ptrace: implement TRAP_NOTIFY and use it for group stop events
[GitHub/mt8127/android_kernel_alcatel_ttab.git] / include / linux / sched.h
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3
4 /*
5 * cloning flags:
6 */
7 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
8 #define CLONE_VM 0x00000100 /* set if VM shared between processes */
9 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
10 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
11 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
12 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
13 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
14 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
15 #define CLONE_THREAD 0x00010000 /* Same thread group? */
16 #define CLONE_NEWNS 0x00020000 /* New namespace group? */
17 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
18 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
19 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
20 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
21 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
22 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
23 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
24 /* 0x02000000 was previously the unused CLONE_STOPPED (Start in stopped state)
25 and is now available for re-use. */
26 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
27 #define CLONE_NEWIPC 0x08000000 /* New ipcs */
28 #define CLONE_NEWUSER 0x10000000 /* New user namespace */
29 #define CLONE_NEWPID 0x20000000 /* New pid namespace */
30 #define CLONE_NEWNET 0x40000000 /* New network namespace */
31 #define CLONE_IO 0x80000000 /* Clone io context */
32
33 /*
34 * Scheduling policies
35 */
36 #define SCHED_NORMAL 0
37 #define SCHED_FIFO 1
38 #define SCHED_RR 2
39 #define SCHED_BATCH 3
40 /* SCHED_ISO: reserved but not implemented yet */
41 #define SCHED_IDLE 5
42 /* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
43 #define SCHED_RESET_ON_FORK 0x40000000
44
45 #ifdef __KERNEL__
46
47 struct sched_param {
48 int sched_priority;
49 };
50
51 #include <asm/param.h> /* for HZ */
52
53 #include <linux/capability.h>
54 #include <linux/threads.h>
55 #include <linux/kernel.h>
56 #include <linux/types.h>
57 #include <linux/timex.h>
58 #include <linux/jiffies.h>
59 #include <linux/rbtree.h>
60 #include <linux/thread_info.h>
61 #include <linux/cpumask.h>
62 #include <linux/errno.h>
63 #include <linux/nodemask.h>
64 #include <linux/mm_types.h>
65
66 #include <asm/system.h>
67 #include <asm/page.h>
68 #include <asm/ptrace.h>
69 #include <asm/cputime.h>
70
71 #include <linux/smp.h>
72 #include <linux/sem.h>
73 #include <linux/signal.h>
74 #include <linux/compiler.h>
75 #include <linux/completion.h>
76 #include <linux/pid.h>
77 #include <linux/percpu.h>
78 #include <linux/topology.h>
79 #include <linux/proportions.h>
80 #include <linux/seccomp.h>
81 #include <linux/rcupdate.h>
82 #include <linux/rculist.h>
83 #include <linux/rtmutex.h>
84
85 #include <linux/time.h>
86 #include <linux/param.h>
87 #include <linux/resource.h>
88 #include <linux/timer.h>
89 #include <linux/hrtimer.h>
90 #include <linux/task_io_accounting.h>
91 #include <linux/latencytop.h>
92 #include <linux/cred.h>
93
94 #include <asm/processor.h>
95
96 struct exec_domain;
97 struct futex_pi_state;
98 struct robust_list_head;
99 struct bio_list;
100 struct fs_struct;
101 struct perf_event_context;
102 struct blk_plug;
103
104 /*
105 * List of flags we want to share for kernel threads,
106 * if only because they are not used by them anyway.
107 */
108 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
109
110 /*
111 * These are the constant used to fake the fixed-point load-average
112 * counting. Some notes:
113 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
114 * a load-average precision of 10 bits integer + 11 bits fractional
115 * - if you want to count load-averages more often, you need more
116 * precision, or rounding will get you. With 2-second counting freq,
117 * the EXP_n values would be 1981, 2034 and 2043 if still using only
118 * 11 bit fractions.
119 */
120 extern unsigned long avenrun[]; /* Load averages */
121 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
122
123 #define FSHIFT 11 /* nr of bits of precision */
124 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
125 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
126 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
127 #define EXP_5 2014 /* 1/exp(5sec/5min) */
128 #define EXP_15 2037 /* 1/exp(5sec/15min) */
129
130 #define CALC_LOAD(load,exp,n) \
131 load *= exp; \
132 load += n*(FIXED_1-exp); \
133 load >>= FSHIFT;
134
135 extern unsigned long total_forks;
136 extern int nr_threads;
137 DECLARE_PER_CPU(unsigned long, process_counts);
138 extern int nr_processes(void);
139 extern unsigned long nr_running(void);
140 extern unsigned long nr_uninterruptible(void);
141 extern unsigned long nr_iowait(void);
142 extern unsigned long nr_iowait_cpu(int cpu);
143 extern unsigned long this_cpu_load(void);
144
145
146 extern void calc_global_load(unsigned long ticks);
147
148 extern unsigned long get_parent_ip(unsigned long addr);
149
150 struct seq_file;
151 struct cfs_rq;
152 struct task_group;
153 #ifdef CONFIG_SCHED_DEBUG
154 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
155 extern void proc_sched_set_task(struct task_struct *p);
156 extern void
157 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
158 #else
159 static inline void
160 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
161 {
162 }
163 static inline void proc_sched_set_task(struct task_struct *p)
164 {
165 }
166 static inline void
167 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
168 {
169 }
170 #endif
171
172 /*
173 * Task state bitmask. NOTE! These bits are also
174 * encoded in fs/proc/array.c: get_task_state().
175 *
176 * We have two separate sets of flags: task->state
177 * is about runnability, while task->exit_state are
178 * about the task exiting. Confusing, but this way
179 * modifying one set can't modify the other one by
180 * mistake.
181 */
182 #define TASK_RUNNING 0
183 #define TASK_INTERRUPTIBLE 1
184 #define TASK_UNINTERRUPTIBLE 2
185 #define __TASK_STOPPED 4
186 #define __TASK_TRACED 8
187 /* in tsk->exit_state */
188 #define EXIT_ZOMBIE 16
189 #define EXIT_DEAD 32
190 /* in tsk->state again */
191 #define TASK_DEAD 64
192 #define TASK_WAKEKILL 128
193 #define TASK_WAKING 256
194 #define TASK_STATE_MAX 512
195
196 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
197
198 extern char ___assert_task_state[1 - 2*!!(
199 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
200
201 /* Convenience macros for the sake of set_task_state */
202 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
203 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
204 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
205
206 /* Convenience macros for the sake of wake_up */
207 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
208 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
209
210 /* get_task_state() */
211 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
212 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
213 __TASK_TRACED)
214
215 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
216 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
217 #define task_is_dead(task) ((task)->exit_state != 0)
218 #define task_is_stopped_or_traced(task) \
219 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
220 #define task_contributes_to_load(task) \
221 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
222 (task->flags & PF_FREEZING) == 0)
223
224 #define __set_task_state(tsk, state_value) \
225 do { (tsk)->state = (state_value); } while (0)
226 #define set_task_state(tsk, state_value) \
227 set_mb((tsk)->state, (state_value))
228
229 /*
230 * set_current_state() includes a barrier so that the write of current->state
231 * is correctly serialised wrt the caller's subsequent test of whether to
232 * actually sleep:
233 *
234 * set_current_state(TASK_UNINTERRUPTIBLE);
235 * if (do_i_need_to_sleep())
236 * schedule();
237 *
238 * If the caller does not need such serialisation then use __set_current_state()
239 */
240 #define __set_current_state(state_value) \
241 do { current->state = (state_value); } while (0)
242 #define set_current_state(state_value) \
243 set_mb(current->state, (state_value))
244
245 /* Task command name length */
246 #define TASK_COMM_LEN 16
247
248 #include <linux/spinlock.h>
249
250 /*
251 * This serializes "schedule()" and also protects
252 * the run-queue from deletions/modifications (but
253 * _adding_ to the beginning of the run-queue has
254 * a separate lock).
255 */
256 extern rwlock_t tasklist_lock;
257 extern spinlock_t mmlist_lock;
258
259 struct task_struct;
260
261 #ifdef CONFIG_PROVE_RCU
262 extern int lockdep_tasklist_lock_is_held(void);
263 #endif /* #ifdef CONFIG_PROVE_RCU */
264
265 extern void sched_init(void);
266 extern void sched_init_smp(void);
267 extern asmlinkage void schedule_tail(struct task_struct *prev);
268 extern void init_idle(struct task_struct *idle, int cpu);
269 extern void init_idle_bootup_task(struct task_struct *idle);
270
271 extern int runqueue_is_locked(int cpu);
272
273 extern cpumask_var_t nohz_cpu_mask;
274 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
275 extern void select_nohz_load_balancer(int stop_tick);
276 extern int get_nohz_timer_target(void);
277 #else
278 static inline void select_nohz_load_balancer(int stop_tick) { }
279 #endif
280
281 /*
282 * Only dump TASK_* tasks. (0 for all tasks)
283 */
284 extern void show_state_filter(unsigned long state_filter);
285
286 static inline void show_state(void)
287 {
288 show_state_filter(0);
289 }
290
291 extern void show_regs(struct pt_regs *);
292
293 /*
294 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
295 * task), SP is the stack pointer of the first frame that should be shown in the back
296 * trace (or NULL if the entire call-chain of the task should be shown).
297 */
298 extern void show_stack(struct task_struct *task, unsigned long *sp);
299
300 void io_schedule(void);
301 long io_schedule_timeout(long timeout);
302
303 extern void cpu_init (void);
304 extern void trap_init(void);
305 extern void update_process_times(int user);
306 extern void scheduler_tick(void);
307
308 extern void sched_show_task(struct task_struct *p);
309
310 #ifdef CONFIG_LOCKUP_DETECTOR
311 extern void touch_softlockup_watchdog(void);
312 extern void touch_softlockup_watchdog_sync(void);
313 extern void touch_all_softlockup_watchdogs(void);
314 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
315 void __user *buffer,
316 size_t *lenp, loff_t *ppos);
317 extern unsigned int softlockup_panic;
318 void lockup_detector_init(void);
319 #else
320 static inline void touch_softlockup_watchdog(void)
321 {
322 }
323 static inline void touch_softlockup_watchdog_sync(void)
324 {
325 }
326 static inline void touch_all_softlockup_watchdogs(void)
327 {
328 }
329 static inline void lockup_detector_init(void)
330 {
331 }
332 #endif
333
334 #ifdef CONFIG_DETECT_HUNG_TASK
335 extern unsigned int sysctl_hung_task_panic;
336 extern unsigned long sysctl_hung_task_check_count;
337 extern unsigned long sysctl_hung_task_timeout_secs;
338 extern unsigned long sysctl_hung_task_warnings;
339 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
340 void __user *buffer,
341 size_t *lenp, loff_t *ppos);
342 #else
343 /* Avoid need for ifdefs elsewhere in the code */
344 enum { sysctl_hung_task_timeout_secs = 0 };
345 #endif
346
347 /* Attach to any functions which should be ignored in wchan output. */
348 #define __sched __attribute__((__section__(".sched.text")))
349
350 /* Linker adds these: start and end of __sched functions */
351 extern char __sched_text_start[], __sched_text_end[];
352
353 /* Is this address in the __sched functions? */
354 extern int in_sched_functions(unsigned long addr);
355
356 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
357 extern signed long schedule_timeout(signed long timeout);
358 extern signed long schedule_timeout_interruptible(signed long timeout);
359 extern signed long schedule_timeout_killable(signed long timeout);
360 extern signed long schedule_timeout_uninterruptible(signed long timeout);
361 asmlinkage void schedule(void);
362 extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
363
364 struct nsproxy;
365 struct user_namespace;
366
367 /*
368 * Default maximum number of active map areas, this limits the number of vmas
369 * per mm struct. Users can overwrite this number by sysctl but there is a
370 * problem.
371 *
372 * When a program's coredump is generated as ELF format, a section is created
373 * per a vma. In ELF, the number of sections is represented in unsigned short.
374 * This means the number of sections should be smaller than 65535 at coredump.
375 * Because the kernel adds some informative sections to a image of program at
376 * generating coredump, we need some margin. The number of extra sections is
377 * 1-3 now and depends on arch. We use "5" as safe margin, here.
378 */
379 #define MAPCOUNT_ELF_CORE_MARGIN (5)
380 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
381
382 extern int sysctl_max_map_count;
383
384 #include <linux/aio.h>
385
386 #ifdef CONFIG_MMU
387 extern void arch_pick_mmap_layout(struct mm_struct *mm);
388 extern unsigned long
389 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
390 unsigned long, unsigned long);
391 extern unsigned long
392 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
393 unsigned long len, unsigned long pgoff,
394 unsigned long flags);
395 extern void arch_unmap_area(struct mm_struct *, unsigned long);
396 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
397 #else
398 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
399 #endif
400
401
402 extern void set_dumpable(struct mm_struct *mm, int value);
403 extern int get_dumpable(struct mm_struct *mm);
404
405 /* mm flags */
406 /* dumpable bits */
407 #define MMF_DUMPABLE 0 /* core dump is permitted */
408 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
409
410 #define MMF_DUMPABLE_BITS 2
411 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
412
413 /* coredump filter bits */
414 #define MMF_DUMP_ANON_PRIVATE 2
415 #define MMF_DUMP_ANON_SHARED 3
416 #define MMF_DUMP_MAPPED_PRIVATE 4
417 #define MMF_DUMP_MAPPED_SHARED 5
418 #define MMF_DUMP_ELF_HEADERS 6
419 #define MMF_DUMP_HUGETLB_PRIVATE 7
420 #define MMF_DUMP_HUGETLB_SHARED 8
421
422 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
423 #define MMF_DUMP_FILTER_BITS 7
424 #define MMF_DUMP_FILTER_MASK \
425 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
426 #define MMF_DUMP_FILTER_DEFAULT \
427 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
428 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
429
430 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
431 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
432 #else
433 # define MMF_DUMP_MASK_DEFAULT_ELF 0
434 #endif
435 /* leave room for more dump flags */
436 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
437 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
438
439 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
440
441 struct sighand_struct {
442 atomic_t count;
443 struct k_sigaction action[_NSIG];
444 spinlock_t siglock;
445 wait_queue_head_t signalfd_wqh;
446 };
447
448 struct pacct_struct {
449 int ac_flag;
450 long ac_exitcode;
451 unsigned long ac_mem;
452 cputime_t ac_utime, ac_stime;
453 unsigned long ac_minflt, ac_majflt;
454 };
455
456 struct cpu_itimer {
457 cputime_t expires;
458 cputime_t incr;
459 u32 error;
460 u32 incr_error;
461 };
462
463 /**
464 * struct task_cputime - collected CPU time counts
465 * @utime: time spent in user mode, in &cputime_t units
466 * @stime: time spent in kernel mode, in &cputime_t units
467 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
468 *
469 * This structure groups together three kinds of CPU time that are
470 * tracked for threads and thread groups. Most things considering
471 * CPU time want to group these counts together and treat all three
472 * of them in parallel.
473 */
474 struct task_cputime {
475 cputime_t utime;
476 cputime_t stime;
477 unsigned long long sum_exec_runtime;
478 };
479 /* Alternate field names when used to cache expirations. */
480 #define prof_exp stime
481 #define virt_exp utime
482 #define sched_exp sum_exec_runtime
483
484 #define INIT_CPUTIME \
485 (struct task_cputime) { \
486 .utime = cputime_zero, \
487 .stime = cputime_zero, \
488 .sum_exec_runtime = 0, \
489 }
490
491 /*
492 * Disable preemption until the scheduler is running.
493 * Reset by start_kernel()->sched_init()->init_idle().
494 *
495 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
496 * before the scheduler is active -- see should_resched().
497 */
498 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
499
500 /**
501 * struct thread_group_cputimer - thread group interval timer counts
502 * @cputime: thread group interval timers.
503 * @running: non-zero when there are timers running and
504 * @cputime receives updates.
505 * @lock: lock for fields in this struct.
506 *
507 * This structure contains the version of task_cputime, above, that is
508 * used for thread group CPU timer calculations.
509 */
510 struct thread_group_cputimer {
511 struct task_cputime cputime;
512 int running;
513 spinlock_t lock;
514 };
515
516 #include <linux/rwsem.h>
517 struct autogroup;
518
519 /*
520 * NOTE! "signal_struct" does not have its own
521 * locking, because a shared signal_struct always
522 * implies a shared sighand_struct, so locking
523 * sighand_struct is always a proper superset of
524 * the locking of signal_struct.
525 */
526 struct signal_struct {
527 atomic_t sigcnt;
528 atomic_t live;
529 int nr_threads;
530
531 wait_queue_head_t wait_chldexit; /* for wait4() */
532
533 /* current thread group signal load-balancing target: */
534 struct task_struct *curr_target;
535
536 /* shared signal handling: */
537 struct sigpending shared_pending;
538
539 /* thread group exit support */
540 int group_exit_code;
541 /* overloaded:
542 * - notify group_exit_task when ->count is equal to notify_count
543 * - everyone except group_exit_task is stopped during signal delivery
544 * of fatal signals, group_exit_task processes the signal.
545 */
546 int notify_count;
547 struct task_struct *group_exit_task;
548
549 /* thread group stop support, overloads group_exit_code too */
550 int group_stop_count;
551 unsigned int flags; /* see SIGNAL_* flags below */
552
553 /* POSIX.1b Interval Timers */
554 struct list_head posix_timers;
555
556 /* ITIMER_REAL timer for the process */
557 struct hrtimer real_timer;
558 struct pid *leader_pid;
559 ktime_t it_real_incr;
560
561 /*
562 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
563 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
564 * values are defined to 0 and 1 respectively
565 */
566 struct cpu_itimer it[2];
567
568 /*
569 * Thread group totals for process CPU timers.
570 * See thread_group_cputimer(), et al, for details.
571 */
572 struct thread_group_cputimer cputimer;
573
574 /* Earliest-expiration cache. */
575 struct task_cputime cputime_expires;
576
577 struct list_head cpu_timers[3];
578
579 struct pid *tty_old_pgrp;
580
581 /* boolean value for session group leader */
582 int leader;
583
584 struct tty_struct *tty; /* NULL if no tty */
585
586 #ifdef CONFIG_SCHED_AUTOGROUP
587 struct autogroup *autogroup;
588 #endif
589 /*
590 * Cumulative resource counters for dead threads in the group,
591 * and for reaped dead child processes forked by this group.
592 * Live threads maintain their own counters and add to these
593 * in __exit_signal, except for the group leader.
594 */
595 cputime_t utime, stime, cutime, cstime;
596 cputime_t gtime;
597 cputime_t cgtime;
598 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
599 cputime_t prev_utime, prev_stime;
600 #endif
601 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
602 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
603 unsigned long inblock, oublock, cinblock, coublock;
604 unsigned long maxrss, cmaxrss;
605 struct task_io_accounting ioac;
606
607 /*
608 * Cumulative ns of schedule CPU time fo dead threads in the
609 * group, not including a zombie group leader, (This only differs
610 * from jiffies_to_ns(utime + stime) if sched_clock uses something
611 * other than jiffies.)
612 */
613 unsigned long long sum_sched_runtime;
614
615 /*
616 * We don't bother to synchronize most readers of this at all,
617 * because there is no reader checking a limit that actually needs
618 * to get both rlim_cur and rlim_max atomically, and either one
619 * alone is a single word that can safely be read normally.
620 * getrlimit/setrlimit use task_lock(current->group_leader) to
621 * protect this instead of the siglock, because they really
622 * have no need to disable irqs.
623 */
624 struct rlimit rlim[RLIM_NLIMITS];
625
626 #ifdef CONFIG_BSD_PROCESS_ACCT
627 struct pacct_struct pacct; /* per-process accounting information */
628 #endif
629 #ifdef CONFIG_TASKSTATS
630 struct taskstats *stats;
631 #endif
632 #ifdef CONFIG_AUDIT
633 unsigned audit_tty;
634 struct tty_audit_buf *tty_audit_buf;
635 #endif
636 #ifdef CONFIG_CGROUPS
637 /*
638 * The threadgroup_fork_lock prevents threads from forking with
639 * CLONE_THREAD while held for writing. Use this for fork-sensitive
640 * threadgroup-wide operations. It's taken for reading in fork.c in
641 * copy_process().
642 * Currently only needed write-side by cgroups.
643 */
644 struct rw_semaphore threadgroup_fork_lock;
645 #endif
646
647 int oom_adj; /* OOM kill score adjustment (bit shift) */
648 int oom_score_adj; /* OOM kill score adjustment */
649 int oom_score_adj_min; /* OOM kill score adjustment minimum value.
650 * Only settable by CAP_SYS_RESOURCE. */
651
652 struct mutex cred_guard_mutex; /* guard against foreign influences on
653 * credential calculations
654 * (notably. ptrace) */
655 };
656
657 /* Context switch must be unlocked if interrupts are to be enabled */
658 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
659 # define __ARCH_WANT_UNLOCKED_CTXSW
660 #endif
661
662 /*
663 * Bits in flags field of signal_struct.
664 */
665 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
666 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
667 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
668 /*
669 * Pending notifications to parent.
670 */
671 #define SIGNAL_CLD_STOPPED 0x00000010
672 #define SIGNAL_CLD_CONTINUED 0x00000020
673 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
674
675 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
676
677 /* If true, all threads except ->group_exit_task have pending SIGKILL */
678 static inline int signal_group_exit(const struct signal_struct *sig)
679 {
680 return (sig->flags & SIGNAL_GROUP_EXIT) ||
681 (sig->group_exit_task != NULL);
682 }
683
684 /*
685 * Some day this will be a full-fledged user tracking system..
686 */
687 struct user_struct {
688 atomic_t __count; /* reference count */
689 atomic_t processes; /* How many processes does this user have? */
690 atomic_t files; /* How many open files does this user have? */
691 atomic_t sigpending; /* How many pending signals does this user have? */
692 #ifdef CONFIG_INOTIFY_USER
693 atomic_t inotify_watches; /* How many inotify watches does this user have? */
694 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
695 #endif
696 #ifdef CONFIG_FANOTIFY
697 atomic_t fanotify_listeners;
698 #endif
699 #ifdef CONFIG_EPOLL
700 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
701 #endif
702 #ifdef CONFIG_POSIX_MQUEUE
703 /* protected by mq_lock */
704 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
705 #endif
706 unsigned long locked_shm; /* How many pages of mlocked shm ? */
707
708 #ifdef CONFIG_KEYS
709 struct key *uid_keyring; /* UID specific keyring */
710 struct key *session_keyring; /* UID's default session keyring */
711 #endif
712
713 /* Hash table maintenance information */
714 struct hlist_node uidhash_node;
715 uid_t uid;
716 struct user_namespace *user_ns;
717
718 #ifdef CONFIG_PERF_EVENTS
719 atomic_long_t locked_vm;
720 #endif
721 };
722
723 extern int uids_sysfs_init(void);
724
725 extern struct user_struct *find_user(uid_t);
726
727 extern struct user_struct root_user;
728 #define INIT_USER (&root_user)
729
730
731 struct backing_dev_info;
732 struct reclaim_state;
733
734 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
735 struct sched_info {
736 /* cumulative counters */
737 unsigned long pcount; /* # of times run on this cpu */
738 unsigned long long run_delay; /* time spent waiting on a runqueue */
739
740 /* timestamps */
741 unsigned long long last_arrival,/* when we last ran on a cpu */
742 last_queued; /* when we were last queued to run */
743 };
744 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
745
746 #ifdef CONFIG_TASK_DELAY_ACCT
747 struct task_delay_info {
748 spinlock_t lock;
749 unsigned int flags; /* Private per-task flags */
750
751 /* For each stat XXX, add following, aligned appropriately
752 *
753 * struct timespec XXX_start, XXX_end;
754 * u64 XXX_delay;
755 * u32 XXX_count;
756 *
757 * Atomicity of updates to XXX_delay, XXX_count protected by
758 * single lock above (split into XXX_lock if contention is an issue).
759 */
760
761 /*
762 * XXX_count is incremented on every XXX operation, the delay
763 * associated with the operation is added to XXX_delay.
764 * XXX_delay contains the accumulated delay time in nanoseconds.
765 */
766 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
767 u64 blkio_delay; /* wait for sync block io completion */
768 u64 swapin_delay; /* wait for swapin block io completion */
769 u32 blkio_count; /* total count of the number of sync block */
770 /* io operations performed */
771 u32 swapin_count; /* total count of the number of swapin block */
772 /* io operations performed */
773
774 struct timespec freepages_start, freepages_end;
775 u64 freepages_delay; /* wait for memory reclaim */
776 u32 freepages_count; /* total count of memory reclaim */
777 };
778 #endif /* CONFIG_TASK_DELAY_ACCT */
779
780 static inline int sched_info_on(void)
781 {
782 #ifdef CONFIG_SCHEDSTATS
783 return 1;
784 #elif defined(CONFIG_TASK_DELAY_ACCT)
785 extern int delayacct_on;
786 return delayacct_on;
787 #else
788 return 0;
789 #endif
790 }
791
792 enum cpu_idle_type {
793 CPU_IDLE,
794 CPU_NOT_IDLE,
795 CPU_NEWLY_IDLE,
796 CPU_MAX_IDLE_TYPES
797 };
798
799 /*
800 * Increase resolution of nice-level calculations for 64-bit architectures.
801 * The extra resolution improves shares distribution and load balancing of
802 * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
803 * hierarchies, especially on larger systems. This is not a user-visible change
804 * and does not change the user-interface for setting shares/weights.
805 *
806 * We increase resolution only if we have enough bits to allow this increased
807 * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
808 * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
809 * increased costs.
810 */
811 #if BITS_PER_LONG > 32
812 # define SCHED_LOAD_RESOLUTION 10
813 # define scale_load(w) ((w) << SCHED_LOAD_RESOLUTION)
814 # define scale_load_down(w) ((w) >> SCHED_LOAD_RESOLUTION)
815 #else
816 # define SCHED_LOAD_RESOLUTION 0
817 # define scale_load(w) (w)
818 # define scale_load_down(w) (w)
819 #endif
820
821 #define SCHED_LOAD_SHIFT (10 + SCHED_LOAD_RESOLUTION)
822 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
823
824 /*
825 * Increase resolution of cpu_power calculations
826 */
827 #define SCHED_POWER_SHIFT 10
828 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
829
830 /*
831 * sched-domains (multiprocessor balancing) declarations:
832 */
833 #ifdef CONFIG_SMP
834 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
835 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
836 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
837 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
838 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
839 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
840 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
841 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
842 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
843 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
844 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
845 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
846 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
847
848 enum powersavings_balance_level {
849 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
850 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
851 * first for long running threads
852 */
853 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
854 * cpu package for power savings
855 */
856 MAX_POWERSAVINGS_BALANCE_LEVELS
857 };
858
859 extern int sched_mc_power_savings, sched_smt_power_savings;
860
861 static inline int sd_balance_for_mc_power(void)
862 {
863 if (sched_smt_power_savings)
864 return SD_POWERSAVINGS_BALANCE;
865
866 if (!sched_mc_power_savings)
867 return SD_PREFER_SIBLING;
868
869 return 0;
870 }
871
872 static inline int sd_balance_for_package_power(void)
873 {
874 if (sched_mc_power_savings | sched_smt_power_savings)
875 return SD_POWERSAVINGS_BALANCE;
876
877 return SD_PREFER_SIBLING;
878 }
879
880 extern int __weak arch_sd_sibiling_asym_packing(void);
881
882 /*
883 * Optimise SD flags for power savings:
884 * SD_BALANCE_NEWIDLE helps aggressive task consolidation and power savings.
885 * Keep default SD flags if sched_{smt,mc}_power_saving=0
886 */
887
888 static inline int sd_power_saving_flags(void)
889 {
890 if (sched_mc_power_savings | sched_smt_power_savings)
891 return SD_BALANCE_NEWIDLE;
892
893 return 0;
894 }
895
896 struct sched_group {
897 struct sched_group *next; /* Must be a circular list */
898 atomic_t ref;
899
900 /*
901 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
902 * single CPU.
903 */
904 unsigned int cpu_power, cpu_power_orig;
905 unsigned int group_weight;
906
907 /*
908 * The CPUs this group covers.
909 *
910 * NOTE: this field is variable length. (Allocated dynamically
911 * by attaching extra space to the end of the structure,
912 * depending on how many CPUs the kernel has booted up with)
913 */
914 unsigned long cpumask[0];
915 };
916
917 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
918 {
919 return to_cpumask(sg->cpumask);
920 }
921
922 struct sched_domain_attr {
923 int relax_domain_level;
924 };
925
926 #define SD_ATTR_INIT (struct sched_domain_attr) { \
927 .relax_domain_level = -1, \
928 }
929
930 extern int sched_domain_level_max;
931
932 struct sched_domain {
933 /* These fields must be setup */
934 struct sched_domain *parent; /* top domain must be null terminated */
935 struct sched_domain *child; /* bottom domain must be null terminated */
936 struct sched_group *groups; /* the balancing groups of the domain */
937 unsigned long min_interval; /* Minimum balance interval ms */
938 unsigned long max_interval; /* Maximum balance interval ms */
939 unsigned int busy_factor; /* less balancing by factor if busy */
940 unsigned int imbalance_pct; /* No balance until over watermark */
941 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
942 unsigned int busy_idx;
943 unsigned int idle_idx;
944 unsigned int newidle_idx;
945 unsigned int wake_idx;
946 unsigned int forkexec_idx;
947 unsigned int smt_gain;
948 int flags; /* See SD_* */
949 int level;
950
951 /* Runtime fields. */
952 unsigned long last_balance; /* init to jiffies. units in jiffies */
953 unsigned int balance_interval; /* initialise to 1. units in ms. */
954 unsigned int nr_balance_failed; /* initialise to 0 */
955
956 u64 last_update;
957
958 #ifdef CONFIG_SCHEDSTATS
959 /* load_balance() stats */
960 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
961 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
962 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
963 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
964 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
965 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
966 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
967 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
968
969 /* Active load balancing */
970 unsigned int alb_count;
971 unsigned int alb_failed;
972 unsigned int alb_pushed;
973
974 /* SD_BALANCE_EXEC stats */
975 unsigned int sbe_count;
976 unsigned int sbe_balanced;
977 unsigned int sbe_pushed;
978
979 /* SD_BALANCE_FORK stats */
980 unsigned int sbf_count;
981 unsigned int sbf_balanced;
982 unsigned int sbf_pushed;
983
984 /* try_to_wake_up() stats */
985 unsigned int ttwu_wake_remote;
986 unsigned int ttwu_move_affine;
987 unsigned int ttwu_move_balance;
988 #endif
989 #ifdef CONFIG_SCHED_DEBUG
990 char *name;
991 #endif
992 union {
993 void *private; /* used during construction */
994 struct rcu_head rcu; /* used during destruction */
995 };
996
997 unsigned int span_weight;
998 /*
999 * Span of all CPUs in this domain.
1000 *
1001 * NOTE: this field is variable length. (Allocated dynamically
1002 * by attaching extra space to the end of the structure,
1003 * depending on how many CPUs the kernel has booted up with)
1004 */
1005 unsigned long span[0];
1006 };
1007
1008 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1009 {
1010 return to_cpumask(sd->span);
1011 }
1012
1013 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1014 struct sched_domain_attr *dattr_new);
1015
1016 /* Allocate an array of sched domains, for partition_sched_domains(). */
1017 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1018 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1019
1020 /* Test a flag in parent sched domain */
1021 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1022 {
1023 if (sd->parent && (sd->parent->flags & flag))
1024 return 1;
1025
1026 return 0;
1027 }
1028
1029 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1030 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1031
1032 #else /* CONFIG_SMP */
1033
1034 struct sched_domain_attr;
1035
1036 static inline void
1037 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1038 struct sched_domain_attr *dattr_new)
1039 {
1040 }
1041 #endif /* !CONFIG_SMP */
1042
1043
1044 struct io_context; /* See blkdev.h */
1045
1046
1047 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1048 extern void prefetch_stack(struct task_struct *t);
1049 #else
1050 static inline void prefetch_stack(struct task_struct *t) { }
1051 #endif
1052
1053 struct audit_context; /* See audit.c */
1054 struct mempolicy;
1055 struct pipe_inode_info;
1056 struct uts_namespace;
1057
1058 struct rq;
1059 struct sched_domain;
1060
1061 /*
1062 * wake flags
1063 */
1064 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1065 #define WF_FORK 0x02 /* child wakeup after fork */
1066
1067 #define ENQUEUE_WAKEUP 1
1068 #define ENQUEUE_HEAD 2
1069 #ifdef CONFIG_SMP
1070 #define ENQUEUE_WAKING 4 /* sched_class::task_waking was called */
1071 #else
1072 #define ENQUEUE_WAKING 0
1073 #endif
1074
1075 #define DEQUEUE_SLEEP 1
1076
1077 struct sched_class {
1078 const struct sched_class *next;
1079
1080 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1081 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1082 void (*yield_task) (struct rq *rq);
1083 bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1084
1085 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1086
1087 struct task_struct * (*pick_next_task) (struct rq *rq);
1088 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1089
1090 #ifdef CONFIG_SMP
1091 int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1092
1093 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1094 void (*post_schedule) (struct rq *this_rq);
1095 void (*task_waking) (struct task_struct *task);
1096 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1097
1098 void (*set_cpus_allowed)(struct task_struct *p,
1099 const struct cpumask *newmask);
1100
1101 void (*rq_online)(struct rq *rq);
1102 void (*rq_offline)(struct rq *rq);
1103 #endif
1104
1105 void (*set_curr_task) (struct rq *rq);
1106 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1107 void (*task_fork) (struct task_struct *p);
1108
1109 void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1110 void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1111 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1112 int oldprio);
1113
1114 unsigned int (*get_rr_interval) (struct rq *rq,
1115 struct task_struct *task);
1116
1117 #ifdef CONFIG_FAIR_GROUP_SCHED
1118 void (*task_move_group) (struct task_struct *p, int on_rq);
1119 #endif
1120 };
1121
1122 struct load_weight {
1123 unsigned long weight, inv_weight;
1124 };
1125
1126 #ifdef CONFIG_SCHEDSTATS
1127 struct sched_statistics {
1128 u64 wait_start;
1129 u64 wait_max;
1130 u64 wait_count;
1131 u64 wait_sum;
1132 u64 iowait_count;
1133 u64 iowait_sum;
1134
1135 u64 sleep_start;
1136 u64 sleep_max;
1137 s64 sum_sleep_runtime;
1138
1139 u64 block_start;
1140 u64 block_max;
1141 u64 exec_max;
1142 u64 slice_max;
1143
1144 u64 nr_migrations_cold;
1145 u64 nr_failed_migrations_affine;
1146 u64 nr_failed_migrations_running;
1147 u64 nr_failed_migrations_hot;
1148 u64 nr_forced_migrations;
1149
1150 u64 nr_wakeups;
1151 u64 nr_wakeups_sync;
1152 u64 nr_wakeups_migrate;
1153 u64 nr_wakeups_local;
1154 u64 nr_wakeups_remote;
1155 u64 nr_wakeups_affine;
1156 u64 nr_wakeups_affine_attempts;
1157 u64 nr_wakeups_passive;
1158 u64 nr_wakeups_idle;
1159 };
1160 #endif
1161
1162 struct sched_entity {
1163 struct load_weight load; /* for load-balancing */
1164 struct rb_node run_node;
1165 struct list_head group_node;
1166 unsigned int on_rq;
1167
1168 u64 exec_start;
1169 u64 sum_exec_runtime;
1170 u64 vruntime;
1171 u64 prev_sum_exec_runtime;
1172
1173 u64 nr_migrations;
1174
1175 #ifdef CONFIG_SCHEDSTATS
1176 struct sched_statistics statistics;
1177 #endif
1178
1179 #ifdef CONFIG_FAIR_GROUP_SCHED
1180 struct sched_entity *parent;
1181 /* rq on which this entity is (to be) queued: */
1182 struct cfs_rq *cfs_rq;
1183 /* rq "owned" by this entity/group: */
1184 struct cfs_rq *my_q;
1185 #endif
1186 };
1187
1188 struct sched_rt_entity {
1189 struct list_head run_list;
1190 unsigned long timeout;
1191 unsigned int time_slice;
1192 int nr_cpus_allowed;
1193
1194 struct sched_rt_entity *back;
1195 #ifdef CONFIG_RT_GROUP_SCHED
1196 struct sched_rt_entity *parent;
1197 /* rq on which this entity is (to be) queued: */
1198 struct rt_rq *rt_rq;
1199 /* rq "owned" by this entity/group: */
1200 struct rt_rq *my_q;
1201 #endif
1202 };
1203
1204 struct rcu_node;
1205
1206 enum perf_event_task_context {
1207 perf_invalid_context = -1,
1208 perf_hw_context = 0,
1209 perf_sw_context,
1210 perf_nr_task_contexts,
1211 };
1212
1213 struct task_struct {
1214 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1215 void *stack;
1216 atomic_t usage;
1217 unsigned int flags; /* per process flags, defined below */
1218 unsigned int ptrace;
1219
1220 #ifdef CONFIG_SMP
1221 struct task_struct *wake_entry;
1222 int on_cpu;
1223 #endif
1224 int on_rq;
1225
1226 int prio, static_prio, normal_prio;
1227 unsigned int rt_priority;
1228 const struct sched_class *sched_class;
1229 struct sched_entity se;
1230 struct sched_rt_entity rt;
1231
1232 #ifdef CONFIG_PREEMPT_NOTIFIERS
1233 /* list of struct preempt_notifier: */
1234 struct hlist_head preempt_notifiers;
1235 #endif
1236
1237 /*
1238 * fpu_counter contains the number of consecutive context switches
1239 * that the FPU is used. If this is over a threshold, the lazy fpu
1240 * saving becomes unlazy to save the trap. This is an unsigned char
1241 * so that after 256 times the counter wraps and the behavior turns
1242 * lazy again; this to deal with bursty apps that only use FPU for
1243 * a short time
1244 */
1245 unsigned char fpu_counter;
1246 #ifdef CONFIG_BLK_DEV_IO_TRACE
1247 unsigned int btrace_seq;
1248 #endif
1249
1250 unsigned int policy;
1251 cpumask_t cpus_allowed;
1252
1253 #ifdef CONFIG_PREEMPT_RCU
1254 int rcu_read_lock_nesting;
1255 char rcu_read_unlock_special;
1256 struct list_head rcu_node_entry;
1257 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1258 #ifdef CONFIG_TREE_PREEMPT_RCU
1259 struct rcu_node *rcu_blocked_node;
1260 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1261 #ifdef CONFIG_RCU_BOOST
1262 struct rt_mutex *rcu_boost_mutex;
1263 #endif /* #ifdef CONFIG_RCU_BOOST */
1264
1265 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1266 struct sched_info sched_info;
1267 #endif
1268
1269 struct list_head tasks;
1270 #ifdef CONFIG_SMP
1271 struct plist_node pushable_tasks;
1272 #endif
1273
1274 struct mm_struct *mm, *active_mm;
1275 #ifdef CONFIG_COMPAT_BRK
1276 unsigned brk_randomized:1;
1277 #endif
1278 #if defined(SPLIT_RSS_COUNTING)
1279 struct task_rss_stat rss_stat;
1280 #endif
1281 /* task state */
1282 int exit_state;
1283 int exit_code, exit_signal;
1284 int pdeath_signal; /* The signal sent when the parent dies */
1285 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1286 /* ??? */
1287 unsigned int personality;
1288 unsigned did_exec:1;
1289 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1290 * execve */
1291 unsigned in_iowait:1;
1292
1293
1294 /* Revert to default priority/policy when forking */
1295 unsigned sched_reset_on_fork:1;
1296 unsigned sched_contributes_to_load:1;
1297
1298 pid_t pid;
1299 pid_t tgid;
1300
1301 #ifdef CONFIG_CC_STACKPROTECTOR
1302 /* Canary value for the -fstack-protector gcc feature */
1303 unsigned long stack_canary;
1304 #endif
1305
1306 /*
1307 * pointers to (original) parent process, youngest child, younger sibling,
1308 * older sibling, respectively. (p->father can be replaced with
1309 * p->real_parent->pid)
1310 */
1311 struct task_struct *real_parent; /* real parent process */
1312 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1313 /*
1314 * children/sibling forms the list of my natural children
1315 */
1316 struct list_head children; /* list of my children */
1317 struct list_head sibling; /* linkage in my parent's children list */
1318 struct task_struct *group_leader; /* threadgroup leader */
1319
1320 /*
1321 * ptraced is the list of tasks this task is using ptrace on.
1322 * This includes both natural children and PTRACE_ATTACH targets.
1323 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1324 */
1325 struct list_head ptraced;
1326 struct list_head ptrace_entry;
1327
1328 /* PID/PID hash table linkage. */
1329 struct pid_link pids[PIDTYPE_MAX];
1330 struct list_head thread_group;
1331
1332 struct completion *vfork_done; /* for vfork() */
1333 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1334 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1335
1336 cputime_t utime, stime, utimescaled, stimescaled;
1337 cputime_t gtime;
1338 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1339 cputime_t prev_utime, prev_stime;
1340 #endif
1341 unsigned long nvcsw, nivcsw; /* context switch counts */
1342 struct timespec start_time; /* monotonic time */
1343 struct timespec real_start_time; /* boot based time */
1344 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1345 unsigned long min_flt, maj_flt;
1346
1347 struct task_cputime cputime_expires;
1348 struct list_head cpu_timers[3];
1349
1350 /* process credentials */
1351 const struct cred __rcu *real_cred; /* objective and real subjective task
1352 * credentials (COW) */
1353 const struct cred __rcu *cred; /* effective (overridable) subjective task
1354 * credentials (COW) */
1355 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1356
1357 char comm[TASK_COMM_LEN]; /* executable name excluding path
1358 - access with [gs]et_task_comm (which lock
1359 it with task_lock())
1360 - initialized normally by setup_new_exec */
1361 /* file system info */
1362 int link_count, total_link_count;
1363 #ifdef CONFIG_SYSVIPC
1364 /* ipc stuff */
1365 struct sysv_sem sysvsem;
1366 #endif
1367 #ifdef CONFIG_DETECT_HUNG_TASK
1368 /* hung task detection */
1369 unsigned long last_switch_count;
1370 #endif
1371 /* CPU-specific state of this task */
1372 struct thread_struct thread;
1373 /* filesystem information */
1374 struct fs_struct *fs;
1375 /* open file information */
1376 struct files_struct *files;
1377 /* namespaces */
1378 struct nsproxy *nsproxy;
1379 /* signal handlers */
1380 struct signal_struct *signal;
1381 struct sighand_struct *sighand;
1382
1383 sigset_t blocked, real_blocked;
1384 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1385 struct sigpending pending;
1386
1387 unsigned long sas_ss_sp;
1388 size_t sas_ss_size;
1389 int (*notifier)(void *priv);
1390 void *notifier_data;
1391 sigset_t *notifier_mask;
1392 struct audit_context *audit_context;
1393 #ifdef CONFIG_AUDITSYSCALL
1394 uid_t loginuid;
1395 unsigned int sessionid;
1396 #endif
1397 seccomp_t seccomp;
1398
1399 /* Thread group tracking */
1400 u32 parent_exec_id;
1401 u32 self_exec_id;
1402 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1403 * mempolicy */
1404 spinlock_t alloc_lock;
1405
1406 #ifdef CONFIG_GENERIC_HARDIRQS
1407 /* IRQ handler threads */
1408 struct irqaction *irqaction;
1409 #endif
1410
1411 /* Protection of the PI data structures: */
1412 raw_spinlock_t pi_lock;
1413
1414 #ifdef CONFIG_RT_MUTEXES
1415 /* PI waiters blocked on a rt_mutex held by this task */
1416 struct plist_head pi_waiters;
1417 /* Deadlock detection and priority inheritance handling */
1418 struct rt_mutex_waiter *pi_blocked_on;
1419 #endif
1420
1421 #ifdef CONFIG_DEBUG_MUTEXES
1422 /* mutex deadlock detection */
1423 struct mutex_waiter *blocked_on;
1424 #endif
1425 #ifdef CONFIG_TRACE_IRQFLAGS
1426 unsigned int irq_events;
1427 unsigned long hardirq_enable_ip;
1428 unsigned long hardirq_disable_ip;
1429 unsigned int hardirq_enable_event;
1430 unsigned int hardirq_disable_event;
1431 int hardirqs_enabled;
1432 int hardirq_context;
1433 unsigned long softirq_disable_ip;
1434 unsigned long softirq_enable_ip;
1435 unsigned int softirq_disable_event;
1436 unsigned int softirq_enable_event;
1437 int softirqs_enabled;
1438 int softirq_context;
1439 #endif
1440 #ifdef CONFIG_LOCKDEP
1441 # define MAX_LOCK_DEPTH 48UL
1442 u64 curr_chain_key;
1443 int lockdep_depth;
1444 unsigned int lockdep_recursion;
1445 struct held_lock held_locks[MAX_LOCK_DEPTH];
1446 gfp_t lockdep_reclaim_gfp;
1447 #endif
1448
1449 /* journalling filesystem info */
1450 void *journal_info;
1451
1452 /* stacked block device info */
1453 struct bio_list *bio_list;
1454
1455 #ifdef CONFIG_BLOCK
1456 /* stack plugging */
1457 struct blk_plug *plug;
1458 #endif
1459
1460 /* VM state */
1461 struct reclaim_state *reclaim_state;
1462
1463 struct backing_dev_info *backing_dev_info;
1464
1465 struct io_context *io_context;
1466
1467 unsigned long ptrace_message;
1468 siginfo_t *last_siginfo; /* For ptrace use. */
1469 struct task_io_accounting ioac;
1470 #if defined(CONFIG_TASK_XACCT)
1471 u64 acct_rss_mem1; /* accumulated rss usage */
1472 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1473 cputime_t acct_timexpd; /* stime + utime since last update */
1474 #endif
1475 #ifdef CONFIG_CPUSETS
1476 nodemask_t mems_allowed; /* Protected by alloc_lock */
1477 int mems_allowed_change_disable;
1478 int cpuset_mem_spread_rotor;
1479 int cpuset_slab_spread_rotor;
1480 #endif
1481 #ifdef CONFIG_CGROUPS
1482 /* Control Group info protected by css_set_lock */
1483 struct css_set __rcu *cgroups;
1484 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1485 struct list_head cg_list;
1486 #endif
1487 #ifdef CONFIG_FUTEX
1488 struct robust_list_head __user *robust_list;
1489 #ifdef CONFIG_COMPAT
1490 struct compat_robust_list_head __user *compat_robust_list;
1491 #endif
1492 struct list_head pi_state_list;
1493 struct futex_pi_state *pi_state_cache;
1494 #endif
1495 #ifdef CONFIG_PERF_EVENTS
1496 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1497 struct mutex perf_event_mutex;
1498 struct list_head perf_event_list;
1499 #endif
1500 #ifdef CONFIG_NUMA
1501 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1502 short il_next;
1503 short pref_node_fork;
1504 #endif
1505 atomic_t fs_excl; /* holding fs exclusive resources */
1506 struct rcu_head rcu;
1507
1508 /*
1509 * cache last used pipe for splice
1510 */
1511 struct pipe_inode_info *splice_pipe;
1512 #ifdef CONFIG_TASK_DELAY_ACCT
1513 struct task_delay_info *delays;
1514 #endif
1515 #ifdef CONFIG_FAULT_INJECTION
1516 int make_it_fail;
1517 #endif
1518 struct prop_local_single dirties;
1519 #ifdef CONFIG_LATENCYTOP
1520 int latency_record_count;
1521 struct latency_record latency_record[LT_SAVECOUNT];
1522 #endif
1523 /*
1524 * time slack values; these are used to round up poll() and
1525 * select() etc timeout values. These are in nanoseconds.
1526 */
1527 unsigned long timer_slack_ns;
1528 unsigned long default_timer_slack_ns;
1529
1530 struct list_head *scm_work_list;
1531 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1532 /* Index of current stored address in ret_stack */
1533 int curr_ret_stack;
1534 /* Stack of return addresses for return function tracing */
1535 struct ftrace_ret_stack *ret_stack;
1536 /* time stamp for last schedule */
1537 unsigned long long ftrace_timestamp;
1538 /*
1539 * Number of functions that haven't been traced
1540 * because of depth overrun.
1541 */
1542 atomic_t trace_overrun;
1543 /* Pause for the tracing */
1544 atomic_t tracing_graph_pause;
1545 #endif
1546 #ifdef CONFIG_TRACING
1547 /* state flags for use by tracers */
1548 unsigned long trace;
1549 /* bitmask and counter of trace recursion */
1550 unsigned long trace_recursion;
1551 #endif /* CONFIG_TRACING */
1552 #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1553 struct memcg_batch_info {
1554 int do_batch; /* incremented when batch uncharge started */
1555 struct mem_cgroup *memcg; /* target memcg of uncharge */
1556 unsigned long nr_pages; /* uncharged usage */
1557 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1558 } memcg_batch;
1559 #endif
1560 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1561 atomic_t ptrace_bp_refcnt;
1562 #endif
1563 };
1564
1565 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1566 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1567
1568 /*
1569 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1570 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1571 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1572 * values are inverted: lower p->prio value means higher priority.
1573 *
1574 * The MAX_USER_RT_PRIO value allows the actual maximum
1575 * RT priority to be separate from the value exported to
1576 * user-space. This allows kernel threads to set their
1577 * priority to a value higher than any user task. Note:
1578 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1579 */
1580
1581 #define MAX_USER_RT_PRIO 100
1582 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1583
1584 #define MAX_PRIO (MAX_RT_PRIO + 40)
1585 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1586
1587 static inline int rt_prio(int prio)
1588 {
1589 if (unlikely(prio < MAX_RT_PRIO))
1590 return 1;
1591 return 0;
1592 }
1593
1594 static inline int rt_task(struct task_struct *p)
1595 {
1596 return rt_prio(p->prio);
1597 }
1598
1599 static inline struct pid *task_pid(struct task_struct *task)
1600 {
1601 return task->pids[PIDTYPE_PID].pid;
1602 }
1603
1604 static inline struct pid *task_tgid(struct task_struct *task)
1605 {
1606 return task->group_leader->pids[PIDTYPE_PID].pid;
1607 }
1608
1609 /*
1610 * Without tasklist or rcu lock it is not safe to dereference
1611 * the result of task_pgrp/task_session even if task == current,
1612 * we can race with another thread doing sys_setsid/sys_setpgid.
1613 */
1614 static inline struct pid *task_pgrp(struct task_struct *task)
1615 {
1616 return task->group_leader->pids[PIDTYPE_PGID].pid;
1617 }
1618
1619 static inline struct pid *task_session(struct task_struct *task)
1620 {
1621 return task->group_leader->pids[PIDTYPE_SID].pid;
1622 }
1623
1624 struct pid_namespace;
1625
1626 /*
1627 * the helpers to get the task's different pids as they are seen
1628 * from various namespaces
1629 *
1630 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1631 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1632 * current.
1633 * task_xid_nr_ns() : id seen from the ns specified;
1634 *
1635 * set_task_vxid() : assigns a virtual id to a task;
1636 *
1637 * see also pid_nr() etc in include/linux/pid.h
1638 */
1639 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1640 struct pid_namespace *ns);
1641
1642 static inline pid_t task_pid_nr(struct task_struct *tsk)
1643 {
1644 return tsk->pid;
1645 }
1646
1647 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1648 struct pid_namespace *ns)
1649 {
1650 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1651 }
1652
1653 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1654 {
1655 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1656 }
1657
1658
1659 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1660 {
1661 return tsk->tgid;
1662 }
1663
1664 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1665
1666 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1667 {
1668 return pid_vnr(task_tgid(tsk));
1669 }
1670
1671
1672 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1673 struct pid_namespace *ns)
1674 {
1675 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1676 }
1677
1678 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1679 {
1680 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1681 }
1682
1683
1684 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1685 struct pid_namespace *ns)
1686 {
1687 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1688 }
1689
1690 static inline pid_t task_session_vnr(struct task_struct *tsk)
1691 {
1692 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1693 }
1694
1695 /* obsolete, do not use */
1696 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1697 {
1698 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1699 }
1700
1701 /**
1702 * pid_alive - check that a task structure is not stale
1703 * @p: Task structure to be checked.
1704 *
1705 * Test if a process is not yet dead (at most zombie state)
1706 * If pid_alive fails, then pointers within the task structure
1707 * can be stale and must not be dereferenced.
1708 */
1709 static inline int pid_alive(struct task_struct *p)
1710 {
1711 return p->pids[PIDTYPE_PID].pid != NULL;
1712 }
1713
1714 /**
1715 * is_global_init - check if a task structure is init
1716 * @tsk: Task structure to be checked.
1717 *
1718 * Check if a task structure is the first user space task the kernel created.
1719 */
1720 static inline int is_global_init(struct task_struct *tsk)
1721 {
1722 return tsk->pid == 1;
1723 }
1724
1725 /*
1726 * is_container_init:
1727 * check whether in the task is init in its own pid namespace.
1728 */
1729 extern int is_container_init(struct task_struct *tsk);
1730
1731 extern struct pid *cad_pid;
1732
1733 extern void free_task(struct task_struct *tsk);
1734 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1735
1736 extern void __put_task_struct(struct task_struct *t);
1737
1738 static inline void put_task_struct(struct task_struct *t)
1739 {
1740 if (atomic_dec_and_test(&t->usage))
1741 __put_task_struct(t);
1742 }
1743
1744 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1745 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1746
1747 /*
1748 * Per process flags
1749 */
1750 #define PF_STARTING 0x00000002 /* being created */
1751 #define PF_EXITING 0x00000004 /* getting shut down */
1752 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1753 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1754 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1755 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1756 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1757 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1758 #define PF_DUMPCORE 0x00000200 /* dumped core */
1759 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1760 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1761 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1762 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1763 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1764 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1765 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1766 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1767 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1768 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1769 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1770 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1771 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1772 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1773 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1774 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1775 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1776 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1777 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1778 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1779
1780 /*
1781 * Only the _current_ task can read/write to tsk->flags, but other
1782 * tasks can access tsk->flags in readonly mode for example
1783 * with tsk_used_math (like during threaded core dumping).
1784 * There is however an exception to this rule during ptrace
1785 * or during fork: the ptracer task is allowed to write to the
1786 * child->flags of its traced child (same goes for fork, the parent
1787 * can write to the child->flags), because we're guaranteed the
1788 * child is not running and in turn not changing child->flags
1789 * at the same time the parent does it.
1790 */
1791 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1792 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1793 #define clear_used_math() clear_stopped_child_used_math(current)
1794 #define set_used_math() set_stopped_child_used_math(current)
1795 #define conditional_stopped_child_used_math(condition, child) \
1796 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1797 #define conditional_used_math(condition) \
1798 conditional_stopped_child_used_math(condition, current)
1799 #define copy_to_stopped_child_used_math(child) \
1800 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1801 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1802 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1803 #define used_math() tsk_used_math(current)
1804
1805 /*
1806 * task->jobctl flags
1807 */
1808 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1809
1810 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1811 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1812 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1813 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1814 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1815 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1816
1817 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1818 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1819 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1820 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1821 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1822 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1823
1824 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1825 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1826
1827 extern bool task_set_jobctl_pending(struct task_struct *task,
1828 unsigned int mask);
1829 extern void task_clear_jobctl_trapping(struct task_struct *task);
1830 extern void task_clear_jobctl_pending(struct task_struct *task,
1831 unsigned int mask);
1832
1833 #ifdef CONFIG_PREEMPT_RCU
1834
1835 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1836 #define RCU_READ_UNLOCK_BOOSTED (1 << 1) /* boosted while in RCU read-side. */
1837 #define RCU_READ_UNLOCK_NEED_QS (1 << 2) /* RCU core needs CPU response. */
1838
1839 static inline void rcu_copy_process(struct task_struct *p)
1840 {
1841 p->rcu_read_lock_nesting = 0;
1842 p->rcu_read_unlock_special = 0;
1843 #ifdef CONFIG_TREE_PREEMPT_RCU
1844 p->rcu_blocked_node = NULL;
1845 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1846 #ifdef CONFIG_RCU_BOOST
1847 p->rcu_boost_mutex = NULL;
1848 #endif /* #ifdef CONFIG_RCU_BOOST */
1849 INIT_LIST_HEAD(&p->rcu_node_entry);
1850 }
1851
1852 #else
1853
1854 static inline void rcu_copy_process(struct task_struct *p)
1855 {
1856 }
1857
1858 #endif
1859
1860 #ifdef CONFIG_SMP
1861 extern void do_set_cpus_allowed(struct task_struct *p,
1862 const struct cpumask *new_mask);
1863
1864 extern int set_cpus_allowed_ptr(struct task_struct *p,
1865 const struct cpumask *new_mask);
1866 #else
1867 static inline void do_set_cpus_allowed(struct task_struct *p,
1868 const struct cpumask *new_mask)
1869 {
1870 }
1871 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1872 const struct cpumask *new_mask)
1873 {
1874 if (!cpumask_test_cpu(0, new_mask))
1875 return -EINVAL;
1876 return 0;
1877 }
1878 #endif
1879
1880 #ifndef CONFIG_CPUMASK_OFFSTACK
1881 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1882 {
1883 return set_cpus_allowed_ptr(p, &new_mask);
1884 }
1885 #endif
1886
1887 /*
1888 * Do not use outside of architecture code which knows its limitations.
1889 *
1890 * sched_clock() has no promise of monotonicity or bounded drift between
1891 * CPUs, use (which you should not) requires disabling IRQs.
1892 *
1893 * Please use one of the three interfaces below.
1894 */
1895 extern unsigned long long notrace sched_clock(void);
1896 /*
1897 * See the comment in kernel/sched_clock.c
1898 */
1899 extern u64 cpu_clock(int cpu);
1900 extern u64 local_clock(void);
1901 extern u64 sched_clock_cpu(int cpu);
1902
1903
1904 extern void sched_clock_init(void);
1905
1906 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1907 static inline void sched_clock_tick(void)
1908 {
1909 }
1910
1911 static inline void sched_clock_idle_sleep_event(void)
1912 {
1913 }
1914
1915 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1916 {
1917 }
1918 #else
1919 /*
1920 * Architectures can set this to 1 if they have specified
1921 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1922 * but then during bootup it turns out that sched_clock()
1923 * is reliable after all:
1924 */
1925 extern int sched_clock_stable;
1926
1927 extern void sched_clock_tick(void);
1928 extern void sched_clock_idle_sleep_event(void);
1929 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1930 #endif
1931
1932 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1933 /*
1934 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1935 * The reason for this explicit opt-in is not to have perf penalty with
1936 * slow sched_clocks.
1937 */
1938 extern void enable_sched_clock_irqtime(void);
1939 extern void disable_sched_clock_irqtime(void);
1940 #else
1941 static inline void enable_sched_clock_irqtime(void) {}
1942 static inline void disable_sched_clock_irqtime(void) {}
1943 #endif
1944
1945 extern unsigned long long
1946 task_sched_runtime(struct task_struct *task);
1947 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1948
1949 /* sched_exec is called by processes performing an exec */
1950 #ifdef CONFIG_SMP
1951 extern void sched_exec(void);
1952 #else
1953 #define sched_exec() {}
1954 #endif
1955
1956 extern void sched_clock_idle_sleep_event(void);
1957 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1958
1959 #ifdef CONFIG_HOTPLUG_CPU
1960 extern void idle_task_exit(void);
1961 #else
1962 static inline void idle_task_exit(void) {}
1963 #endif
1964
1965 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1966 extern void wake_up_idle_cpu(int cpu);
1967 #else
1968 static inline void wake_up_idle_cpu(int cpu) { }
1969 #endif
1970
1971 extern unsigned int sysctl_sched_latency;
1972 extern unsigned int sysctl_sched_min_granularity;
1973 extern unsigned int sysctl_sched_wakeup_granularity;
1974 extern unsigned int sysctl_sched_child_runs_first;
1975
1976 enum sched_tunable_scaling {
1977 SCHED_TUNABLESCALING_NONE,
1978 SCHED_TUNABLESCALING_LOG,
1979 SCHED_TUNABLESCALING_LINEAR,
1980 SCHED_TUNABLESCALING_END,
1981 };
1982 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1983
1984 #ifdef CONFIG_SCHED_DEBUG
1985 extern unsigned int sysctl_sched_migration_cost;
1986 extern unsigned int sysctl_sched_nr_migrate;
1987 extern unsigned int sysctl_sched_time_avg;
1988 extern unsigned int sysctl_timer_migration;
1989 extern unsigned int sysctl_sched_shares_window;
1990
1991 int sched_proc_update_handler(struct ctl_table *table, int write,
1992 void __user *buffer, size_t *length,
1993 loff_t *ppos);
1994 #endif
1995 #ifdef CONFIG_SCHED_DEBUG
1996 static inline unsigned int get_sysctl_timer_migration(void)
1997 {
1998 return sysctl_timer_migration;
1999 }
2000 #else
2001 static inline unsigned int get_sysctl_timer_migration(void)
2002 {
2003 return 1;
2004 }
2005 #endif
2006 extern unsigned int sysctl_sched_rt_period;
2007 extern int sysctl_sched_rt_runtime;
2008
2009 int sched_rt_handler(struct ctl_table *table, int write,
2010 void __user *buffer, size_t *lenp,
2011 loff_t *ppos);
2012
2013 #ifdef CONFIG_SCHED_AUTOGROUP
2014 extern unsigned int sysctl_sched_autogroup_enabled;
2015
2016 extern void sched_autogroup_create_attach(struct task_struct *p);
2017 extern void sched_autogroup_detach(struct task_struct *p);
2018 extern void sched_autogroup_fork(struct signal_struct *sig);
2019 extern void sched_autogroup_exit(struct signal_struct *sig);
2020 #ifdef CONFIG_PROC_FS
2021 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2022 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice);
2023 #endif
2024 #else
2025 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2026 static inline void sched_autogroup_detach(struct task_struct *p) { }
2027 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2028 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2029 #endif
2030
2031 #ifdef CONFIG_RT_MUTEXES
2032 extern int rt_mutex_getprio(struct task_struct *p);
2033 extern void rt_mutex_setprio(struct task_struct *p, int prio);
2034 extern void rt_mutex_adjust_pi(struct task_struct *p);
2035 #else
2036 static inline int rt_mutex_getprio(struct task_struct *p)
2037 {
2038 return p->normal_prio;
2039 }
2040 # define rt_mutex_adjust_pi(p) do { } while (0)
2041 #endif
2042
2043 extern bool yield_to(struct task_struct *p, bool preempt);
2044 extern void set_user_nice(struct task_struct *p, long nice);
2045 extern int task_prio(const struct task_struct *p);
2046 extern int task_nice(const struct task_struct *p);
2047 extern int can_nice(const struct task_struct *p, const int nice);
2048 extern int task_curr(const struct task_struct *p);
2049 extern int idle_cpu(int cpu);
2050 extern int sched_setscheduler(struct task_struct *, int,
2051 const struct sched_param *);
2052 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2053 const struct sched_param *);
2054 extern struct task_struct *idle_task(int cpu);
2055 extern struct task_struct *curr_task(int cpu);
2056 extern void set_curr_task(int cpu, struct task_struct *p);
2057
2058 void yield(void);
2059
2060 /*
2061 * The default (Linux) execution domain.
2062 */
2063 extern struct exec_domain default_exec_domain;
2064
2065 union thread_union {
2066 struct thread_info thread_info;
2067 unsigned long stack[THREAD_SIZE/sizeof(long)];
2068 };
2069
2070 #ifndef __HAVE_ARCH_KSTACK_END
2071 static inline int kstack_end(void *addr)
2072 {
2073 /* Reliable end of stack detection:
2074 * Some APM bios versions misalign the stack
2075 */
2076 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2077 }
2078 #endif
2079
2080 extern union thread_union init_thread_union;
2081 extern struct task_struct init_task;
2082
2083 extern struct mm_struct init_mm;
2084
2085 extern struct pid_namespace init_pid_ns;
2086
2087 /*
2088 * find a task by one of its numerical ids
2089 *
2090 * find_task_by_pid_ns():
2091 * finds a task by its pid in the specified namespace
2092 * find_task_by_vpid():
2093 * finds a task by its virtual pid
2094 *
2095 * see also find_vpid() etc in include/linux/pid.h
2096 */
2097
2098 extern struct task_struct *find_task_by_vpid(pid_t nr);
2099 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2100 struct pid_namespace *ns);
2101
2102 extern void __set_special_pids(struct pid *pid);
2103
2104 /* per-UID process charging. */
2105 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
2106 static inline struct user_struct *get_uid(struct user_struct *u)
2107 {
2108 atomic_inc(&u->__count);
2109 return u;
2110 }
2111 extern void free_uid(struct user_struct *);
2112 extern void release_uids(struct user_namespace *ns);
2113
2114 #include <asm/current.h>
2115
2116 extern void xtime_update(unsigned long ticks);
2117
2118 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2119 extern int wake_up_process(struct task_struct *tsk);
2120 extern void wake_up_new_task(struct task_struct *tsk);
2121 #ifdef CONFIG_SMP
2122 extern void kick_process(struct task_struct *tsk);
2123 #else
2124 static inline void kick_process(struct task_struct *tsk) { }
2125 #endif
2126 extern void sched_fork(struct task_struct *p);
2127 extern void sched_dead(struct task_struct *p);
2128
2129 extern void proc_caches_init(void);
2130 extern void flush_signals(struct task_struct *);
2131 extern void __flush_signals(struct task_struct *);
2132 extern void ignore_signals(struct task_struct *);
2133 extern void flush_signal_handlers(struct task_struct *, int force_default);
2134 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2135
2136 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2137 {
2138 unsigned long flags;
2139 int ret;
2140
2141 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2142 ret = dequeue_signal(tsk, mask, info);
2143 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2144
2145 return ret;
2146 }
2147
2148 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2149 sigset_t *mask);
2150 extern void unblock_all_signals(void);
2151 extern void release_task(struct task_struct * p);
2152 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2153 extern int force_sigsegv(int, struct task_struct *);
2154 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2155 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2156 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2157 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2158 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2159 extern int kill_pid(struct pid *pid, int sig, int priv);
2160 extern int kill_proc_info(int, struct siginfo *, pid_t);
2161 extern int do_notify_parent(struct task_struct *, int);
2162 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2163 extern void force_sig(int, struct task_struct *);
2164 extern int send_sig(int, struct task_struct *, int);
2165 extern int zap_other_threads(struct task_struct *p);
2166 extern struct sigqueue *sigqueue_alloc(void);
2167 extern void sigqueue_free(struct sigqueue *);
2168 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2169 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2170 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2171
2172 static inline int kill_cad_pid(int sig, int priv)
2173 {
2174 return kill_pid(cad_pid, sig, priv);
2175 }
2176
2177 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2178 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2179 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2180 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2181
2182 /*
2183 * True if we are on the alternate signal stack.
2184 */
2185 static inline int on_sig_stack(unsigned long sp)
2186 {
2187 #ifdef CONFIG_STACK_GROWSUP
2188 return sp >= current->sas_ss_sp &&
2189 sp - current->sas_ss_sp < current->sas_ss_size;
2190 #else
2191 return sp > current->sas_ss_sp &&
2192 sp - current->sas_ss_sp <= current->sas_ss_size;
2193 #endif
2194 }
2195
2196 static inline int sas_ss_flags(unsigned long sp)
2197 {
2198 return (current->sas_ss_size == 0 ? SS_DISABLE
2199 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2200 }
2201
2202 /*
2203 * Routines for handling mm_structs
2204 */
2205 extern struct mm_struct * mm_alloc(void);
2206
2207 /* mmdrop drops the mm and the page tables */
2208 extern void __mmdrop(struct mm_struct *);
2209 static inline void mmdrop(struct mm_struct * mm)
2210 {
2211 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2212 __mmdrop(mm);
2213 }
2214
2215 /* mmput gets rid of the mappings and all user-space */
2216 extern void mmput(struct mm_struct *);
2217 /* Grab a reference to a task's mm, if it is not already going away */
2218 extern struct mm_struct *get_task_mm(struct task_struct *task);
2219 /* Remove the current tasks stale references to the old mm_struct */
2220 extern void mm_release(struct task_struct *, struct mm_struct *);
2221 /* Allocate a new mm structure and copy contents from tsk->mm */
2222 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2223
2224 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2225 struct task_struct *, struct pt_regs *);
2226 extern void flush_thread(void);
2227 extern void exit_thread(void);
2228
2229 extern void exit_files(struct task_struct *);
2230 extern void __cleanup_sighand(struct sighand_struct *);
2231
2232 extern void exit_itimers(struct signal_struct *);
2233 extern void flush_itimer_signals(void);
2234
2235 extern NORET_TYPE void do_group_exit(int);
2236
2237 extern void daemonize(const char *, ...);
2238 extern int allow_signal(int);
2239 extern int disallow_signal(int);
2240
2241 extern int do_execve(const char *,
2242 const char __user * const __user *,
2243 const char __user * const __user *, struct pt_regs *);
2244 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2245 struct task_struct *fork_idle(int);
2246
2247 extern void set_task_comm(struct task_struct *tsk, char *from);
2248 extern char *get_task_comm(char *to, struct task_struct *tsk);
2249
2250 #ifdef CONFIG_SMP
2251 void scheduler_ipi(void);
2252 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2253 #else
2254 static inline void scheduler_ipi(void) { }
2255 static inline unsigned long wait_task_inactive(struct task_struct *p,
2256 long match_state)
2257 {
2258 return 1;
2259 }
2260 #endif
2261
2262 #define next_task(p) \
2263 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2264
2265 #define for_each_process(p) \
2266 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2267
2268 extern bool current_is_single_threaded(void);
2269
2270 /*
2271 * Careful: do_each_thread/while_each_thread is a double loop so
2272 * 'break' will not work as expected - use goto instead.
2273 */
2274 #define do_each_thread(g, t) \
2275 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2276
2277 #define while_each_thread(g, t) \
2278 while ((t = next_thread(t)) != g)
2279
2280 static inline int get_nr_threads(struct task_struct *tsk)
2281 {
2282 return tsk->signal->nr_threads;
2283 }
2284
2285 /* de_thread depends on thread_group_leader not being a pid based check */
2286 #define thread_group_leader(p) (p == p->group_leader)
2287
2288 /* Do to the insanities of de_thread it is possible for a process
2289 * to have the pid of the thread group leader without actually being
2290 * the thread group leader. For iteration through the pids in proc
2291 * all we care about is that we have a task with the appropriate
2292 * pid, we don't actually care if we have the right task.
2293 */
2294 static inline int has_group_leader_pid(struct task_struct *p)
2295 {
2296 return p->pid == p->tgid;
2297 }
2298
2299 static inline
2300 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2301 {
2302 return p1->tgid == p2->tgid;
2303 }
2304
2305 static inline struct task_struct *next_thread(const struct task_struct *p)
2306 {
2307 return list_entry_rcu(p->thread_group.next,
2308 struct task_struct, thread_group);
2309 }
2310
2311 static inline int thread_group_empty(struct task_struct *p)
2312 {
2313 return list_empty(&p->thread_group);
2314 }
2315
2316 #define delay_group_leader(p) \
2317 (thread_group_leader(p) && !thread_group_empty(p))
2318
2319 static inline int task_detached(struct task_struct *p)
2320 {
2321 return p->exit_signal == -1;
2322 }
2323
2324 /*
2325 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2326 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2327 * pins the final release of task.io_context. Also protects ->cpuset and
2328 * ->cgroup.subsys[].
2329 *
2330 * Nests both inside and outside of read_lock(&tasklist_lock).
2331 * It must not be nested with write_lock_irq(&tasklist_lock),
2332 * neither inside nor outside.
2333 */
2334 static inline void task_lock(struct task_struct *p)
2335 {
2336 spin_lock(&p->alloc_lock);
2337 }
2338
2339 static inline void task_unlock(struct task_struct *p)
2340 {
2341 spin_unlock(&p->alloc_lock);
2342 }
2343
2344 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2345 unsigned long *flags);
2346
2347 #define lock_task_sighand(tsk, flags) \
2348 ({ struct sighand_struct *__ss; \
2349 __cond_lock(&(tsk)->sighand->siglock, \
2350 (__ss = __lock_task_sighand(tsk, flags))); \
2351 __ss; \
2352 }) \
2353
2354 static inline void unlock_task_sighand(struct task_struct *tsk,
2355 unsigned long *flags)
2356 {
2357 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2358 }
2359
2360 /* See the declaration of threadgroup_fork_lock in signal_struct. */
2361 #ifdef CONFIG_CGROUPS
2362 static inline void threadgroup_fork_read_lock(struct task_struct *tsk)
2363 {
2364 down_read(&tsk->signal->threadgroup_fork_lock);
2365 }
2366 static inline void threadgroup_fork_read_unlock(struct task_struct *tsk)
2367 {
2368 up_read(&tsk->signal->threadgroup_fork_lock);
2369 }
2370 static inline void threadgroup_fork_write_lock(struct task_struct *tsk)
2371 {
2372 down_write(&tsk->signal->threadgroup_fork_lock);
2373 }
2374 static inline void threadgroup_fork_write_unlock(struct task_struct *tsk)
2375 {
2376 up_write(&tsk->signal->threadgroup_fork_lock);
2377 }
2378 #else
2379 static inline void threadgroup_fork_read_lock(struct task_struct *tsk) {}
2380 static inline void threadgroup_fork_read_unlock(struct task_struct *tsk) {}
2381 static inline void threadgroup_fork_write_lock(struct task_struct *tsk) {}
2382 static inline void threadgroup_fork_write_unlock(struct task_struct *tsk) {}
2383 #endif
2384
2385 #ifndef __HAVE_THREAD_FUNCTIONS
2386
2387 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2388 #define task_stack_page(task) ((task)->stack)
2389
2390 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2391 {
2392 *task_thread_info(p) = *task_thread_info(org);
2393 task_thread_info(p)->task = p;
2394 }
2395
2396 static inline unsigned long *end_of_stack(struct task_struct *p)
2397 {
2398 return (unsigned long *)(task_thread_info(p) + 1);
2399 }
2400
2401 #endif
2402
2403 static inline int object_is_on_stack(void *obj)
2404 {
2405 void *stack = task_stack_page(current);
2406
2407 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2408 }
2409
2410 extern void thread_info_cache_init(void);
2411
2412 #ifdef CONFIG_DEBUG_STACK_USAGE
2413 static inline unsigned long stack_not_used(struct task_struct *p)
2414 {
2415 unsigned long *n = end_of_stack(p);
2416
2417 do { /* Skip over canary */
2418 n++;
2419 } while (!*n);
2420
2421 return (unsigned long)n - (unsigned long)end_of_stack(p);
2422 }
2423 #endif
2424
2425 /* set thread flags in other task's structures
2426 * - see asm/thread_info.h for TIF_xxxx flags available
2427 */
2428 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2429 {
2430 set_ti_thread_flag(task_thread_info(tsk), flag);
2431 }
2432
2433 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2434 {
2435 clear_ti_thread_flag(task_thread_info(tsk), flag);
2436 }
2437
2438 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2439 {
2440 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2441 }
2442
2443 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2444 {
2445 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2446 }
2447
2448 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2449 {
2450 return test_ti_thread_flag(task_thread_info(tsk), flag);
2451 }
2452
2453 static inline void set_tsk_need_resched(struct task_struct *tsk)
2454 {
2455 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2456 }
2457
2458 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2459 {
2460 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2461 }
2462
2463 static inline int test_tsk_need_resched(struct task_struct *tsk)
2464 {
2465 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2466 }
2467
2468 static inline int restart_syscall(void)
2469 {
2470 set_tsk_thread_flag(current, TIF_SIGPENDING);
2471 return -ERESTARTNOINTR;
2472 }
2473
2474 static inline int signal_pending(struct task_struct *p)
2475 {
2476 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2477 }
2478
2479 static inline int __fatal_signal_pending(struct task_struct *p)
2480 {
2481 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2482 }
2483
2484 static inline int fatal_signal_pending(struct task_struct *p)
2485 {
2486 return signal_pending(p) && __fatal_signal_pending(p);
2487 }
2488
2489 static inline int signal_pending_state(long state, struct task_struct *p)
2490 {
2491 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2492 return 0;
2493 if (!signal_pending(p))
2494 return 0;
2495
2496 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2497 }
2498
2499 static inline int need_resched(void)
2500 {
2501 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2502 }
2503
2504 /*
2505 * cond_resched() and cond_resched_lock(): latency reduction via
2506 * explicit rescheduling in places that are safe. The return
2507 * value indicates whether a reschedule was done in fact.
2508 * cond_resched_lock() will drop the spinlock before scheduling,
2509 * cond_resched_softirq() will enable bhs before scheduling.
2510 */
2511 extern int _cond_resched(void);
2512
2513 #define cond_resched() ({ \
2514 __might_sleep(__FILE__, __LINE__, 0); \
2515 _cond_resched(); \
2516 })
2517
2518 extern int __cond_resched_lock(spinlock_t *lock);
2519
2520 #ifdef CONFIG_PREEMPT
2521 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2522 #else
2523 #define PREEMPT_LOCK_OFFSET 0
2524 #endif
2525
2526 #define cond_resched_lock(lock) ({ \
2527 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2528 __cond_resched_lock(lock); \
2529 })
2530
2531 extern int __cond_resched_softirq(void);
2532
2533 #define cond_resched_softirq() ({ \
2534 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2535 __cond_resched_softirq(); \
2536 })
2537
2538 /*
2539 * Does a critical section need to be broken due to another
2540 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2541 * but a general need for low latency)
2542 */
2543 static inline int spin_needbreak(spinlock_t *lock)
2544 {
2545 #ifdef CONFIG_PREEMPT
2546 return spin_is_contended(lock);
2547 #else
2548 return 0;
2549 #endif
2550 }
2551
2552 /*
2553 * Thread group CPU time accounting.
2554 */
2555 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2556 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2557
2558 static inline void thread_group_cputime_init(struct signal_struct *sig)
2559 {
2560 spin_lock_init(&sig->cputimer.lock);
2561 }
2562
2563 /*
2564 * Reevaluate whether the task has signals pending delivery.
2565 * Wake the task if so.
2566 * This is required every time the blocked sigset_t changes.
2567 * callers must hold sighand->siglock.
2568 */
2569 extern void recalc_sigpending_and_wake(struct task_struct *t);
2570 extern void recalc_sigpending(void);
2571
2572 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2573
2574 /*
2575 * Wrappers for p->thread_info->cpu access. No-op on UP.
2576 */
2577 #ifdef CONFIG_SMP
2578
2579 static inline unsigned int task_cpu(const struct task_struct *p)
2580 {
2581 return task_thread_info(p)->cpu;
2582 }
2583
2584 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2585
2586 #else
2587
2588 static inline unsigned int task_cpu(const struct task_struct *p)
2589 {
2590 return 0;
2591 }
2592
2593 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2594 {
2595 }
2596
2597 #endif /* CONFIG_SMP */
2598
2599 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2600 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2601
2602 extern void normalize_rt_tasks(void);
2603
2604 #ifdef CONFIG_CGROUP_SCHED
2605
2606 extern struct task_group root_task_group;
2607
2608 extern struct task_group *sched_create_group(struct task_group *parent);
2609 extern void sched_destroy_group(struct task_group *tg);
2610 extern void sched_move_task(struct task_struct *tsk);
2611 #ifdef CONFIG_FAIR_GROUP_SCHED
2612 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2613 extern unsigned long sched_group_shares(struct task_group *tg);
2614 #endif
2615 #ifdef CONFIG_RT_GROUP_SCHED
2616 extern int sched_group_set_rt_runtime(struct task_group *tg,
2617 long rt_runtime_us);
2618 extern long sched_group_rt_runtime(struct task_group *tg);
2619 extern int sched_group_set_rt_period(struct task_group *tg,
2620 long rt_period_us);
2621 extern long sched_group_rt_period(struct task_group *tg);
2622 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2623 #endif
2624 #endif
2625
2626 extern int task_can_switch_user(struct user_struct *up,
2627 struct task_struct *tsk);
2628
2629 #ifdef CONFIG_TASK_XACCT
2630 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2631 {
2632 tsk->ioac.rchar += amt;
2633 }
2634
2635 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2636 {
2637 tsk->ioac.wchar += amt;
2638 }
2639
2640 static inline void inc_syscr(struct task_struct *tsk)
2641 {
2642 tsk->ioac.syscr++;
2643 }
2644
2645 static inline void inc_syscw(struct task_struct *tsk)
2646 {
2647 tsk->ioac.syscw++;
2648 }
2649 #else
2650 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2651 {
2652 }
2653
2654 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2655 {
2656 }
2657
2658 static inline void inc_syscr(struct task_struct *tsk)
2659 {
2660 }
2661
2662 static inline void inc_syscw(struct task_struct *tsk)
2663 {
2664 }
2665 #endif
2666
2667 #ifndef TASK_SIZE_OF
2668 #define TASK_SIZE_OF(tsk) TASK_SIZE
2669 #endif
2670
2671 #ifdef CONFIG_MM_OWNER
2672 extern void mm_update_next_owner(struct mm_struct *mm);
2673 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2674 #else
2675 static inline void mm_update_next_owner(struct mm_struct *mm)
2676 {
2677 }
2678
2679 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2680 {
2681 }
2682 #endif /* CONFIG_MM_OWNER */
2683
2684 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2685 unsigned int limit)
2686 {
2687 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2688 }
2689
2690 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2691 unsigned int limit)
2692 {
2693 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2694 }
2695
2696 static inline unsigned long rlimit(unsigned int limit)
2697 {
2698 return task_rlimit(current, limit);
2699 }
2700
2701 static inline unsigned long rlimit_max(unsigned int limit)
2702 {
2703 return task_rlimit_max(current, limit);
2704 }
2705
2706 #endif /* __KERNEL__ */
2707
2708 #endif