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