262 atomic_inc(&u->__count);
2263 return u;
2264}
2265extern void free_uid(struct user_struct *);
1da177e4
LT
2266
2267#include <asm/current.h>
2268
f0af911a 2269extern void xtime_update(unsigned long ticks);
1da177e4 2270
b3c97528
HH
2271extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2272extern int wake_up_process(struct task_struct *tsk);
3e51e3ed 2273extern void wake_up_new_task(struct task_struct *tsk);
1da177e4
LT
2274#ifdef CONFIG_SMP
2275 extern void kick_process(struct task_struct *tsk);
2276#else
2277 static inline void kick_process(struct task_struct *tsk) { }
2278#endif
aab03e05 2279extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
ad46c2c4 2280extern void sched_dead(struct task_struct *p);
1da177e4 2281
1da177e4
LT
2282extern void proc_caches_init(void);
2283extern void flush_signals(struct task_struct *);
3bcac026 2284extern void __flush_signals(struct task_struct *);
10ab825b 2285extern void ignore_signals(struct task_struct *);
1da177e4
LT
2286extern void flush_signal_handlers(struct task_struct *, int force_default);
2287extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2288
2289static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2290{
2291 unsigned long flags;
2292 int ret;
2293
2294 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2295 ret = dequeue_signal(tsk, mask, info);
2296 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2297
2298 return ret;
53c8f9f1 2299}
1da177e4
LT
2300
2301extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2302 sigset_t *mask);
2303extern void unblock_all_signals(void);
2304extern void release_task(struct task_struct * p);
2305extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1da177e4
LT
2306extern int force_sigsegv(int, struct task_struct *);
2307extern int force_sig_info(int, struct siginfo *, struct task_struct *);
c4b92fc1 2308extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
c4b92fc1 2309extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
d178bc3a
SH
2310extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2311 const struct cred *, u32);
c4b92fc1
EB
2312extern int kill_pgrp(struct pid *pid, int sig, int priv);
2313extern int kill_pid(struct pid *pid, int sig, int priv);
c3de4b38 2314extern int kill_proc_info(int, struct siginfo *, pid_t);
86773473 2315extern __must_check bool do_notify_parent(struct task_struct *, int);
a7f0765e 2316extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
1da177e4 2317extern void force_sig(int, struct task_struct *);
1da177e4 2318extern int send_sig(int, struct task_struct *, int);
09faef11 2319extern int zap_other_threads(struct task_struct *p);
1da177e4
LT
2320extern struct sigqueue *sigqueue_alloc(void);
2321extern void sigqueue_free(struct sigqueue *);
ac5c2153 2322extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
9ac95f2f 2323extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1da177e4 2324
51a7b448
AV
2325static inline void restore_saved_sigmask(void)
2326{
2327 if (test_and_clear_restore_sigmask())
77097ae5 2328 __set_current_blocked(&current->saved_sigmask);
51a7b448
AV
2329}
2330
b7f9a11a
AV
2331static inline sigset_t *sigmask_to_save(void)
2332{
2333 sigset_t *res = &current->blocked;
2334 if (unlikely(test_restore_sigmask()))
2335 res = &current->saved_sigmask;
2336 return res;
2337}
2338
9ec52099
CLG
2339static inline int kill_cad_pid(int sig, int priv)
2340{
2341 return kill_pid(cad_pid, sig, priv);
2342}
2343
1da177e4
LT
2344/* These can be the second arg to send_sig_info/send_group_sig_info. */
2345#define SEND_SIG_NOINFO ((struct siginfo *) 0)
2346#define SEND_SIG_PRIV ((struct siginfo *) 1)
2347#define SEND_SIG_FORCED ((struct siginfo *) 2)
2348
2a855dd0
SAS
2349/*
2350 * True if we are on the alternate signal stack.
2351 */
1da177e4
LT
2352static inline int on_sig_stack(unsigned long sp)
2353{
2a855dd0
SAS
2354#ifdef CONFIG_STACK_GROWSUP
2355 return sp >= current->sas_ss_sp &&
2356 sp - current->sas_ss_sp < current->sas_ss_size;
2357#else
2358 return sp > current->sas_ss_sp &&
2359 sp - current->sas_ss_sp <= current->sas_ss_size;
2360#endif
1da177e4
LT
2361}
2362
2363static inline int sas_ss_flags(unsigned long sp)
2364{
72f15c03
RW
2365 if (!current->sas_ss_size)
2366 return SS_DISABLE;
2367
2368 return on_sig_stack(sp) ? SS_ONSTACK : 0;
1da177e4
LT
2369}
2370
5a1b98d3
AV
2371static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2372{
2373 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2374#ifdef CONFIG_STACK_GROWSUP
2375 return current->sas_ss_sp;
2376#else
2377 return current->sas_ss_sp + current->sas_ss_size;
2378#endif
2379 return sp;
2380}
2381
1da177e4
LT
2382/*
2383 * Routines for handling mm_structs
2384 */
2385extern struct mm_struct * mm_alloc(void);
2386
2387/* mmdrop drops the mm and the page tables */
b3c97528 2388extern void __mmdrop(struct mm_struct *);
1da177e4
LT
2389static inline void mmdrop(struct mm_struct * mm)
2390{
6fb43d7b 2391 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1da177e4
LT
2392 __mmdrop(mm);
2393}
2394
2395/* mmput gets rid of the mappings and all user-space */
2396extern void mmput(struct mm_struct *);
2397/* Grab a reference to a task's mm, if it is not already going away */
2398extern struct mm_struct *get_task_mm(struct task_struct *task);
8cdb878d
CY
2399/*
2400 * Grab a reference to a task's mm, if it is not already going away
2401 * and ptrace_may_access with the mode parameter passed to it
2402 * succeeds.
2403 */
2404extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
1da177e4
LT
2405/* Remove the current tasks stale references to the old mm_struct */
2406extern void mm_release(struct task_struct *, struct mm_struct *);
2407
6f2c55b8 2408extern int copy_thread(unsigned long, unsigned long, unsigned long,
afa86fc4 2409 struct task_struct *);
1da177e4
LT
2410extern void flush_thread(void);
2411extern void exit_thread(void);
2412
1da177e4 2413extern void exit_files(struct task_struct *);
a7e5328a 2414extern void __cleanup_sighand(struct sighand_struct *);
cbaffba1 2415
1da177e4 2416extern void exit_itimers(struct signal_struct *);
cbaffba1 2417extern void flush_itimer_signals(void);
1da177e4 2418
9402c95f 2419extern void do_group_exit(int);
1da177e4 2420
c4ad8f98 2421extern int do_execve(struct filename *,
d7627467 2422 const char __user * const __user *,
da3d4c5f 2423 const char __user * const __user *);
e80d6661 2424extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
36c8b586 2425struct task_struct *fork_idle(int);
2aa3a7f8 2426extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
1da177e4 2427
82b89778
AH
2428extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2429static inline void set_task_comm(struct task_struct *tsk, const char *from)
2430{
2431 __set_task_comm(tsk, from, false);
2432}
59714d65 2433extern char *get_task_comm(char *to, struct task_struct *tsk);
1da177e4
LT
2434
2435#ifdef CONFIG_SMP
317f3941 2436void scheduler_ipi(void);
85ba2d86 2437extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1da177e4 2438#else
184748cc 2439static inline void scheduler_ipi(void) { }
85ba2d86
RM
2440static inline unsigned long wait_task_inactive(struct task_struct *p,
2441 long match_state)
2442{
2443 return 1;
2444}
1da177e4
LT
2445#endif
2446
05725f7e
JP
2447#define next_task(p) \
2448 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
1da177e4
LT
2449
2450#define for_each_process(p) \
2451 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2452
5bb459bb 2453extern bool current_is_single_threaded(void);
d84f4f99 2454
1da177e4
LT
2455/*
2456 * Careful: do_each_thread/while_each_thread is a double loop so
2457 * 'break' will not work as expected - use goto instead.
2458 */
2459#define do_each_thread(g, t) \
2460 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2461
2462#define while_each_thread(g, t) \
2463 while ((t = next_thread(t)) != g)
2464
0c740d0a
ON
2465#define __for_each_thread(signal, t) \
2466 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2467
2468#define for_each_thread(p, t) \
2469 __for_each_thread((p)->signal, t)
2470
2471/* Careful: this is a double loop, 'break' won't work as expected. */
2472#define for_each_process_thread(p, t) \
2473 for_each_process(p) for_each_thread(p, t)
2474
7e49827c
ON
2475static inline int get_nr_threads(struct task_struct *tsk)
2476{
b3ac022c 2477 return tsk->signal->nr_threads;
7e49827c
ON
2478}
2479
087806b1
ON
2480static inline bool thread_group_leader(struct task_struct *p)
2481{
2482 return p->exit_signal >= 0;
2483}
1da177e4 2484
0804ef4b
EB
2485/* Do to the insanities of de_thread it is possible for a process
2486 * to have the pid of the thread group leader without actually being
2487 * the thread group leader. For iteration through the pids in proc
2488 * all we care about is that we have a task with the appropriate
2489 * pid, we don't actually care if we have the right task.
2490 */
e1403b8e 2491static inline bool has_group_leader_pid(struct task_struct *p)
0804ef4b 2492{
e1403b8e 2493 return task_pid(p) == p->signal->leader_pid;
0804ef4b
EB
2494}
2495
bac0abd6 2496static inline
e1403b8e 2497bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
bac0abd6 2498{
e1403b8e 2499 return p1->signal == p2->signal;
bac0abd6
PE
2500}
2501
36c8b586 2502static inline struct task_struct *next_thread(const struct task_struct *p)
47e65328 2503{
05725f7e
JP
2504 return list_entry_rcu(p->thread_group.next,
2505 struct task_struct, thread_group);
47e65328
ON
2506}
2507
e868171a 2508static inline int thread_group_empty(struct task_struct *p)
1da177e4 2509{
47e65328 2510 return list_empty(&p->thread_group);
1da177e4
LT
2511}
2512
2513#define delay_group_leader(p) \
2514 (thread_group_leader(p) && !thread_group_empty(p))
2515
1da177e4 2516/*
260ea101 2517 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
22e2c507 2518 * subscriptions and synchronises with wait4(). Also used in procfs. Also
ddbcc7e8 2519 * pins the final release of task.io_context. Also protects ->cpuset and
d68b46fe 2520 * ->cgroup.subsys[]. And ->vfork_done.
1da177e4
LT
2521 *
2522 * Nests both inside and outside of read_lock(&tasklist_lock).
2523 * It must not be nested with write_lock_irq(&tasklist_lock),
2524 * neither inside nor outside.
2525 */
2526static inline void task_lock(struct task_struct *p)
2527{
2528 spin_lock(&p->alloc_lock);
2529}
2530
2531static inline void task_unlock(struct task_struct *p)
2532{
2533 spin_unlock(&p->alloc_lock);
2534}
2535
b8ed374e 2536extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
f63ee72e
ON
2537 unsigned long *flags);
2538
9388dc30
AV
2539static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2540 unsigned long *flags)
2541{
2542 struct sighand_struct *ret;
2543
2544 ret = __lock_task_sighand(tsk, flags);
2545 (void)__cond_lock(&tsk->sighand->siglock, ret);
2546 return ret;
2547}
b8ed374e 2548
f63ee72e
ON
2549static inline void unlock_task_sighand(struct task_struct *tsk,
2550 unsigned long *flags)
2551{
2552 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2553}
2554
4714d1d3 2555#ifdef CONFIG_CGROUPS
257058ae 2556static inline void threadgroup_change_begin(struct task_struct *tsk)
4714d1d3 2557{
257058ae 2558 down_read(&tsk->signal->group_rwsem);
4714d1d3 2559}
257058ae 2560static inline void threadgroup_change_end(struct task_struct *tsk)
4714d1d3 2561{
257058ae 2562 up_read(&tsk->signal->group_rwsem);
4714d1d3 2563}
77e4ef99
TH
2564
2565/**
2566 * threadgroup_lock - lock threadgroup
2567 * @tsk: member task of the threadgroup to lock
2568 *
2569 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2570 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
e56fb287
ON
2571 * change ->group_leader/pid. This is useful for cases where the threadgroup
2572 * needs to stay stable across blockable operations.
77e4ef99
TH
2573 *
2574 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2575 * synchronization. While held, no new task will be added to threadgroup
2576 * and no existing live task will have its PF_EXITING set.
2577 *
e56fb287
ON
2578 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2579 * sub-thread becomes a new leader.
77e4ef99 2580 */
257058ae 2581static inline void threadgroup_lock(struct task_struct *tsk)
4714d1d3 2582{
257058ae 2583 down_write(&tsk->signal->group_rwsem);
4714d1d3 2584}
77e4ef99
TH
2585
2586/**
2587 * threadgroup_unlock - unlock threadgroup
2588 * @tsk: member task of the threadgroup to unlock
2589 *
2590 * Reverse threadgroup_lock().
2591 */
257058ae 2592static inline void threadgroup_unlock(struct task_struct *tsk)
4714d1d3 2593{
257058ae 2594 up_write(&tsk->signal->group_rwsem);
4714d1d3
BB
2595}
2596#else
257058ae
TH
2597static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2598static inline void threadgroup_change_end(struct task_struct *tsk) {}
2599static inline void threadgroup_lock(struct task_struct *tsk) {}
2600static inline void threadgroup_unlock(struct task_struct *tsk) {}
4714d1d3
BB
2601#endif
2602
f037360f
AV
2603#ifndef __HAVE_THREAD_FUNCTIONS
2604
f7e4217b
RZ
2605#define task_thread_info(task) ((struct thread_info *)(task)->stack)
2606#define task_stack_page(task) ((task)->stack)
a1261f54 2607
10ebffde
AV
2608static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2609{
2610 *task_thread_info(p) = *task_thread_info(org);
2611 task_thread_info(p)->task = p;
2612}
2613
2614static inline unsigned long *end_of_stack(struct task_struct *p)
2615{
f7e4217b 2616 return (unsigned long *)(task_thread_info(p) + 1);
10ebffde
AV
2617}
2618
f037360f
AV
2619#endif
2620
8b05c7e6
FT
2621static inline int object_is_on_stack(void *obj)
2622{
2623 void *stack = task_stack_page(current);
2624
2625 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2626}
2627
8c9843e5
BH
2628extern void thread_info_cache_init(void);
2629
7c9f8861
ES
2630#ifdef CONFIG_DEBUG_STACK_USAGE
2631static inline unsigned long stack_not_used(struct task_struct *p)
2632{
2633 unsigned long *n = end_of_stack(p);
2634
2635 do { /* Skip over canary */
2636 n++;
2637 } while (!*n);
2638
2639 return (unsigned long)n - (unsigned long)end_of_stack(p);
2640}
2641#endif
d4311ff1 2642extern void set_task_stack_end_magic(struct task_struct *tsk);
7c9f8861 2643
1da177e4
LT
2644/* set thread flags in other task's structures
2645 * - see asm/thread_info.h for TIF_xxxx flags available
2646 */
2647static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2648{
a1261f54 2649 set_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
2650}
2651
2652static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2653{
a1261f54 2654 clear_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
2655}
2656
2657static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2658{
a1261f54 2659 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
2660}
2661
2662static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2663{
a1261f54 2664 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
2665}
2666
2667static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2668{
a1261f54 2669 return test_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
2670}
2671
2672static inline void set_tsk_need_resched(struct task_struct *tsk)
2673{
2674 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2675}
2676
2677static inline void clear_tsk_need_resched(struct task_struct *tsk)
2678{
2679 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2680}
2681
8ae121ac
GH
2682static inline int test_tsk_need_resched(struct task_struct *tsk)
2683{
2684 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2685}
2686
690cc3ff
EB
2687static inline int restart_syscall(void)
2688{
2689 set_tsk_thread_flag(current, TIF_SIGPENDING);
2690 return -ERESTARTNOINTR;
2691}
2692
1da177e4
LT
2693static inline int signal_pending(struct task_struct *p)
2694{
2695 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2696}
f776d12d 2697
d9588725
RM
2698static inline int __fatal_signal_pending(struct task_struct *p)
2699{
2700 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2701}
f776d12d
MW
2702
2703static inline int fatal_signal_pending(struct task_struct *p)
2704{
2705 return signal_pending(p) && __fatal_signal_pending(p);
2706}
2707
16882c1e
ON
2708static inline int signal_pending_state(long state, struct task_struct *p)
2709{
2710 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2711 return 0;
2712 if (!signal_pending(p))
2713 return 0;
2714
16882c1e
ON
2715 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2716}
2717
1da177e4
LT
2718/*
2719 * cond_resched() and cond_resched_lock(): latency reduction via
2720 * explicit rescheduling in places that are safe. The return
2721 * value indicates whether a reschedule was done in fact.
2722 * cond_resched_lock() will drop the spinlock before scheduling,
2723 * cond_resched_softirq() will enable bhs before scheduling.
2724 */
c3921ab7 2725extern int _cond_resched(void);
6f80bd98 2726
613afbf8
FW
2727#define cond_resched() ({ \
2728 __might_sleep(__FILE__, __LINE__, 0); \
2729 _cond_resched(); \
2730})
6f80bd98 2731
613afbf8
FW
2732extern int __cond_resched_lock(spinlock_t *lock);
2733
bdd4e85d 2734#ifdef CONFIG_PREEMPT_COUNT
716a4234 2735#define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
02b67cc3 2736#else
716a4234 2737#define PREEMPT_LOCK_OFFSET 0
02b67cc3 2738#endif
716a4234 2739
613afbf8 2740#define cond_resched_lock(lock) ({ \
716a4234 2741 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
613afbf8
FW
2742 __cond_resched_lock(lock); \
2743})
2744
2745extern int __cond_resched_softirq(void);
2746
75e1056f
VP
2747#define cond_resched_softirq() ({ \
2748 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2749 __cond_resched_softirq(); \
613afbf8 2750})
1da177e4 2751
f6f3c437
SH
2752static inline void cond_resched_rcu(void)
2753{
2754#if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2755 rcu_read_unlock();
2756 cond_resched();
2757 rcu_read_lock();
2758#endif
2759}
2760
1da177e4
LT
2761/*
2762 * Does a critical section need to be broken due to another
95c354fe
NP
2763 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2764 * but a general need for low latency)
1da177e4 2765 */
95c354fe 2766static inline int spin_needbreak(spinlock_t *lock)
1da177e4 2767{
95c354fe
NP
2768#ifdef CONFIG_PREEMPT
2769 return spin_is_contended(lock);
2770#else
1da177e4 2771 return 0;
95c354fe 2772#endif
1da177e4
LT
2773}
2774
ee761f62
TG
2775/*
2776 * Idle thread specific functions to determine the need_resched
69dd0f84 2777 * polling state.
ee761f62 2778 */
69dd0f84 2779#ifdef TIF_POLLING_NRFLAG
ee761f62
TG
2780static inline int tsk_is_polling(struct task_struct *p)
2781{
2782 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2783}
ea811747
PZ
2784
2785static inline void __current_set_polling(void)
3a98f871
TG
2786{
2787 set_thread_flag(TIF_POLLING_NRFLAG);
2788}
2789
ea811747
PZ
2790static inline bool __must_check current_set_polling_and_test(void)
2791{
2792 __current_set_polling();
2793
2794 /*
2795 * Polling state must be visible before we test NEED_RESCHED,
8875125e 2796 * paired by resched_curr()
ea811747 2797 */
4e857c58 2798 smp_mb__after_atomic();
ea811747
PZ
2799
2800 return unlikely(tif_need_resched());
2801}
2802
2803static inline void __current_clr_polling(void)
3a98f871
TG
2804{
2805 clear_thread_flag(TIF_POLLING_NRFLAG);
2806}
ea811747
PZ
2807
2808static inline bool __must_check current_clr_polling_and_test(void)
2809{
2810 __current_clr_polling();
2811
2812 /*
2813 * Polling state must be visible before we test NEED_RESCHED,
8875125e 2814 * paired by resched_curr()
ea811747 2815 */
4e857c58 2816 smp_mb__after_atomic();
ea811747
PZ
2817
2818 return unlikely(tif_need_resched());
2819}
2820
ee761f62
TG
2821#else
2822static inline int tsk_is_polling(struct task_struct *p) { return 0; }
ea811747
PZ
2823static inline void __current_set_polling(void) { }
2824static inline void __current_clr_polling(void) { }
2825
2826static inline bool __must_check current_set_polling_and_test(void)
2827{
2828 return unlikely(tif_need_resched());
2829}
2830static inline bool __must_check current_clr_polling_and_test(void)
2831{
2832 return unlikely(tif_need_resched());
2833}
ee761f62
TG
2834#endif
2835
8cb75e0c
PZ
2836static inline void current_clr_polling(void)
2837{
2838 __current_clr_polling();
2839
2840 /*
2841 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2842 * Once the bit is cleared, we'll get IPIs with every new
2843 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2844 * fold.
2845 */
8875125e 2846 smp_mb(); /* paired with resched_curr() */
8cb75e0c
PZ
2847
2848 preempt_fold_need_resched();
2849}
2850
75f93fed
PZ
2851static __always_inline bool need_resched(void)
2852{
2853 return unlikely(tif_need_resched());
2854}
2855
f06febc9
FM
2856/*
2857 * Thread group CPU time accounting.
2858 */
4cd4c1b4 2859void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
4da94d49 2860void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
f06febc9 2861
490dea45 2862static inline void thread_group_cputime_init(struct signal_struct *sig)
f06febc9 2863{
ee30a7b2 2864 raw_spin_lock_init(&sig->cputimer.lock);
f06febc9
FM
2865}
2866
7bb44ade
RM
2867/*
2868 * Reevaluate whether the task has signals pending delivery.
2869 * Wake the task if so.
2870 * This is required every time the blocked sigset_t changes.
2871 * callers must hold sighand->siglock.
2872 */
2873extern void recalc_sigpending_and_wake(struct task_struct *t);
1da177e4
LT
2874extern void recalc_sigpending(void);
2875
910ffdb1
ON
2876extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2877
2878static inline void signal_wake_up(struct task_struct *t, bool resume)
2879{
2880 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2881}
2882static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2883{
2884 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2885}
1da177e4
LT
2886
2887/*
2888 * Wrappers for p->thread_info->cpu access. No-op on UP.
2889 */
2890#ifdef CONFIG_SMP
2891
2892static inline unsigned int task_cpu(const struct task_struct *p)
2893{
a1261f54 2894 return task_thread_info(p)->cpu;
1da177e4
LT
2895}
2896
b32e86b4
IM
2897static inline int task_node(const struct task_struct *p)
2898{
2899 return cpu_to_node(task_cpu(p));
2900}
2901
c65cc870 2902extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
1da177e4
LT
2903
2904#else
2905
2906static inline unsigned int task_cpu(const struct task_struct *p)
2907{
2908 return 0;
2909}
2910
2911static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2912{
2913}
2914
2915#endif /* CONFIG_SMP */
2916
96f874e2
RR
2917extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2918extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
5c45bf27 2919
7c941438 2920#ifdef CONFIG_CGROUP_SCHED
07e06b01 2921extern struct task_group root_task_group;
8323f26c 2922#endif /* CONFIG_CGROUP_SCHED */
9b5b7751 2923
54e99124
DG
2924extern int task_can_switch_user(struct user_struct *up,
2925 struct task_struct *tsk);
2926
4b98d11b
AD
2927#ifdef CONFIG_TASK_XACCT
2928static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2929{
940389b8 2930 tsk->ioac.rchar += amt;
4b98d11b
AD
2931}
2932
2933static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2934{
940389b8 2935 tsk->ioac.wchar += amt;
4b98d11b
AD
2936}
2937
2938static inline void inc_syscr(struct task_struct *tsk)
2939{
940389b8 2940 tsk->ioac.syscr++;
4b98d11b
AD
2941}
2942
2943static inline void inc_syscw(struct task_struct *tsk)
2944{
940389b8 2945 tsk->ioac.syscw++;
4b98d11b
AD
2946}
2947#else
2948static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2949{
2950}
2951
2952static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2953{
2954}
2955
2956static inline void inc_syscr(struct task_struct *tsk)
2957{
2958}
2959
2960static inline void inc_syscw(struct task_struct *tsk)
2961{
2962}
2963#endif
2964
82455257
DH
2965#ifndef TASK_SIZE_OF
2966#define TASK_SIZE_OF(tsk) TASK_SIZE
2967#endif
2968
f98bafa0 2969#ifdef CONFIG_MEMCG
cf475ad2 2970extern void mm_update_next_owner(struct mm_struct *mm);
cf475ad2
BS
2971#else
2972static inline void mm_update_next_owner(struct mm_struct *mm)
2973{
2974}
f98bafa0 2975#endif /* CONFIG_MEMCG */
cf475ad2 2976
3e10e716
JS
2977static inline unsigned long task_rlimit(const struct task_struct *tsk,
2978 unsigned int limit)
2979{
2980 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2981}
2982
2983static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2984 unsigned int limit)
2985{
2986 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2987}
2988
2989static inline unsigned long rlimit(unsigned int limit)
2990{
2991 return task_rlimit(current, limit);
2992}
2993
2994static inline unsigned long rlimit_max(unsigned int limit)
2995{
2996 return task_rlimit_max(current, limit);
2997}
2998
1da177e4 2999#endif
This page took 1.942591 seconds and 5 git commands to generate.