iwlwifi: rs: remove unneeded check of average tpt in window
[linux-2.6/btrfs-unstable.git] / include / linux / sched.h
blobf74d4cc3a3e54f72026449a8f70ff287dc807248
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
4 #include <uapi/linux/sched.h>
6 #include <linux/sched/prio.h>
9 struct sched_param {
10 int sched_priority;
13 #include <asm/param.h> /* for HZ */
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>
21 #include <linux/plist.h>
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>
27 #include <linux/mm_types.h>
28 #include <linux/preempt_mask.h>
30 #include <asm/page.h>
31 #include <asm/ptrace.h>
32 #include <linux/cputime.h>
34 #include <linux/smp.h>
35 #include <linux/sem.h>
36 #include <linux/shm.h>
37 #include <linux/signal.h>
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>
43 #include <linux/proportions.h>
44 #include <linux/seccomp.h>
45 #include <linux/rcupdate.h>
46 #include <linux/rculist.h>
47 #include <linux/rtmutex.h>
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>
54 #include <linux/task_io_accounting.h>
55 #include <linux/latencytop.h>
56 #include <linux/cred.h>
57 #include <linux/llist.h>
58 #include <linux/uidgid.h>
59 #include <linux/gfp.h>
60 #include <linux/magic.h>
62 #include <asm/processor.h>
64 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
67 * Extended scheduling parameters data structure.
69 * This is needed because the original struct sched_param can not be
70 * altered without introducing ABI issues with legacy applications
71 * (e.g., in sched_getparam()).
73 * However, the possibility of specifying more than just a priority for
74 * the tasks may be useful for a wide variety of application fields, e.g.,
75 * multimedia, streaming, automation and control, and many others.
77 * This variant (sched_attr) is meant at describing a so-called
78 * sporadic time-constrained task. In such model a task is specified by:
79 * - the activation period or minimum instance inter-arrival time;
80 * - the maximum (or average, depending on the actual scheduling
81 * discipline) computation time of all instances, a.k.a. runtime;
82 * - the deadline (relative to the actual activation time) of each
83 * instance.
84 * Very briefly, a periodic (sporadic) task asks for the execution of
85 * some specific computation --which is typically called an instance--
86 * (at most) every period. Moreover, each instance typically lasts no more
87 * than the runtime and must be completed by time instant t equal to
88 * the instance activation time + the deadline.
90 * This is reflected by the actual fields of the sched_attr structure:
92 * @size size of the structure, for fwd/bwd compat.
94 * @sched_policy task's scheduling policy
95 * @sched_flags for customizing the scheduler behaviour
96 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
97 * @sched_priority task's static priority (SCHED_FIFO/RR)
98 * @sched_deadline representative of the task's deadline
99 * @sched_runtime representative of the task's runtime
100 * @sched_period representative of the task's period
102 * Given this task model, there are a multiplicity of scheduling algorithms
103 * and policies, that can be used to ensure all the tasks will make their
104 * timing constraints.
106 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
107 * only user of this new interface. More information about the algorithm
108 * available in the scheduling class file or in Documentation/.
110 struct sched_attr {
111 u32 size;
113 u32 sched_policy;
114 u64 sched_flags;
116 /* SCHED_NORMAL, SCHED_BATCH */
117 s32 sched_nice;
119 /* SCHED_FIFO, SCHED_RR */
120 u32 sched_priority;
122 /* SCHED_DEADLINE */
123 u64 sched_runtime;
124 u64 sched_deadline;
125 u64 sched_period;
128 struct exec_domain;
129 struct futex_pi_state;
130 struct robust_list_head;
131 struct bio_list;
132 struct fs_struct;
133 struct perf_event_context;
134 struct blk_plug;
135 struct filename;
137 #define VMACACHE_BITS 2
138 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
139 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
142 * These are the constant used to fake the fixed-point load-average
143 * counting. Some notes:
144 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
145 * a load-average precision of 10 bits integer + 11 bits fractional
146 * - if you want to count load-averages more often, you need more
147 * precision, or rounding will get you. With 2-second counting freq,
148 * the EXP_n values would be 1981, 2034 and 2043 if still using only
149 * 11 bit fractions.
151 extern unsigned long avenrun[]; /* Load averages */
152 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
154 #define FSHIFT 11 /* nr of bits of precision */
155 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
156 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
157 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
158 #define EXP_5 2014 /* 1/exp(5sec/5min) */
159 #define EXP_15 2037 /* 1/exp(5sec/15min) */
161 #define CALC_LOAD(load,exp,n) \
162 load *= exp; \
163 load += n*(FIXED_1-exp); \
164 load >>= FSHIFT;
166 extern unsigned long total_forks;
167 extern int nr_threads;
168 DECLARE_PER_CPU(unsigned long, process_counts);
169 extern int nr_processes(void);
170 extern unsigned long nr_running(void);
171 extern bool single_task_running(void);
172 extern unsigned long nr_iowait(void);
173 extern unsigned long nr_iowait_cpu(int cpu);
174 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
176 extern void calc_global_load(unsigned long ticks);
177 extern void update_cpu_load_nohz(void);
179 /* Notifier for when a task gets migrated to a new CPU */
180 struct task_migration_notifier {
181 struct task_struct *task;
182 int from_cpu;
183 int to_cpu;
185 extern void register_task_migration_notifier(struct notifier_block *n);
187 extern unsigned long get_parent_ip(unsigned long addr);
189 extern void dump_cpu_task(int cpu);
191 struct seq_file;
192 struct cfs_rq;
193 struct task_group;
194 #ifdef CONFIG_SCHED_DEBUG
195 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
196 extern void proc_sched_set_task(struct task_struct *p);
197 extern void
198 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
199 #endif
202 * Task state bitmask. NOTE! These bits are also
203 * encoded in fs/proc/array.c: get_task_state().
205 * We have two separate sets of flags: task->state
206 * is about runnability, while task->exit_state are
207 * about the task exiting. Confusing, but this way
208 * modifying one set can't modify the other one by
209 * mistake.
211 #define TASK_RUNNING 0
212 #define TASK_INTERRUPTIBLE 1
213 #define TASK_UNINTERRUPTIBLE 2
214 #define __TASK_STOPPED 4
215 #define __TASK_TRACED 8
216 /* in tsk->exit_state */
217 #define EXIT_DEAD 16
218 #define EXIT_ZOMBIE 32
219 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
220 /* in tsk->state again */
221 #define TASK_DEAD 64
222 #define TASK_WAKEKILL 128
223 #define TASK_WAKING 256
224 #define TASK_PARKED 512
225 #define TASK_STATE_MAX 1024
227 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWP"
229 extern char ___assert_task_state[1 - 2*!!(
230 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
232 /* Convenience macros for the sake of set_task_state */
233 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
234 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
235 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
237 /* Convenience macros for the sake of wake_up */
238 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
239 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
241 /* get_task_state() */
242 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
243 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
244 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
246 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
247 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
248 #define task_is_stopped_or_traced(task) \
249 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
250 #define task_contributes_to_load(task) \
251 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
252 (task->flags & PF_FROZEN) == 0)
254 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
256 #define __set_task_state(tsk, state_value) \
257 do { \
258 (tsk)->task_state_change = _THIS_IP_; \
259 (tsk)->state = (state_value); \
260 } while (0)
261 #define set_task_state(tsk, state_value) \
262 do { \
263 (tsk)->task_state_change = _THIS_IP_; \
264 set_mb((tsk)->state, (state_value)); \
265 } while (0)
268 * set_current_state() includes a barrier so that the write of current->state
269 * is correctly serialised wrt the caller's subsequent test of whether to
270 * actually sleep:
272 * set_current_state(TASK_UNINTERRUPTIBLE);
273 * if (do_i_need_to_sleep())
274 * schedule();
276 * If the caller does not need such serialisation then use __set_current_state()
278 #define __set_current_state(state_value) \
279 do { \
280 current->task_state_change = _THIS_IP_; \
281 current->state = (state_value); \
282 } while (0)
283 #define set_current_state(state_value) \
284 do { \
285 current->task_state_change = _THIS_IP_; \
286 set_mb(current->state, (state_value)); \
287 } while (0)
289 #else
291 #define __set_task_state(tsk, state_value) \
292 do { (tsk)->state = (state_value); } while (0)
293 #define set_task_state(tsk, state_value) \
294 set_mb((tsk)->state, (state_value))
297 * set_current_state() includes a barrier so that the write of current->state
298 * is correctly serialised wrt the caller's subsequent test of whether to
299 * actually sleep:
301 * set_current_state(TASK_UNINTERRUPTIBLE);
302 * if (do_i_need_to_sleep())
303 * schedule();
305 * If the caller does not need such serialisation then use __set_current_state()
307 #define __set_current_state(state_value) \
308 do { current->state = (state_value); } while (0)
309 #define set_current_state(state_value) \
310 set_mb(current->state, (state_value))
312 #endif
314 /* Task command name length */
315 #define TASK_COMM_LEN 16
317 #include <linux/spinlock.h>
320 * This serializes "schedule()" and also protects
321 * the run-queue from deletions/modifications (but
322 * _adding_ to the beginning of the run-queue has
323 * a separate lock).
325 extern rwlock_t tasklist_lock;
326 extern spinlock_t mmlist_lock;
328 struct task_struct;
330 #ifdef CONFIG_PROVE_RCU
331 extern int lockdep_tasklist_lock_is_held(void);
332 #endif /* #ifdef CONFIG_PROVE_RCU */
334 extern void sched_init(void);
335 extern void sched_init_smp(void);
336 extern asmlinkage void schedule_tail(struct task_struct *prev);
337 extern void init_idle(struct task_struct *idle, int cpu);
338 extern void init_idle_bootup_task(struct task_struct *idle);
340 extern cpumask_var_t cpu_isolated_map;
342 extern int runqueue_is_locked(int cpu);
344 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
345 extern void nohz_balance_enter_idle(int cpu);
346 extern void set_cpu_sd_state_idle(void);
347 extern int get_nohz_timer_target(int pinned);
348 #else
349 static inline void nohz_balance_enter_idle(int cpu) { }
350 static inline void set_cpu_sd_state_idle(void) { }
351 static inline int get_nohz_timer_target(int pinned)
353 return smp_processor_id();
355 #endif
358 * Only dump TASK_* tasks. (0 for all tasks)
360 extern void show_state_filter(unsigned long state_filter);
362 static inline void show_state(void)
364 show_state_filter(0);
367 extern void show_regs(struct pt_regs *);
370 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
371 * task), SP is the stack pointer of the first frame that should be shown in the back
372 * trace (or NULL if the entire call-chain of the task should be shown).
374 extern void show_stack(struct task_struct *task, unsigned long *sp);
376 extern void cpu_init (void);
377 extern void trap_init(void);
378 extern void update_process_times(int user);
379 extern void scheduler_tick(void);
381 extern void sched_show_task(struct task_struct *p);
383 #ifdef CONFIG_LOCKUP_DETECTOR
384 extern void touch_softlockup_watchdog(void);
385 extern void touch_softlockup_watchdog_sync(void);
386 extern void touch_all_softlockup_watchdogs(void);
387 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
388 void __user *buffer,
389 size_t *lenp, loff_t *ppos);
390 extern unsigned int softlockup_panic;
391 void lockup_detector_init(void);
392 #else
393 static inline void touch_softlockup_watchdog(void)
396 static inline void touch_softlockup_watchdog_sync(void)
399 static inline void touch_all_softlockup_watchdogs(void)
402 static inline void lockup_detector_init(void)
405 #endif
407 #ifdef CONFIG_DETECT_HUNG_TASK
408 void reset_hung_task_detector(void);
409 #else
410 static inline void reset_hung_task_detector(void)
413 #endif
415 /* Attach to any functions which should be ignored in wchan output. */
416 #define __sched __attribute__((__section__(".sched.text")))
418 /* Linker adds these: start and end of __sched functions */
419 extern char __sched_text_start[], __sched_text_end[];
421 /* Is this address in the __sched functions? */
422 extern int in_sched_functions(unsigned long addr);
424 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
425 extern signed long schedule_timeout(signed long timeout);
426 extern signed long schedule_timeout_interruptible(signed long timeout);
427 extern signed long schedule_timeout_killable(signed long timeout);
428 extern signed long schedule_timeout_uninterruptible(signed long timeout);
429 asmlinkage void schedule(void);
430 extern void schedule_preempt_disabled(void);
432 extern long io_schedule_timeout(long timeout);
434 static inline void io_schedule(void)
436 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
439 struct nsproxy;
440 struct user_namespace;
442 #ifdef CONFIG_MMU
443 extern void arch_pick_mmap_layout(struct mm_struct *mm);
444 extern unsigned long
445 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
446 unsigned long, unsigned long);
447 extern unsigned long
448 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
449 unsigned long len, unsigned long pgoff,
450 unsigned long flags);
451 #else
452 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
453 #endif
455 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
456 #define SUID_DUMP_USER 1 /* Dump as user of process */
457 #define SUID_DUMP_ROOT 2 /* Dump as root */
459 /* mm flags */
461 /* for SUID_DUMP_* above */
462 #define MMF_DUMPABLE_BITS 2
463 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
465 extern void set_dumpable(struct mm_struct *mm, int value);
467 * This returns the actual value of the suid_dumpable flag. For things
468 * that are using this for checking for privilege transitions, it must
469 * test against SUID_DUMP_USER rather than treating it as a boolean
470 * value.
472 static inline int __get_dumpable(unsigned long mm_flags)
474 return mm_flags & MMF_DUMPABLE_MASK;
477 static inline int get_dumpable(struct mm_struct *mm)
479 return __get_dumpable(mm->flags);
482 /* coredump filter bits */
483 #define MMF_DUMP_ANON_PRIVATE 2
484 #define MMF_DUMP_ANON_SHARED 3
485 #define MMF_DUMP_MAPPED_PRIVATE 4
486 #define MMF_DUMP_MAPPED_SHARED 5
487 #define MMF_DUMP_ELF_HEADERS 6
488 #define MMF_DUMP_HUGETLB_PRIVATE 7
489 #define MMF_DUMP_HUGETLB_SHARED 8
491 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
492 #define MMF_DUMP_FILTER_BITS 7
493 #define MMF_DUMP_FILTER_MASK \
494 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
495 #define MMF_DUMP_FILTER_DEFAULT \
496 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
497 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
499 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
500 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
501 #else
502 # define MMF_DUMP_MASK_DEFAULT_ELF 0
503 #endif
504 /* leave room for more dump flags */
505 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
506 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
507 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
509 #define MMF_HAS_UPROBES 19 /* has uprobes */
510 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
512 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
514 struct sighand_struct {
515 atomic_t count;
516 struct k_sigaction action[_NSIG];
517 spinlock_t siglock;
518 wait_queue_head_t signalfd_wqh;
521 struct pacct_struct {
522 int ac_flag;
523 long ac_exitcode;
524 unsigned long ac_mem;
525 cputime_t ac_utime, ac_stime;
526 unsigned long ac_minflt, ac_majflt;
529 struct cpu_itimer {
530 cputime_t expires;
531 cputime_t incr;
532 u32 error;
533 u32 incr_error;
537 * struct cputime - snaphsot of system and user cputime
538 * @utime: time spent in user mode
539 * @stime: time spent in system mode
541 * Gathers a generic snapshot of user and system time.
543 struct cputime {
544 cputime_t utime;
545 cputime_t stime;
549 * struct task_cputime - collected CPU time counts
550 * @utime: time spent in user mode, in &cputime_t units
551 * @stime: time spent in kernel mode, in &cputime_t units
552 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
554 * This is an extension of struct cputime that includes the total runtime
555 * spent by the task from the scheduler point of view.
557 * As a result, this structure groups together three kinds of CPU time
558 * that are tracked for threads and thread groups. Most things considering
559 * CPU time want to group these counts together and treat all three
560 * of them in parallel.
562 struct task_cputime {
563 cputime_t utime;
564 cputime_t stime;
565 unsigned long long sum_exec_runtime;
567 /* Alternate field names when used to cache expirations. */
568 #define prof_exp stime
569 #define virt_exp utime
570 #define sched_exp sum_exec_runtime
572 #define INIT_CPUTIME \
573 (struct task_cputime) { \
574 .utime = 0, \
575 .stime = 0, \
576 .sum_exec_runtime = 0, \
579 #ifdef CONFIG_PREEMPT_COUNT
580 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
581 #else
582 #define PREEMPT_DISABLED PREEMPT_ENABLED
583 #endif
586 * Disable preemption until the scheduler is running.
587 * Reset by start_kernel()->sched_init()->init_idle().
589 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
590 * before the scheduler is active -- see should_resched().
592 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
595 * struct thread_group_cputimer - thread group interval timer counts
596 * @cputime: thread group interval timers.
597 * @running: non-zero when there are timers running and
598 * @cputime receives updates.
599 * @lock: lock for fields in this struct.
601 * This structure contains the version of task_cputime, above, that is
602 * used for thread group CPU timer calculations.
604 struct thread_group_cputimer {
605 struct task_cputime cputime;
606 int running;
607 raw_spinlock_t lock;
610 #include <linux/rwsem.h>
611 struct autogroup;
614 * NOTE! "signal_struct" does not have its own
615 * locking, because a shared signal_struct always
616 * implies a shared sighand_struct, so locking
617 * sighand_struct is always a proper superset of
618 * the locking of signal_struct.
620 struct signal_struct {
621 atomic_t sigcnt;
622 atomic_t live;
623 int nr_threads;
624 struct list_head thread_head;
626 wait_queue_head_t wait_chldexit; /* for wait4() */
628 /* current thread group signal load-balancing target: */
629 struct task_struct *curr_target;
631 /* shared signal handling: */
632 struct sigpending shared_pending;
634 /* thread group exit support */
635 int group_exit_code;
636 /* overloaded:
637 * - notify group_exit_task when ->count is equal to notify_count
638 * - everyone except group_exit_task is stopped during signal delivery
639 * of fatal signals, group_exit_task processes the signal.
641 int notify_count;
642 struct task_struct *group_exit_task;
644 /* thread group stop support, overloads group_exit_code too */
645 int group_stop_count;
646 unsigned int flags; /* see SIGNAL_* flags below */
649 * PR_SET_CHILD_SUBREAPER marks a process, like a service
650 * manager, to re-parent orphan (double-forking) child processes
651 * to this process instead of 'init'. The service manager is
652 * able to receive SIGCHLD signals and is able to investigate
653 * the process until it calls wait(). All children of this
654 * process will inherit a flag if they should look for a
655 * child_subreaper process at exit.
657 unsigned int is_child_subreaper:1;
658 unsigned int has_child_subreaper:1;
660 /* POSIX.1b Interval Timers */
661 int posix_timer_id;
662 struct list_head posix_timers;
664 /* ITIMER_REAL timer for the process */
665 struct hrtimer real_timer;
666 struct pid *leader_pid;
667 ktime_t it_real_incr;
670 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
671 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
672 * values are defined to 0 and 1 respectively
674 struct cpu_itimer it[2];
677 * Thread group totals for process CPU timers.
678 * See thread_group_cputimer(), et al, for details.
680 struct thread_group_cputimer cputimer;
682 /* Earliest-expiration cache. */
683 struct task_cputime cputime_expires;
685 struct list_head cpu_timers[3];
687 struct pid *tty_old_pgrp;
689 /* boolean value for session group leader */
690 int leader;
692 struct tty_struct *tty; /* NULL if no tty */
694 #ifdef CONFIG_SCHED_AUTOGROUP
695 struct autogroup *autogroup;
696 #endif
698 * Cumulative resource counters for dead threads in the group,
699 * and for reaped dead child processes forked by this group.
700 * Live threads maintain their own counters and add to these
701 * in __exit_signal, except for the group leader.
703 seqlock_t stats_lock;
704 cputime_t utime, stime, cutime, cstime;
705 cputime_t gtime;
706 cputime_t cgtime;
707 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
708 struct cputime prev_cputime;
709 #endif
710 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
711 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
712 unsigned long inblock, oublock, cinblock, coublock;
713 unsigned long maxrss, cmaxrss;
714 struct task_io_accounting ioac;
717 * Cumulative ns of schedule CPU time fo dead threads in the
718 * group, not including a zombie group leader, (This only differs
719 * from jiffies_to_ns(utime + stime) if sched_clock uses something
720 * other than jiffies.)
722 unsigned long long sum_sched_runtime;
725 * We don't bother to synchronize most readers of this at all,
726 * because there is no reader checking a limit that actually needs
727 * to get both rlim_cur and rlim_max atomically, and either one
728 * alone is a single word that can safely be read normally.
729 * getrlimit/setrlimit use task_lock(current->group_leader) to
730 * protect this instead of the siglock, because they really
731 * have no need to disable irqs.
733 struct rlimit rlim[RLIM_NLIMITS];
735 #ifdef CONFIG_BSD_PROCESS_ACCT
736 struct pacct_struct pacct; /* per-process accounting information */
737 #endif
738 #ifdef CONFIG_TASKSTATS
739 struct taskstats *stats;
740 #endif
741 #ifdef CONFIG_AUDIT
742 unsigned audit_tty;
743 unsigned audit_tty_log_passwd;
744 struct tty_audit_buf *tty_audit_buf;
745 #endif
746 #ifdef CONFIG_CGROUPS
748 * group_rwsem prevents new tasks from entering the threadgroup and
749 * member tasks from exiting,a more specifically, setting of
750 * PF_EXITING. fork and exit paths are protected with this rwsem
751 * using threadgroup_change_begin/end(). Users which require
752 * threadgroup to remain stable should use threadgroup_[un]lock()
753 * which also takes care of exec path. Currently, cgroup is the
754 * only user.
756 struct rw_semaphore group_rwsem;
757 #endif
759 oom_flags_t oom_flags;
760 short oom_score_adj; /* OOM kill score adjustment */
761 short oom_score_adj_min; /* OOM kill score adjustment min value.
762 * Only settable by CAP_SYS_RESOURCE. */
764 struct mutex cred_guard_mutex; /* guard against foreign influences on
765 * credential calculations
766 * (notably. ptrace) */
770 * Bits in flags field of signal_struct.
772 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
773 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
774 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
775 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
777 * Pending notifications to parent.
779 #define SIGNAL_CLD_STOPPED 0x00000010
780 #define SIGNAL_CLD_CONTINUED 0x00000020
781 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
783 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
785 /* If true, all threads except ->group_exit_task have pending SIGKILL */
786 static inline int signal_group_exit(const struct signal_struct *sig)
788 return (sig->flags & SIGNAL_GROUP_EXIT) ||
789 (sig->group_exit_task != NULL);
793 * Some day this will be a full-fledged user tracking system..
795 struct user_struct {
796 atomic_t __count; /* reference count */
797 atomic_t processes; /* How many processes does this user have? */
798 atomic_t sigpending; /* How many pending signals does this user have? */
799 #ifdef CONFIG_INOTIFY_USER
800 atomic_t inotify_watches; /* How many inotify watches does this user have? */
801 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
802 #endif
803 #ifdef CONFIG_FANOTIFY
804 atomic_t fanotify_listeners;
805 #endif
806 #ifdef CONFIG_EPOLL
807 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
808 #endif
809 #ifdef CONFIG_POSIX_MQUEUE
810 /* protected by mq_lock */
811 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
812 #endif
813 unsigned long locked_shm; /* How many pages of mlocked shm ? */
815 #ifdef CONFIG_KEYS
816 struct key *uid_keyring; /* UID specific keyring */
817 struct key *session_keyring; /* UID's default session keyring */
818 #endif
820 /* Hash table maintenance information */
821 struct hlist_node uidhash_node;
822 kuid_t uid;
824 #ifdef CONFIG_PERF_EVENTS
825 atomic_long_t locked_vm;
826 #endif
829 extern int uids_sysfs_init(void);
831 extern struct user_struct *find_user(kuid_t);
833 extern struct user_struct root_user;
834 #define INIT_USER (&root_user)
837 struct backing_dev_info;
838 struct reclaim_state;
840 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
841 struct sched_info {
842 /* cumulative counters */
843 unsigned long pcount; /* # of times run on this cpu */
844 unsigned long long run_delay; /* time spent waiting on a runqueue */
846 /* timestamps */
847 unsigned long long last_arrival,/* when we last ran on a cpu */
848 last_queued; /* when we were last queued to run */
850 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
852 #ifdef CONFIG_TASK_DELAY_ACCT
853 struct task_delay_info {
854 spinlock_t lock;
855 unsigned int flags; /* Private per-task flags */
857 /* For each stat XXX, add following, aligned appropriately
859 * struct timespec XXX_start, XXX_end;
860 * u64 XXX_delay;
861 * u32 XXX_count;
863 * Atomicity of updates to XXX_delay, XXX_count protected by
864 * single lock above (split into XXX_lock if contention is an issue).
868 * XXX_count is incremented on every XXX operation, the delay
869 * associated with the operation is added to XXX_delay.
870 * XXX_delay contains the accumulated delay time in nanoseconds.
872 u64 blkio_start; /* Shared by blkio, swapin */
873 u64 blkio_delay; /* wait for sync block io completion */
874 u64 swapin_delay; /* wait for swapin block io completion */
875 u32 blkio_count; /* total count of the number of sync block */
876 /* io operations performed */
877 u32 swapin_count; /* total count of the number of swapin block */
878 /* io operations performed */
880 u64 freepages_start;
881 u64 freepages_delay; /* wait for memory reclaim */
882 u32 freepages_count; /* total count of memory reclaim */
884 #endif /* CONFIG_TASK_DELAY_ACCT */
886 static inline int sched_info_on(void)
888 #ifdef CONFIG_SCHEDSTATS
889 return 1;
890 #elif defined(CONFIG_TASK_DELAY_ACCT)
891 extern int delayacct_on;
892 return delayacct_on;
893 #else
894 return 0;
895 #endif
898 enum cpu_idle_type {
899 CPU_IDLE,
900 CPU_NOT_IDLE,
901 CPU_NEWLY_IDLE,
902 CPU_MAX_IDLE_TYPES
906 * Increase resolution of cpu_capacity calculations
908 #define SCHED_CAPACITY_SHIFT 10
909 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
912 * sched-domains (multiprocessor balancing) declarations:
914 #ifdef CONFIG_SMP
915 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
916 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
917 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
918 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
919 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
920 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
921 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
922 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
923 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
924 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
925 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
926 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
927 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
928 #define SD_NUMA 0x4000 /* cross-node balancing */
930 #ifdef CONFIG_SCHED_SMT
931 static inline int cpu_smt_flags(void)
933 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
935 #endif
937 #ifdef CONFIG_SCHED_MC
938 static inline int cpu_core_flags(void)
940 return SD_SHARE_PKG_RESOURCES;
942 #endif
944 #ifdef CONFIG_NUMA
945 static inline int cpu_numa_flags(void)
947 return SD_NUMA;
949 #endif
951 struct sched_domain_attr {
952 int relax_domain_level;
955 #define SD_ATTR_INIT (struct sched_domain_attr) { \
956 .relax_domain_level = -1, \
959 extern int sched_domain_level_max;
961 struct sched_group;
963 struct sched_domain {
964 /* These fields must be setup */
965 struct sched_domain *parent; /* top domain must be null terminated */
966 struct sched_domain *child; /* bottom domain must be null terminated */
967 struct sched_group *groups; /* the balancing groups of the domain */
968 unsigned long min_interval; /* Minimum balance interval ms */
969 unsigned long max_interval; /* Maximum balance interval ms */
970 unsigned int busy_factor; /* less balancing by factor if busy */
971 unsigned int imbalance_pct; /* No balance until over watermark */
972 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
973 unsigned int busy_idx;
974 unsigned int idle_idx;
975 unsigned int newidle_idx;
976 unsigned int wake_idx;
977 unsigned int forkexec_idx;
978 unsigned int smt_gain;
980 int nohz_idle; /* NOHZ IDLE status */
981 int flags; /* See SD_* */
982 int level;
984 /* Runtime fields. */
985 unsigned long last_balance; /* init to jiffies. units in jiffies */
986 unsigned int balance_interval; /* initialise to 1. units in ms. */
987 unsigned int nr_balance_failed; /* initialise to 0 */
989 /* idle_balance() stats */
990 u64 max_newidle_lb_cost;
991 unsigned long next_decay_max_lb_cost;
993 #ifdef CONFIG_SCHEDSTATS
994 /* load_balance() stats */
995 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
996 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
997 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
998 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
999 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1000 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1001 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1002 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1004 /* Active load balancing */
1005 unsigned int alb_count;
1006 unsigned int alb_failed;
1007 unsigned int alb_pushed;
1009 /* SD_BALANCE_EXEC stats */
1010 unsigned int sbe_count;
1011 unsigned int sbe_balanced;
1012 unsigned int sbe_pushed;
1014 /* SD_BALANCE_FORK stats */
1015 unsigned int sbf_count;
1016 unsigned int sbf_balanced;
1017 unsigned int sbf_pushed;
1019 /* try_to_wake_up() stats */
1020 unsigned int ttwu_wake_remote;
1021 unsigned int ttwu_move_affine;
1022 unsigned int ttwu_move_balance;
1023 #endif
1024 #ifdef CONFIG_SCHED_DEBUG
1025 char *name;
1026 #endif
1027 union {
1028 void *private; /* used during construction */
1029 struct rcu_head rcu; /* used during destruction */
1032 unsigned int span_weight;
1034 * Span of all CPUs in this domain.
1036 * NOTE: this field is variable length. (Allocated dynamically
1037 * by attaching extra space to the end of the structure,
1038 * depending on how many CPUs the kernel has booted up with)
1040 unsigned long span[0];
1043 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1045 return to_cpumask(sd->span);
1048 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1049 struct sched_domain_attr *dattr_new);
1051 /* Allocate an array of sched domains, for partition_sched_domains(). */
1052 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1053 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1055 bool cpus_share_cache(int this_cpu, int that_cpu);
1057 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1058 typedef int (*sched_domain_flags_f)(void);
1060 #define SDTL_OVERLAP 0x01
1062 struct sd_data {
1063 struct sched_domain **__percpu sd;
1064 struct sched_group **__percpu sg;
1065 struct sched_group_capacity **__percpu sgc;
1068 struct sched_domain_topology_level {
1069 sched_domain_mask_f mask;
1070 sched_domain_flags_f sd_flags;
1071 int flags;
1072 int numa_level;
1073 struct sd_data data;
1074 #ifdef CONFIG_SCHED_DEBUG
1075 char *name;
1076 #endif
1079 extern struct sched_domain_topology_level *sched_domain_topology;
1081 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1082 extern void wake_up_if_idle(int cpu);
1084 #ifdef CONFIG_SCHED_DEBUG
1085 # define SD_INIT_NAME(type) .name = #type
1086 #else
1087 # define SD_INIT_NAME(type)
1088 #endif
1090 #else /* CONFIG_SMP */
1092 struct sched_domain_attr;
1094 static inline void
1095 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1096 struct sched_domain_attr *dattr_new)
1100 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1102 return true;
1105 #endif /* !CONFIG_SMP */
1108 struct io_context; /* See blkdev.h */
1111 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1112 extern void prefetch_stack(struct task_struct *t);
1113 #else
1114 static inline void prefetch_stack(struct task_struct *t) { }
1115 #endif
1117 struct audit_context; /* See audit.c */
1118 struct mempolicy;
1119 struct pipe_inode_info;
1120 struct uts_namespace;
1122 struct load_weight {
1123 unsigned long weight;
1124 u32 inv_weight;
1127 struct sched_avg {
1128 u64 last_runnable_update;
1129 s64 decay_count;
1131 * utilization_avg_contrib describes the amount of time that a
1132 * sched_entity is running on a CPU. It is based on running_avg_sum
1133 * and is scaled in the range [0..SCHED_LOAD_SCALE].
1134 * load_avg_contrib described the amount of time that a sched_entity
1135 * is runnable on a rq. It is based on both runnable_avg_sum and the
1136 * weight of the task.
1138 unsigned long load_avg_contrib, utilization_avg_contrib;
1140 * These sums represent an infinite geometric series and so are bound
1141 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1142 * choices of y < 1-2^(-32)*1024.
1143 * running_avg_sum reflects the time that the sched_entity is
1144 * effectively running on the CPU.
1145 * runnable_avg_sum represents the amount of time a sched_entity is on
1146 * a runqueue which includes the running time that is monitored by
1147 * running_avg_sum.
1149 u32 runnable_avg_sum, avg_period, running_avg_sum;
1152 #ifdef CONFIG_SCHEDSTATS
1153 struct sched_statistics {
1154 u64 wait_start;
1155 u64 wait_max;
1156 u64 wait_count;
1157 u64 wait_sum;
1158 u64 iowait_count;
1159 u64 iowait_sum;
1161 u64 sleep_start;
1162 u64 sleep_max;
1163 s64 sum_sleep_runtime;
1165 u64 block_start;
1166 u64 block_max;
1167 u64 exec_max;
1168 u64 slice_max;
1170 u64 nr_migrations_cold;
1171 u64 nr_failed_migrations_affine;
1172 u64 nr_failed_migrations_running;
1173 u64 nr_failed_migrations_hot;
1174 u64 nr_forced_migrations;
1176 u64 nr_wakeups;
1177 u64 nr_wakeups_sync;
1178 u64 nr_wakeups_migrate;
1179 u64 nr_wakeups_local;
1180 u64 nr_wakeups_remote;
1181 u64 nr_wakeups_affine;
1182 u64 nr_wakeups_affine_attempts;
1183 u64 nr_wakeups_passive;
1184 u64 nr_wakeups_idle;
1186 #endif
1188 struct sched_entity {
1189 struct load_weight load; /* for load-balancing */
1190 struct rb_node run_node;
1191 struct list_head group_node;
1192 unsigned int on_rq;
1194 u64 exec_start;
1195 u64 sum_exec_runtime;
1196 u64 vruntime;
1197 u64 prev_sum_exec_runtime;
1199 u64 nr_migrations;
1201 #ifdef CONFIG_SCHEDSTATS
1202 struct sched_statistics statistics;
1203 #endif
1205 #ifdef CONFIG_FAIR_GROUP_SCHED
1206 int depth;
1207 struct sched_entity *parent;
1208 /* rq on which this entity is (to be) queued: */
1209 struct cfs_rq *cfs_rq;
1210 /* rq "owned" by this entity/group: */
1211 struct cfs_rq *my_q;
1212 #endif
1214 #ifdef CONFIG_SMP
1215 /* Per-entity load-tracking */
1216 struct sched_avg avg;
1217 #endif
1220 struct sched_rt_entity {
1221 struct list_head run_list;
1222 unsigned long timeout;
1223 unsigned long watchdog_stamp;
1224 unsigned int time_slice;
1226 struct sched_rt_entity *back;
1227 #ifdef CONFIG_RT_GROUP_SCHED
1228 struct sched_rt_entity *parent;
1229 /* rq on which this entity is (to be) queued: */
1230 struct rt_rq *rt_rq;
1231 /* rq "owned" by this entity/group: */
1232 struct rt_rq *my_q;
1233 #endif
1236 struct sched_dl_entity {
1237 struct rb_node rb_node;
1240 * Original scheduling parameters. Copied here from sched_attr
1241 * during sched_setattr(), they will remain the same until
1242 * the next sched_setattr().
1244 u64 dl_runtime; /* maximum runtime for each instance */
1245 u64 dl_deadline; /* relative deadline of each instance */
1246 u64 dl_period; /* separation of two instances (period) */
1247 u64 dl_bw; /* dl_runtime / dl_deadline */
1250 * Actual scheduling parameters. Initialized with the values above,
1251 * they are continously updated during task execution. Note that
1252 * the remaining runtime could be < 0 in case we are in overrun.
1254 s64 runtime; /* remaining runtime for this instance */
1255 u64 deadline; /* absolute deadline for this instance */
1256 unsigned int flags; /* specifying the scheduler behaviour */
1259 * Some bool flags:
1261 * @dl_throttled tells if we exhausted the runtime. If so, the
1262 * task has to wait for a replenishment to be performed at the
1263 * next firing of dl_timer.
1265 * @dl_new tells if a new instance arrived. If so we must
1266 * start executing it with full runtime and reset its absolute
1267 * deadline;
1269 * @dl_boosted tells if we are boosted due to DI. If so we are
1270 * outside bandwidth enforcement mechanism (but only until we
1271 * exit the critical section);
1273 * @dl_yielded tells if task gave up the cpu before consuming
1274 * all its available runtime during the last job.
1276 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1279 * Bandwidth enforcement timer. Each -deadline task has its
1280 * own bandwidth to be enforced, thus we need one timer per task.
1282 struct hrtimer dl_timer;
1285 union rcu_special {
1286 struct {
1287 bool blocked;
1288 bool need_qs;
1289 } b;
1290 short s;
1292 struct rcu_node;
1294 enum perf_event_task_context {
1295 perf_invalid_context = -1,
1296 perf_hw_context = 0,
1297 perf_sw_context,
1298 perf_nr_task_contexts,
1301 struct task_struct {
1302 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1303 void *stack;
1304 atomic_t usage;
1305 unsigned int flags; /* per process flags, defined below */
1306 unsigned int ptrace;
1308 #ifdef CONFIG_SMP
1309 struct llist_node wake_entry;
1310 int on_cpu;
1311 struct task_struct *last_wakee;
1312 unsigned long wakee_flips;
1313 unsigned long wakee_flip_decay_ts;
1315 int wake_cpu;
1316 #endif
1317 int on_rq;
1319 int prio, static_prio, normal_prio;
1320 unsigned int rt_priority;
1321 const struct sched_class *sched_class;
1322 struct sched_entity se;
1323 struct sched_rt_entity rt;
1324 #ifdef CONFIG_CGROUP_SCHED
1325 struct task_group *sched_task_group;
1326 #endif
1327 struct sched_dl_entity dl;
1329 #ifdef CONFIG_PREEMPT_NOTIFIERS
1330 /* list of struct preempt_notifier: */
1331 struct hlist_head preempt_notifiers;
1332 #endif
1334 #ifdef CONFIG_BLK_DEV_IO_TRACE
1335 unsigned int btrace_seq;
1336 #endif
1338 unsigned int policy;
1339 int nr_cpus_allowed;
1340 cpumask_t cpus_allowed;
1342 #ifdef CONFIG_PREEMPT_RCU
1343 int rcu_read_lock_nesting;
1344 union rcu_special rcu_read_unlock_special;
1345 struct list_head rcu_node_entry;
1346 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1347 #ifdef CONFIG_PREEMPT_RCU
1348 struct rcu_node *rcu_blocked_node;
1349 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1350 #ifdef CONFIG_TASKS_RCU
1351 unsigned long rcu_tasks_nvcsw;
1352 bool rcu_tasks_holdout;
1353 struct list_head rcu_tasks_holdout_list;
1354 int rcu_tasks_idle_cpu;
1355 #endif /* #ifdef CONFIG_TASKS_RCU */
1357 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1358 struct sched_info sched_info;
1359 #endif
1361 struct list_head tasks;
1362 #ifdef CONFIG_SMP
1363 struct plist_node pushable_tasks;
1364 struct rb_node pushable_dl_tasks;
1365 #endif
1367 struct mm_struct *mm, *active_mm;
1368 #ifdef CONFIG_COMPAT_BRK
1369 unsigned brk_randomized:1;
1370 #endif
1371 /* per-thread vma caching */
1372 u32 vmacache_seqnum;
1373 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1374 #if defined(SPLIT_RSS_COUNTING)
1375 struct task_rss_stat rss_stat;
1376 #endif
1377 /* task state */
1378 int exit_state;
1379 int exit_code, exit_signal;
1380 int pdeath_signal; /* The signal sent when the parent dies */
1381 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1383 /* Used for emulating ABI behavior of previous Linux versions */
1384 unsigned int personality;
1386 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1387 * execve */
1388 unsigned in_iowait:1;
1390 /* Revert to default priority/policy when forking */
1391 unsigned sched_reset_on_fork:1;
1392 unsigned sched_contributes_to_load:1;
1394 #ifdef CONFIG_MEMCG_KMEM
1395 unsigned memcg_kmem_skip_account:1;
1396 #endif
1398 unsigned long atomic_flags; /* Flags needing atomic access. */
1400 struct restart_block restart_block;
1402 pid_t pid;
1403 pid_t tgid;
1405 #ifdef CONFIG_CC_STACKPROTECTOR
1406 /* Canary value for the -fstack-protector gcc feature */
1407 unsigned long stack_canary;
1408 #endif
1410 * pointers to (original) parent process, youngest child, younger sibling,
1411 * older sibling, respectively. (p->father can be replaced with
1412 * p->real_parent->pid)
1414 struct task_struct __rcu *real_parent; /* real parent process */
1415 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1417 * children/sibling forms the list of my natural children
1419 struct list_head children; /* list of my children */
1420 struct list_head sibling; /* linkage in my parent's children list */
1421 struct task_struct *group_leader; /* threadgroup leader */
1424 * ptraced is the list of tasks this task is using ptrace on.
1425 * This includes both natural children and PTRACE_ATTACH targets.
1426 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1428 struct list_head ptraced;
1429 struct list_head ptrace_entry;
1431 /* PID/PID hash table linkage. */
1432 struct pid_link pids[PIDTYPE_MAX];
1433 struct list_head thread_group;
1434 struct list_head thread_node;
1436 struct completion *vfork_done; /* for vfork() */
1437 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1438 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1440 cputime_t utime, stime, utimescaled, stimescaled;
1441 cputime_t gtime;
1442 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1443 struct cputime prev_cputime;
1444 #endif
1445 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1446 seqlock_t vtime_seqlock;
1447 unsigned long long vtime_snap;
1448 enum {
1449 VTIME_SLEEPING = 0,
1450 VTIME_USER,
1451 VTIME_SYS,
1452 } vtime_snap_whence;
1453 #endif
1454 unsigned long nvcsw, nivcsw; /* context switch counts */
1455 u64 start_time; /* monotonic time in nsec */
1456 u64 real_start_time; /* boot based time in nsec */
1457 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1458 unsigned long min_flt, maj_flt;
1460 struct task_cputime cputime_expires;
1461 struct list_head cpu_timers[3];
1463 /* process credentials */
1464 const struct cred __rcu *real_cred; /* objective and real subjective task
1465 * credentials (COW) */
1466 const struct cred __rcu *cred; /* effective (overridable) subjective task
1467 * credentials (COW) */
1468 char comm[TASK_COMM_LEN]; /* executable name excluding path
1469 - access with [gs]et_task_comm (which lock
1470 it with task_lock())
1471 - initialized normally by setup_new_exec */
1472 /* file system info */
1473 int link_count, total_link_count;
1474 #ifdef CONFIG_SYSVIPC
1475 /* ipc stuff */
1476 struct sysv_sem sysvsem;
1477 struct sysv_shm sysvshm;
1478 #endif
1479 #ifdef CONFIG_DETECT_HUNG_TASK
1480 /* hung task detection */
1481 unsigned long last_switch_count;
1482 #endif
1483 /* CPU-specific state of this task */
1484 struct thread_struct thread;
1485 /* filesystem information */
1486 struct fs_struct *fs;
1487 /* open file information */
1488 struct files_struct *files;
1489 /* namespaces */
1490 struct nsproxy *nsproxy;
1491 /* signal handlers */
1492 struct signal_struct *signal;
1493 struct sighand_struct *sighand;
1495 sigset_t blocked, real_blocked;
1496 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1497 struct sigpending pending;
1499 unsigned long sas_ss_sp;
1500 size_t sas_ss_size;
1501 int (*notifier)(void *priv);
1502 void *notifier_data;
1503 sigset_t *notifier_mask;
1504 struct callback_head *task_works;
1506 struct audit_context *audit_context;
1507 #ifdef CONFIG_AUDITSYSCALL
1508 kuid_t loginuid;
1509 unsigned int sessionid;
1510 #endif
1511 struct seccomp seccomp;
1513 /* Thread group tracking */
1514 u32 parent_exec_id;
1515 u32 self_exec_id;
1516 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1517 * mempolicy */
1518 spinlock_t alloc_lock;
1520 /* Protection of the PI data structures: */
1521 raw_spinlock_t pi_lock;
1523 #ifdef CONFIG_RT_MUTEXES
1524 /* PI waiters blocked on a rt_mutex held by this task */
1525 struct rb_root pi_waiters;
1526 struct rb_node *pi_waiters_leftmost;
1527 /* Deadlock detection and priority inheritance handling */
1528 struct rt_mutex_waiter *pi_blocked_on;
1529 #endif
1531 #ifdef CONFIG_DEBUG_MUTEXES
1532 /* mutex deadlock detection */
1533 struct mutex_waiter *blocked_on;
1534 #endif
1535 #ifdef CONFIG_TRACE_IRQFLAGS
1536 unsigned int irq_events;
1537 unsigned long hardirq_enable_ip;
1538 unsigned long hardirq_disable_ip;
1539 unsigned int hardirq_enable_event;
1540 unsigned int hardirq_disable_event;
1541 int hardirqs_enabled;
1542 int hardirq_context;
1543 unsigned long softirq_disable_ip;
1544 unsigned long softirq_enable_ip;
1545 unsigned int softirq_disable_event;
1546 unsigned int softirq_enable_event;
1547 int softirqs_enabled;
1548 int softirq_context;
1549 #endif
1550 #ifdef CONFIG_LOCKDEP
1551 # define MAX_LOCK_DEPTH 48UL
1552 u64 curr_chain_key;
1553 int lockdep_depth;
1554 unsigned int lockdep_recursion;
1555 struct held_lock held_locks[MAX_LOCK_DEPTH];
1556 gfp_t lockdep_reclaim_gfp;
1557 #endif
1559 /* journalling filesystem info */
1560 void *journal_info;
1562 /* stacked block device info */
1563 struct bio_list *bio_list;
1565 #ifdef CONFIG_BLOCK
1566 /* stack plugging */
1567 struct blk_plug *plug;
1568 #endif
1570 /* VM state */
1571 struct reclaim_state *reclaim_state;
1573 struct backing_dev_info *backing_dev_info;
1575 struct io_context *io_context;
1577 unsigned long ptrace_message;
1578 siginfo_t *last_siginfo; /* For ptrace use. */
1579 struct task_io_accounting ioac;
1580 #if defined(CONFIG_TASK_XACCT)
1581 u64 acct_rss_mem1; /* accumulated rss usage */
1582 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1583 cputime_t acct_timexpd; /* stime + utime since last update */
1584 #endif
1585 #ifdef CONFIG_CPUSETS
1586 nodemask_t mems_allowed; /* Protected by alloc_lock */
1587 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1588 int cpuset_mem_spread_rotor;
1589 int cpuset_slab_spread_rotor;
1590 #endif
1591 #ifdef CONFIG_CGROUPS
1592 /* Control Group info protected by css_set_lock */
1593 struct css_set __rcu *cgroups;
1594 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1595 struct list_head cg_list;
1596 #endif
1597 #ifdef CONFIG_FUTEX
1598 struct robust_list_head __user *robust_list;
1599 #ifdef CONFIG_COMPAT
1600 struct compat_robust_list_head __user *compat_robust_list;
1601 #endif
1602 struct list_head pi_state_list;
1603 struct futex_pi_state *pi_state_cache;
1604 #endif
1605 #ifdef CONFIG_PERF_EVENTS
1606 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1607 struct mutex perf_event_mutex;
1608 struct list_head perf_event_list;
1609 #endif
1610 #ifdef CONFIG_DEBUG_PREEMPT
1611 unsigned long preempt_disable_ip;
1612 #endif
1613 #ifdef CONFIG_NUMA
1614 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1615 short il_next;
1616 short pref_node_fork;
1617 #endif
1618 #ifdef CONFIG_NUMA_BALANCING
1619 int numa_scan_seq;
1620 unsigned int numa_scan_period;
1621 unsigned int numa_scan_period_max;
1622 int numa_preferred_nid;
1623 unsigned long numa_migrate_retry;
1624 u64 node_stamp; /* migration stamp */
1625 u64 last_task_numa_placement;
1626 u64 last_sum_exec_runtime;
1627 struct callback_head numa_work;
1629 struct list_head numa_entry;
1630 struct numa_group *numa_group;
1633 * numa_faults is an array split into four regions:
1634 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1635 * in this precise order.
1637 * faults_memory: Exponential decaying average of faults on a per-node
1638 * basis. Scheduling placement decisions are made based on these
1639 * counts. The values remain static for the duration of a PTE scan.
1640 * faults_cpu: Track the nodes the process was running on when a NUMA
1641 * hinting fault was incurred.
1642 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1643 * during the current scan window. When the scan completes, the counts
1644 * in faults_memory and faults_cpu decay and these values are copied.
1646 unsigned long *numa_faults;
1647 unsigned long total_numa_faults;
1650 * numa_faults_locality tracks if faults recorded during the last
1651 * scan window were remote/local or failed to migrate. The task scan
1652 * period is adapted based on the locality of the faults with different
1653 * weights depending on whether they were shared or private faults
1655 unsigned long numa_faults_locality[3];
1657 unsigned long numa_pages_migrated;
1658 #endif /* CONFIG_NUMA_BALANCING */
1660 struct rcu_head rcu;
1663 * cache last used pipe for splice
1665 struct pipe_inode_info *splice_pipe;
1667 struct page_frag task_frag;
1669 #ifdef CONFIG_TASK_DELAY_ACCT
1670 struct task_delay_info *delays;
1671 #endif
1672 #ifdef CONFIG_FAULT_INJECTION
1673 int make_it_fail;
1674 #endif
1676 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1677 * balance_dirty_pages() for some dirty throttling pause
1679 int nr_dirtied;
1680 int nr_dirtied_pause;
1681 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1683 #ifdef CONFIG_LATENCYTOP
1684 int latency_record_count;
1685 struct latency_record latency_record[LT_SAVECOUNT];
1686 #endif
1688 * time slack values; these are used to round up poll() and
1689 * select() etc timeout values. These are in nanoseconds.
1691 unsigned long timer_slack_ns;
1692 unsigned long default_timer_slack_ns;
1694 #ifdef CONFIG_KASAN
1695 unsigned int kasan_depth;
1696 #endif
1697 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1698 /* Index of current stored address in ret_stack */
1699 int curr_ret_stack;
1700 /* Stack of return addresses for return function tracing */
1701 struct ftrace_ret_stack *ret_stack;
1702 /* time stamp for last schedule */
1703 unsigned long long ftrace_timestamp;
1705 * Number of functions that haven't been traced
1706 * because of depth overrun.
1708 atomic_t trace_overrun;
1709 /* Pause for the tracing */
1710 atomic_t tracing_graph_pause;
1711 #endif
1712 #ifdef CONFIG_TRACING
1713 /* state flags for use by tracers */
1714 unsigned long trace;
1715 /* bitmask and counter of trace recursion */
1716 unsigned long trace_recursion;
1717 #endif /* CONFIG_TRACING */
1718 #ifdef CONFIG_MEMCG
1719 struct memcg_oom_info {
1720 struct mem_cgroup *memcg;
1721 gfp_t gfp_mask;
1722 int order;
1723 unsigned int may_oom:1;
1724 } memcg_oom;
1725 #endif
1726 #ifdef CONFIG_UPROBES
1727 struct uprobe_task *utask;
1728 #endif
1729 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1730 unsigned int sequential_io;
1731 unsigned int sequential_io_avg;
1732 #endif
1733 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1734 unsigned long task_state_change;
1735 #endif
1738 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1739 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1741 #define TNF_MIGRATED 0x01
1742 #define TNF_NO_GROUP 0x02
1743 #define TNF_SHARED 0x04
1744 #define TNF_FAULT_LOCAL 0x08
1745 #define TNF_MIGRATE_FAIL 0x10
1747 #ifdef CONFIG_NUMA_BALANCING
1748 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1749 extern pid_t task_numa_group_id(struct task_struct *p);
1750 extern void set_numabalancing_state(bool enabled);
1751 extern void task_numa_free(struct task_struct *p);
1752 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1753 int src_nid, int dst_cpu);
1754 #else
1755 static inline void task_numa_fault(int last_node, int node, int pages,
1756 int flags)
1759 static inline pid_t task_numa_group_id(struct task_struct *p)
1761 return 0;
1763 static inline void set_numabalancing_state(bool enabled)
1766 static inline void task_numa_free(struct task_struct *p)
1769 static inline bool should_numa_migrate_memory(struct task_struct *p,
1770 struct page *page, int src_nid, int dst_cpu)
1772 return true;
1774 #endif
1776 static inline struct pid *task_pid(struct task_struct *task)
1778 return task->pids[PIDTYPE_PID].pid;
1781 static inline struct pid *task_tgid(struct task_struct *task)
1783 return task->group_leader->pids[PIDTYPE_PID].pid;
1787 * Without tasklist or rcu lock it is not safe to dereference
1788 * the result of task_pgrp/task_session even if task == current,
1789 * we can race with another thread doing sys_setsid/sys_setpgid.
1791 static inline struct pid *task_pgrp(struct task_struct *task)
1793 return task->group_leader->pids[PIDTYPE_PGID].pid;
1796 static inline struct pid *task_session(struct task_struct *task)
1798 return task->group_leader->pids[PIDTYPE_SID].pid;
1801 struct pid_namespace;
1804 * the helpers to get the task's different pids as they are seen
1805 * from various namespaces
1807 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1808 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1809 * current.
1810 * task_xid_nr_ns() : id seen from the ns specified;
1812 * set_task_vxid() : assigns a virtual id to a task;
1814 * see also pid_nr() etc in include/linux/pid.h
1816 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1817 struct pid_namespace *ns);
1819 static inline pid_t task_pid_nr(struct task_struct *tsk)
1821 return tsk->pid;
1824 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1825 struct pid_namespace *ns)
1827 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1830 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1832 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1836 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1838 return tsk->tgid;
1841 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1843 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1845 return pid_vnr(task_tgid(tsk));
1849 static inline int pid_alive(const struct task_struct *p);
1850 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1852 pid_t pid = 0;
1854 rcu_read_lock();
1855 if (pid_alive(tsk))
1856 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1857 rcu_read_unlock();
1859 return pid;
1862 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1864 return task_ppid_nr_ns(tsk, &init_pid_ns);
1867 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1868 struct pid_namespace *ns)
1870 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1873 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1875 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1879 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1880 struct pid_namespace *ns)
1882 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1885 static inline pid_t task_session_vnr(struct task_struct *tsk)
1887 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1890 /* obsolete, do not use */
1891 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1893 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1897 * pid_alive - check that a task structure is not stale
1898 * @p: Task structure to be checked.
1900 * Test if a process is not yet dead (at most zombie state)
1901 * If pid_alive fails, then pointers within the task structure
1902 * can be stale and must not be dereferenced.
1904 * Return: 1 if the process is alive. 0 otherwise.
1906 static inline int pid_alive(const struct task_struct *p)
1908 return p->pids[PIDTYPE_PID].pid != NULL;
1912 * is_global_init - check if a task structure is init
1913 * @tsk: Task structure to be checked.
1915 * Check if a task structure is the first user space task the kernel created.
1917 * Return: 1 if the task structure is init. 0 otherwise.
1919 static inline int is_global_init(struct task_struct *tsk)
1921 return tsk->pid == 1;
1924 extern struct pid *cad_pid;
1926 extern void free_task(struct task_struct *tsk);
1927 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1929 extern void __put_task_struct(struct task_struct *t);
1931 static inline void put_task_struct(struct task_struct *t)
1933 if (atomic_dec_and_test(&t->usage))
1934 __put_task_struct(t);
1937 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1938 extern void task_cputime(struct task_struct *t,
1939 cputime_t *utime, cputime_t *stime);
1940 extern void task_cputime_scaled(struct task_struct *t,
1941 cputime_t *utimescaled, cputime_t *stimescaled);
1942 extern cputime_t task_gtime(struct task_struct *t);
1943 #else
1944 static inline void task_cputime(struct task_struct *t,
1945 cputime_t *utime, cputime_t *stime)
1947 if (utime)
1948 *utime = t->utime;
1949 if (stime)
1950 *stime = t->stime;
1953 static inline void task_cputime_scaled(struct task_struct *t,
1954 cputime_t *utimescaled,
1955 cputime_t *stimescaled)
1957 if (utimescaled)
1958 *utimescaled = t->utimescaled;
1959 if (stimescaled)
1960 *stimescaled = t->stimescaled;
1963 static inline cputime_t task_gtime(struct task_struct *t)
1965 return t->gtime;
1967 #endif
1968 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1969 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1972 * Per process flags
1974 #define PF_EXITING 0x00000004 /* getting shut down */
1975 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1976 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1977 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1978 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1979 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1980 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1981 #define PF_DUMPCORE 0x00000200 /* dumped core */
1982 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1983 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1984 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1985 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1986 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1987 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1988 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1989 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1990 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1991 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1992 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1993 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1994 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1995 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1996 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1997 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1998 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1999 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2000 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2003 * Only the _current_ task can read/write to tsk->flags, but other
2004 * tasks can access tsk->flags in readonly mode for example
2005 * with tsk_used_math (like during threaded core dumping).
2006 * There is however an exception to this rule during ptrace
2007 * or during fork: the ptracer task is allowed to write to the
2008 * child->flags of its traced child (same goes for fork, the parent
2009 * can write to the child->flags), because we're guaranteed the
2010 * child is not running and in turn not changing child->flags
2011 * at the same time the parent does it.
2013 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2014 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2015 #define clear_used_math() clear_stopped_child_used_math(current)
2016 #define set_used_math() set_stopped_child_used_math(current)
2017 #define conditional_stopped_child_used_math(condition, child) \
2018 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2019 #define conditional_used_math(condition) \
2020 conditional_stopped_child_used_math(condition, current)
2021 #define copy_to_stopped_child_used_math(child) \
2022 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2023 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2024 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2025 #define used_math() tsk_used_math(current)
2027 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2028 * __GFP_FS is also cleared as it implies __GFP_IO.
2030 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2032 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2033 flags &= ~(__GFP_IO | __GFP_FS);
2034 return flags;
2037 static inline unsigned int memalloc_noio_save(void)
2039 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2040 current->flags |= PF_MEMALLOC_NOIO;
2041 return flags;
2044 static inline void memalloc_noio_restore(unsigned int flags)
2046 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2049 /* Per-process atomic flags. */
2050 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2051 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2052 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2055 #define TASK_PFA_TEST(name, func) \
2056 static inline bool task_##func(struct task_struct *p) \
2057 { return test_bit(PFA_##name, &p->atomic_flags); }
2058 #define TASK_PFA_SET(name, func) \
2059 static inline void task_set_##func(struct task_struct *p) \
2060 { set_bit(PFA_##name, &p->atomic_flags); }
2061 #define TASK_PFA_CLEAR(name, func) \
2062 static inline void task_clear_##func(struct task_struct *p) \
2063 { clear_bit(PFA_##name, &p->atomic_flags); }
2065 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2066 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2068 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2069 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2070 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2072 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2073 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2074 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2077 * task->jobctl flags
2079 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2081 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2082 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2083 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2084 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2085 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2086 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2087 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2089 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
2090 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
2091 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
2092 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
2093 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
2094 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
2095 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
2097 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2098 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2100 extern bool task_set_jobctl_pending(struct task_struct *task,
2101 unsigned int mask);
2102 extern void task_clear_jobctl_trapping(struct task_struct *task);
2103 extern void task_clear_jobctl_pending(struct task_struct *task,
2104 unsigned int mask);
2106 static inline void rcu_copy_process(struct task_struct *p)
2108 #ifdef CONFIG_PREEMPT_RCU
2109 p->rcu_read_lock_nesting = 0;
2110 p->rcu_read_unlock_special.s = 0;
2111 p->rcu_blocked_node = NULL;
2112 INIT_LIST_HEAD(&p->rcu_node_entry);
2113 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2114 #ifdef CONFIG_TASKS_RCU
2115 p->rcu_tasks_holdout = false;
2116 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2117 p->rcu_tasks_idle_cpu = -1;
2118 #endif /* #ifdef CONFIG_TASKS_RCU */
2121 static inline void tsk_restore_flags(struct task_struct *task,
2122 unsigned long orig_flags, unsigned long flags)
2124 task->flags &= ~flags;
2125 task->flags |= orig_flags & flags;
2128 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2129 const struct cpumask *trial);
2130 extern int task_can_attach(struct task_struct *p,
2131 const struct cpumask *cs_cpus_allowed);
2132 #ifdef CONFIG_SMP
2133 extern void do_set_cpus_allowed(struct task_struct *p,
2134 const struct cpumask *new_mask);
2136 extern int set_cpus_allowed_ptr(struct task_struct *p,
2137 const struct cpumask *new_mask);
2138 #else
2139 static inline void do_set_cpus_allowed(struct task_struct *p,
2140 const struct cpumask *new_mask)
2143 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2144 const struct cpumask *new_mask)
2146 if (!cpumask_test_cpu(0, new_mask))
2147 return -EINVAL;
2148 return 0;
2150 #endif
2152 #ifdef CONFIG_NO_HZ_COMMON
2153 void calc_load_enter_idle(void);
2154 void calc_load_exit_idle(void);
2155 #else
2156 static inline void calc_load_enter_idle(void) { }
2157 static inline void calc_load_exit_idle(void) { }
2158 #endif /* CONFIG_NO_HZ_COMMON */
2160 #ifndef CONFIG_CPUMASK_OFFSTACK
2161 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2163 return set_cpus_allowed_ptr(p, &new_mask);
2165 #endif
2168 * Do not use outside of architecture code which knows its limitations.
2170 * sched_clock() has no promise of monotonicity or bounded drift between
2171 * CPUs, use (which you should not) requires disabling IRQs.
2173 * Please use one of the three interfaces below.
2175 extern unsigned long long notrace sched_clock(void);
2177 * See the comment in kernel/sched/clock.c
2179 extern u64 cpu_clock(int cpu);
2180 extern u64 local_clock(void);
2181 extern u64 running_clock(void);
2182 extern u64 sched_clock_cpu(int cpu);
2185 extern void sched_clock_init(void);
2187 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2188 static inline void sched_clock_tick(void)
2192 static inline void sched_clock_idle_sleep_event(void)
2196 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2199 #else
2201 * Architectures can set this to 1 if they have specified
2202 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2203 * but then during bootup it turns out that sched_clock()
2204 * is reliable after all:
2206 extern int sched_clock_stable(void);
2207 extern void set_sched_clock_stable(void);
2208 extern void clear_sched_clock_stable(void);
2210 extern void sched_clock_tick(void);
2211 extern void sched_clock_idle_sleep_event(void);
2212 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2213 #endif
2215 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2217 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2218 * The reason for this explicit opt-in is not to have perf penalty with
2219 * slow sched_clocks.
2221 extern void enable_sched_clock_irqtime(void);
2222 extern void disable_sched_clock_irqtime(void);
2223 #else
2224 static inline void enable_sched_clock_irqtime(void) {}
2225 static inline void disable_sched_clock_irqtime(void) {}
2226 #endif
2228 extern unsigned long long
2229 task_sched_runtime(struct task_struct *task);
2231 /* sched_exec is called by processes performing an exec */
2232 #ifdef CONFIG_SMP
2233 extern void sched_exec(void);
2234 #else
2235 #define sched_exec() {}
2236 #endif
2238 extern void sched_clock_idle_sleep_event(void);
2239 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2241 #ifdef CONFIG_HOTPLUG_CPU
2242 extern void idle_task_exit(void);
2243 #else
2244 static inline void idle_task_exit(void) {}
2245 #endif
2247 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2248 extern void wake_up_nohz_cpu(int cpu);
2249 #else
2250 static inline void wake_up_nohz_cpu(int cpu) { }
2251 #endif
2253 #ifdef CONFIG_NO_HZ_FULL
2254 extern bool sched_can_stop_tick(void);
2255 extern u64 scheduler_tick_max_deferment(void);
2256 #else
2257 static inline bool sched_can_stop_tick(void) { return false; }
2258 #endif
2260 #ifdef CONFIG_SCHED_AUTOGROUP
2261 extern void sched_autogroup_create_attach(struct task_struct *p);
2262 extern void sched_autogroup_detach(struct task_struct *p);
2263 extern void sched_autogroup_fork(struct signal_struct *sig);
2264 extern void sched_autogroup_exit(struct signal_struct *sig);
2265 #ifdef CONFIG_PROC_FS
2266 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2267 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2268 #endif
2269 #else
2270 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2271 static inline void sched_autogroup_detach(struct task_struct *p) { }
2272 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2273 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2274 #endif
2276 extern int yield_to(struct task_struct *p, bool preempt);
2277 extern void set_user_nice(struct task_struct *p, long nice);
2278 extern int task_prio(const struct task_struct *p);
2280 * task_nice - return the nice value of a given task.
2281 * @p: the task in question.
2283 * Return: The nice value [ -20 ... 0 ... 19 ].
2285 static inline int task_nice(const struct task_struct *p)
2287 return PRIO_TO_NICE((p)->static_prio);
2289 extern int can_nice(const struct task_struct *p, const int nice);
2290 extern int task_curr(const struct task_struct *p);
2291 extern int idle_cpu(int cpu);
2292 extern int sched_setscheduler(struct task_struct *, int,
2293 const struct sched_param *);
2294 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2295 const struct sched_param *);
2296 extern int sched_setattr(struct task_struct *,
2297 const struct sched_attr *);
2298 extern struct task_struct *idle_task(int cpu);
2300 * is_idle_task - is the specified task an idle task?
2301 * @p: the task in question.
2303 * Return: 1 if @p is an idle task. 0 otherwise.
2305 static inline bool is_idle_task(const struct task_struct *p)
2307 return p->pid == 0;
2309 extern struct task_struct *curr_task(int cpu);
2310 extern void set_curr_task(int cpu, struct task_struct *p);
2312 void yield(void);
2315 * The default (Linux) execution domain.
2317 extern struct exec_domain default_exec_domain;
2319 union thread_union {
2320 struct thread_info thread_info;
2321 unsigned long stack[THREAD_SIZE/sizeof(long)];
2324 #ifndef __HAVE_ARCH_KSTACK_END
2325 static inline int kstack_end(void *addr)
2327 /* Reliable end of stack detection:
2328 * Some APM bios versions misalign the stack
2330 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2332 #endif
2334 extern union thread_union init_thread_union;
2335 extern struct task_struct init_task;
2337 extern struct mm_struct init_mm;
2339 extern struct pid_namespace init_pid_ns;
2342 * find a task by one of its numerical ids
2344 * find_task_by_pid_ns():
2345 * finds a task by its pid in the specified namespace
2346 * find_task_by_vpid():
2347 * finds a task by its virtual pid
2349 * see also find_vpid() etc in include/linux/pid.h
2352 extern struct task_struct *find_task_by_vpid(pid_t nr);
2353 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2354 struct pid_namespace *ns);
2356 /* per-UID process charging. */
2357 extern struct user_struct * alloc_uid(kuid_t);
2358 static inline struct user_struct *get_uid(struct user_struct *u)
2360 atomic_inc(&u->__count);
2361 return u;
2363 extern void free_uid(struct user_struct *);
2365 #include <asm/current.h>
2367 extern void xtime_update(unsigned long ticks);
2369 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2370 extern int wake_up_process(struct task_struct *tsk);
2371 extern void wake_up_new_task(struct task_struct *tsk);
2372 #ifdef CONFIG_SMP
2373 extern void kick_process(struct task_struct *tsk);
2374 #else
2375 static inline void kick_process(struct task_struct *tsk) { }
2376 #endif
2377 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2378 extern void sched_dead(struct task_struct *p);
2380 extern void proc_caches_init(void);
2381 extern void flush_signals(struct task_struct *);
2382 extern void __flush_signals(struct task_struct *);
2383 extern void ignore_signals(struct task_struct *);
2384 extern void flush_signal_handlers(struct task_struct *, int force_default);
2385 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2387 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2389 unsigned long flags;
2390 int ret;
2392 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2393 ret = dequeue_signal(tsk, mask, info);
2394 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2396 return ret;
2399 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2400 sigset_t *mask);
2401 extern void unblock_all_signals(void);
2402 extern void release_task(struct task_struct * p);
2403 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2404 extern int force_sigsegv(int, struct task_struct *);
2405 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2406 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2407 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2408 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2409 const struct cred *, u32);
2410 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2411 extern int kill_pid(struct pid *pid, int sig, int priv);
2412 extern int kill_proc_info(int, struct siginfo *, pid_t);
2413 extern __must_check bool do_notify_parent(struct task_struct *, int);
2414 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2415 extern void force_sig(int, struct task_struct *);
2416 extern int send_sig(int, struct task_struct *, int);
2417 extern int zap_other_threads(struct task_struct *p);
2418 extern struct sigqueue *sigqueue_alloc(void);
2419 extern void sigqueue_free(struct sigqueue *);
2420 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2421 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2423 static inline void restore_saved_sigmask(void)
2425 if (test_and_clear_restore_sigmask())
2426 __set_current_blocked(&current->saved_sigmask);
2429 static inline sigset_t *sigmask_to_save(void)
2431 sigset_t *res = &current->blocked;
2432 if (unlikely(test_restore_sigmask()))
2433 res = &current->saved_sigmask;
2434 return res;
2437 static inline int kill_cad_pid(int sig, int priv)
2439 return kill_pid(cad_pid, sig, priv);
2442 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2443 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2444 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2445 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2448 * True if we are on the alternate signal stack.
2450 static inline int on_sig_stack(unsigned long sp)
2452 #ifdef CONFIG_STACK_GROWSUP
2453 return sp >= current->sas_ss_sp &&
2454 sp - current->sas_ss_sp < current->sas_ss_size;
2455 #else
2456 return sp > current->sas_ss_sp &&
2457 sp - current->sas_ss_sp <= current->sas_ss_size;
2458 #endif
2461 static inline int sas_ss_flags(unsigned long sp)
2463 if (!current->sas_ss_size)
2464 return SS_DISABLE;
2466 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2469 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2471 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2472 #ifdef CONFIG_STACK_GROWSUP
2473 return current->sas_ss_sp;
2474 #else
2475 return current->sas_ss_sp + current->sas_ss_size;
2476 #endif
2477 return sp;
2481 * Routines for handling mm_structs
2483 extern struct mm_struct * mm_alloc(void);
2485 /* mmdrop drops the mm and the page tables */
2486 extern void __mmdrop(struct mm_struct *);
2487 static inline void mmdrop(struct mm_struct * mm)
2489 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2490 __mmdrop(mm);
2493 /* mmput gets rid of the mappings and all user-space */
2494 extern void mmput(struct mm_struct *);
2495 /* Grab a reference to a task's mm, if it is not already going away */
2496 extern struct mm_struct *get_task_mm(struct task_struct *task);
2498 * Grab a reference to a task's mm, if it is not already going away
2499 * and ptrace_may_access with the mode parameter passed to it
2500 * succeeds.
2502 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2503 /* Remove the current tasks stale references to the old mm_struct */
2504 extern void mm_release(struct task_struct *, struct mm_struct *);
2506 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2507 struct task_struct *);
2508 extern void flush_thread(void);
2509 extern void exit_thread(void);
2511 extern void exit_files(struct task_struct *);
2512 extern void __cleanup_sighand(struct sighand_struct *);
2514 extern void exit_itimers(struct signal_struct *);
2515 extern void flush_itimer_signals(void);
2517 extern void do_group_exit(int);
2519 extern int do_execve(struct filename *,
2520 const char __user * const __user *,
2521 const char __user * const __user *);
2522 extern int do_execveat(int, struct filename *,
2523 const char __user * const __user *,
2524 const char __user * const __user *,
2525 int);
2526 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2527 struct task_struct *fork_idle(int);
2528 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2530 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2531 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2533 __set_task_comm(tsk, from, false);
2535 extern char *get_task_comm(char *to, struct task_struct *tsk);
2537 #ifdef CONFIG_SMP
2538 void scheduler_ipi(void);
2539 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2540 #else
2541 static inline void scheduler_ipi(void) { }
2542 static inline unsigned long wait_task_inactive(struct task_struct *p,
2543 long match_state)
2545 return 1;
2547 #endif
2549 #define next_task(p) \
2550 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2552 #define for_each_process(p) \
2553 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2555 extern bool current_is_single_threaded(void);
2558 * Careful: do_each_thread/while_each_thread is a double loop so
2559 * 'break' will not work as expected - use goto instead.
2561 #define do_each_thread(g, t) \
2562 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2564 #define while_each_thread(g, t) \
2565 while ((t = next_thread(t)) != g)
2567 #define __for_each_thread(signal, t) \
2568 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2570 #define for_each_thread(p, t) \
2571 __for_each_thread((p)->signal, t)
2573 /* Careful: this is a double loop, 'break' won't work as expected. */
2574 #define for_each_process_thread(p, t) \
2575 for_each_process(p) for_each_thread(p, t)
2577 static inline int get_nr_threads(struct task_struct *tsk)
2579 return tsk->signal->nr_threads;
2582 static inline bool thread_group_leader(struct task_struct *p)
2584 return p->exit_signal >= 0;
2587 /* Do to the insanities of de_thread it is possible for a process
2588 * to have the pid of the thread group leader without actually being
2589 * the thread group leader. For iteration through the pids in proc
2590 * all we care about is that we have a task with the appropriate
2591 * pid, we don't actually care if we have the right task.
2593 static inline bool has_group_leader_pid(struct task_struct *p)
2595 return task_pid(p) == p->signal->leader_pid;
2598 static inline
2599 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2601 return p1->signal == p2->signal;
2604 static inline struct task_struct *next_thread(const struct task_struct *p)
2606 return list_entry_rcu(p->thread_group.next,
2607 struct task_struct, thread_group);
2610 static inline int thread_group_empty(struct task_struct *p)
2612 return list_empty(&p->thread_group);
2615 #define delay_group_leader(p) \
2616 (thread_group_leader(p) && !thread_group_empty(p))
2619 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2620 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2621 * pins the final release of task.io_context. Also protects ->cpuset and
2622 * ->cgroup.subsys[]. And ->vfork_done.
2624 * Nests both inside and outside of read_lock(&tasklist_lock).
2625 * It must not be nested with write_lock_irq(&tasklist_lock),
2626 * neither inside nor outside.
2628 static inline void task_lock(struct task_struct *p)
2630 spin_lock(&p->alloc_lock);
2633 static inline void task_unlock(struct task_struct *p)
2635 spin_unlock(&p->alloc_lock);
2638 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2639 unsigned long *flags);
2641 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2642 unsigned long *flags)
2644 struct sighand_struct *ret;
2646 ret = __lock_task_sighand(tsk, flags);
2647 (void)__cond_lock(&tsk->sighand->siglock, ret);
2648 return ret;
2651 static inline void unlock_task_sighand(struct task_struct *tsk,
2652 unsigned long *flags)
2654 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2657 #ifdef CONFIG_CGROUPS
2658 static inline void threadgroup_change_begin(struct task_struct *tsk)
2660 down_read(&tsk->signal->group_rwsem);
2662 static inline void threadgroup_change_end(struct task_struct *tsk)
2664 up_read(&tsk->signal->group_rwsem);
2668 * threadgroup_lock - lock threadgroup
2669 * @tsk: member task of the threadgroup to lock
2671 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2672 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2673 * change ->group_leader/pid. This is useful for cases where the threadgroup
2674 * needs to stay stable across blockable operations.
2676 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2677 * synchronization. While held, no new task will be added to threadgroup
2678 * and no existing live task will have its PF_EXITING set.
2680 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2681 * sub-thread becomes a new leader.
2683 static inline void threadgroup_lock(struct task_struct *tsk)
2685 down_write(&tsk->signal->group_rwsem);
2689 * threadgroup_unlock - unlock threadgroup
2690 * @tsk: member task of the threadgroup to unlock
2692 * Reverse threadgroup_lock().
2694 static inline void threadgroup_unlock(struct task_struct *tsk)
2696 up_write(&tsk->signal->group_rwsem);
2698 #else
2699 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2700 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2701 static inline void threadgroup_lock(struct task_struct *tsk) {}
2702 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2703 #endif
2705 #ifndef __HAVE_THREAD_FUNCTIONS
2707 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2708 #define task_stack_page(task) ((task)->stack)
2710 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2712 *task_thread_info(p) = *task_thread_info(org);
2713 task_thread_info(p)->task = p;
2717 * Return the address of the last usable long on the stack.
2719 * When the stack grows down, this is just above the thread
2720 * info struct. Going any lower will corrupt the threadinfo.
2722 * When the stack grows up, this is the highest address.
2723 * Beyond that position, we corrupt data on the next page.
2725 static inline unsigned long *end_of_stack(struct task_struct *p)
2727 #ifdef CONFIG_STACK_GROWSUP
2728 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2729 #else
2730 return (unsigned long *)(task_thread_info(p) + 1);
2731 #endif
2734 #endif
2735 #define task_stack_end_corrupted(task) \
2736 (*(end_of_stack(task)) != STACK_END_MAGIC)
2738 static inline int object_is_on_stack(void *obj)
2740 void *stack = task_stack_page(current);
2742 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2745 extern void thread_info_cache_init(void);
2747 #ifdef CONFIG_DEBUG_STACK_USAGE
2748 static inline unsigned long stack_not_used(struct task_struct *p)
2750 unsigned long *n = end_of_stack(p);
2752 do { /* Skip over canary */
2753 n++;
2754 } while (!*n);
2756 return (unsigned long)n - (unsigned long)end_of_stack(p);
2758 #endif
2759 extern void set_task_stack_end_magic(struct task_struct *tsk);
2761 /* set thread flags in other task's structures
2762 * - see asm/thread_info.h for TIF_xxxx flags available
2764 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2766 set_ti_thread_flag(task_thread_info(tsk), flag);
2769 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2771 clear_ti_thread_flag(task_thread_info(tsk), flag);
2774 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2776 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2779 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2781 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2784 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2786 return test_ti_thread_flag(task_thread_info(tsk), flag);
2789 static inline void set_tsk_need_resched(struct task_struct *tsk)
2791 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2794 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2796 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2799 static inline int test_tsk_need_resched(struct task_struct *tsk)
2801 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2804 static inline int restart_syscall(void)
2806 set_tsk_thread_flag(current, TIF_SIGPENDING);
2807 return -ERESTARTNOINTR;
2810 static inline int signal_pending(struct task_struct *p)
2812 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2815 static inline int __fatal_signal_pending(struct task_struct *p)
2817 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2820 static inline int fatal_signal_pending(struct task_struct *p)
2822 return signal_pending(p) && __fatal_signal_pending(p);
2825 static inline int signal_pending_state(long state, struct task_struct *p)
2827 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2828 return 0;
2829 if (!signal_pending(p))
2830 return 0;
2832 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2836 * cond_resched() and cond_resched_lock(): latency reduction via
2837 * explicit rescheduling in places that are safe. The return
2838 * value indicates whether a reschedule was done in fact.
2839 * cond_resched_lock() will drop the spinlock before scheduling,
2840 * cond_resched_softirq() will enable bhs before scheduling.
2842 extern int _cond_resched(void);
2844 #define cond_resched() ({ \
2845 ___might_sleep(__FILE__, __LINE__, 0); \
2846 _cond_resched(); \
2849 extern int __cond_resched_lock(spinlock_t *lock);
2851 #ifdef CONFIG_PREEMPT_COUNT
2852 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2853 #else
2854 #define PREEMPT_LOCK_OFFSET 0
2855 #endif
2857 #define cond_resched_lock(lock) ({ \
2858 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2859 __cond_resched_lock(lock); \
2862 extern int __cond_resched_softirq(void);
2864 #define cond_resched_softirq() ({ \
2865 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2866 __cond_resched_softirq(); \
2869 static inline void cond_resched_rcu(void)
2871 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2872 rcu_read_unlock();
2873 cond_resched();
2874 rcu_read_lock();
2875 #endif
2879 * Does a critical section need to be broken due to another
2880 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2881 * but a general need for low latency)
2883 static inline int spin_needbreak(spinlock_t *lock)
2885 #ifdef CONFIG_PREEMPT
2886 return spin_is_contended(lock);
2887 #else
2888 return 0;
2889 #endif
2893 * Idle thread specific functions to determine the need_resched
2894 * polling state.
2896 #ifdef TIF_POLLING_NRFLAG
2897 static inline int tsk_is_polling(struct task_struct *p)
2899 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2902 static inline void __current_set_polling(void)
2904 set_thread_flag(TIF_POLLING_NRFLAG);
2907 static inline bool __must_check current_set_polling_and_test(void)
2909 __current_set_polling();
2912 * Polling state must be visible before we test NEED_RESCHED,
2913 * paired by resched_curr()
2915 smp_mb__after_atomic();
2917 return unlikely(tif_need_resched());
2920 static inline void __current_clr_polling(void)
2922 clear_thread_flag(TIF_POLLING_NRFLAG);
2925 static inline bool __must_check current_clr_polling_and_test(void)
2927 __current_clr_polling();
2930 * Polling state must be visible before we test NEED_RESCHED,
2931 * paired by resched_curr()
2933 smp_mb__after_atomic();
2935 return unlikely(tif_need_resched());
2938 #else
2939 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2940 static inline void __current_set_polling(void) { }
2941 static inline void __current_clr_polling(void) { }
2943 static inline bool __must_check current_set_polling_and_test(void)
2945 return unlikely(tif_need_resched());
2947 static inline bool __must_check current_clr_polling_and_test(void)
2949 return unlikely(tif_need_resched());
2951 #endif
2953 static inline void current_clr_polling(void)
2955 __current_clr_polling();
2958 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2959 * Once the bit is cleared, we'll get IPIs with every new
2960 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2961 * fold.
2963 smp_mb(); /* paired with resched_curr() */
2965 preempt_fold_need_resched();
2968 static __always_inline bool need_resched(void)
2970 return unlikely(tif_need_resched());
2974 * Thread group CPU time accounting.
2976 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2977 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2979 static inline void thread_group_cputime_init(struct signal_struct *sig)
2981 raw_spin_lock_init(&sig->cputimer.lock);
2985 * Reevaluate whether the task has signals pending delivery.
2986 * Wake the task if so.
2987 * This is required every time the blocked sigset_t changes.
2988 * callers must hold sighand->siglock.
2990 extern void recalc_sigpending_and_wake(struct task_struct *t);
2991 extern void recalc_sigpending(void);
2993 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2995 static inline void signal_wake_up(struct task_struct *t, bool resume)
2997 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2999 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3001 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3005 * Wrappers for p->thread_info->cpu access. No-op on UP.
3007 #ifdef CONFIG_SMP
3009 static inline unsigned int task_cpu(const struct task_struct *p)
3011 return task_thread_info(p)->cpu;
3014 static inline int task_node(const struct task_struct *p)
3016 return cpu_to_node(task_cpu(p));
3019 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3021 #else
3023 static inline unsigned int task_cpu(const struct task_struct *p)
3025 return 0;
3028 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3032 #endif /* CONFIG_SMP */
3034 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3035 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3037 #ifdef CONFIG_CGROUP_SCHED
3038 extern struct task_group root_task_group;
3039 #endif /* CONFIG_CGROUP_SCHED */
3041 extern int task_can_switch_user(struct user_struct *up,
3042 struct task_struct *tsk);
3044 #ifdef CONFIG_TASK_XACCT
3045 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3047 tsk->ioac.rchar += amt;
3050 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3052 tsk->ioac.wchar += amt;
3055 static inline void inc_syscr(struct task_struct *tsk)
3057 tsk->ioac.syscr++;
3060 static inline void inc_syscw(struct task_struct *tsk)
3062 tsk->ioac.syscw++;
3064 #else
3065 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3069 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3073 static inline void inc_syscr(struct task_struct *tsk)
3077 static inline void inc_syscw(struct task_struct *tsk)
3080 #endif
3082 #ifndef TASK_SIZE_OF
3083 #define TASK_SIZE_OF(tsk) TASK_SIZE
3084 #endif
3086 #ifdef CONFIG_MEMCG
3087 extern void mm_update_next_owner(struct mm_struct *mm);
3088 #else
3089 static inline void mm_update_next_owner(struct mm_struct *mm)
3092 #endif /* CONFIG_MEMCG */
3094 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3095 unsigned int limit)
3097 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
3100 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3101 unsigned int limit)
3103 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
3106 static inline unsigned long rlimit(unsigned int limit)
3108 return task_rlimit(current, limit);
3111 static inline unsigned long rlimit_max(unsigned int limit)
3113 return task_rlimit_max(current, limit);
3116 #endif