PageRenderTime 74ms CodeModel.GetById 31ms RepoModel.GetById 0ms app.codeStats 1ms

/include/linux/sched.h

https://github.com/airy09/android_kernel_sony_apq8064
C Header | 2848 lines | 1843 code | 433 blank | 572 comment | 51 complexity | 50ea772be0b9726b241677024a712a43 MD5 | raw file
Possible License(s): GPL-2.0
  1. #ifndef _LINUX_SCHED_H
  2. #define _LINUX_SCHED_H
  3. /*
  4. * cloning flags:
  5. */
  6. #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
  7. #define CLONE_VM 0x00000100 /* set if VM shared between processes */
  8. #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
  9. #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
  10. #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
  11. #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
  12. #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
  13. #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
  14. #define CLONE_THREAD 0x00010000 /* Same thread group? */
  15. #define CLONE_NEWNS 0x00020000 /* New namespace group? */
  16. #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
  17. #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
  18. #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
  19. #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
  20. #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
  21. #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
  22. #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
  23. /* 0x02000000 was previously the unused CLONE_STOPPED (Start in stopped state)
  24. and is now available for re-use. */
  25. #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
  26. #define CLONE_NEWIPC 0x08000000 /* New ipcs */
  27. #define CLONE_NEWUSER 0x10000000 /* New user namespace */
  28. #define CLONE_NEWPID 0x20000000 /* New pid namespace */
  29. #define CLONE_NEWNET 0x40000000 /* New network namespace */
  30. #define CLONE_IO 0x80000000 /* Clone io context */
  31. /*
  32. * Scheduling policies
  33. */
  34. #define SCHED_NORMAL 0
  35. #define SCHED_FIFO 1
  36. #define SCHED_RR 2
  37. #define SCHED_BATCH 3
  38. /* SCHED_ISO: reserved but not implemented yet */
  39. #define SCHED_IDLE 5
  40. /* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
  41. #define SCHED_RESET_ON_FORK 0x40000000
  42. #ifdef __KERNEL__
  43. struct sched_param {
  44. int sched_priority;
  45. };
  46. #include <asm/param.h> /* for HZ */
  47. #include <linux/capability.h>
  48. #include <linux/threads.h>
  49. #include <linux/kernel.h>
  50. #include <linux/types.h>
  51. #include <linux/timex.h>
  52. #include <linux/jiffies.h>
  53. #include <linux/rbtree.h>
  54. #include <linux/thread_info.h>
  55. #include <linux/cpumask.h>
  56. #include <linux/errno.h>
  57. #include <linux/nodemask.h>
  58. #include <linux/mm_types.h>
  59. #include <asm/page.h>
  60. #include <asm/ptrace.h>
  61. #include <asm/cputime.h>
  62. #include <linux/smp.h>
  63. #include <linux/sem.h>
  64. #include <linux/signal.h>
  65. #include <linux/compiler.h>
  66. #include <linux/completion.h>
  67. #include <linux/pid.h>
  68. #include <linux/percpu.h>
  69. #include <linux/topology.h>
  70. #include <linux/proportions.h>
  71. #include <linux/seccomp.h>
  72. #include <linux/rcupdate.h>
  73. #include <linux/rculist.h>
  74. #include <linux/rtmutex.h>
  75. #include <linux/time.h>
  76. #include <linux/param.h>
  77. #include <linux/resource.h>
  78. #include <linux/timer.h>
  79. #include <linux/hrtimer.h>
  80. #include <linux/task_io_accounting.h>
  81. #include <linux/latencytop.h>
  82. #include <linux/cred.h>
  83. #include <linux/llist.h>
  84. #include <asm/processor.h>
  85. struct exec_domain;
  86. struct futex_pi_state;
  87. struct robust_list_head;
  88. struct bio_list;
  89. struct fs_struct;
  90. struct perf_event_context;
  91. struct blk_plug;
  92. /*
  93. * List of flags we want to share for kernel threads,
  94. * if only because they are not used by them anyway.
  95. */
  96. #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
  97. /*
  98. * These are the constant used to fake the fixed-point load-average
  99. * counting. Some notes:
  100. * - 11 bit fractions expand to 22 bits by the multiplies: this gives
  101. * a load-average precision of 10 bits integer + 11 bits fractional
  102. * - if you want to count load-averages more often, you need more
  103. * precision, or rounding will get you. With 2-second counting freq,
  104. * the EXP_n values would be 1981, 2034 and 2043 if still using only
  105. * 11 bit fractions.
  106. */
  107. extern unsigned long avenrun[]; /* Load averages */
  108. extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
  109. #define FSHIFT 11 /* nr of bits of precision */
  110. #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
  111. #define LOAD_FREQ (4*HZ+61) /* 4.61 sec intervals */
  112. #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
  113. #define EXP_5 2014 /* 1/exp(5sec/5min) */
  114. #define EXP_15 2037 /* 1/exp(5sec/15min) */
  115. #define CALC_LOAD(load,exp,n) \
  116. load *= exp; \
  117. load += n*(FIXED_1-exp); \
  118. load >>= FSHIFT;
  119. extern unsigned long total_forks;
  120. extern int nr_threads;
  121. DECLARE_PER_CPU(unsigned long, process_counts);
  122. extern int nr_processes(void);
  123. extern unsigned long nr_running(void);
  124. extern unsigned long nr_uninterruptible(void);
  125. extern unsigned long nr_iowait(void);
  126. extern unsigned long avg_nr_running(void);
  127. extern unsigned long nr_iowait_cpu(int cpu);
  128. extern unsigned long this_cpu_load(void);
  129. extern void sched_update_nr_prod(int cpu, unsigned long nr, bool inc);
  130. extern void sched_get_nr_running_avg(int *avg, int *iowait_avg);
  131. extern void calc_global_load(unsigned long ticks);
  132. extern unsigned long get_parent_ip(unsigned long addr);
  133. struct seq_file;
  134. struct cfs_rq;
  135. struct task_group;
  136. #ifdef CONFIG_SCHED_DEBUG
  137. extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
  138. extern void proc_sched_set_task(struct task_struct *p);
  139. extern void
  140. print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
  141. #else
  142. static inline void
  143. proc_sched_show_task(struct task_struct *p, struct seq_file *m)
  144. {
  145. }
  146. static inline void proc_sched_set_task(struct task_struct *p)
  147. {
  148. }
  149. static inline void
  150. print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
  151. {
  152. }
  153. #endif
  154. /*
  155. * Task state bitmask. NOTE! These bits are also
  156. * encoded in fs/proc/array.c: get_task_state().
  157. *
  158. * We have two separate sets of flags: task->state
  159. * is about runnability, while task->exit_state are
  160. * about the task exiting. Confusing, but this way
  161. * modifying one set can't modify the other one by
  162. * mistake.
  163. */
  164. #define TASK_RUNNING 0
  165. #define TASK_INTERRUPTIBLE 1
  166. #define TASK_UNINTERRUPTIBLE 2
  167. #define __TASK_STOPPED 4
  168. #define __TASK_TRACED 8
  169. /* in tsk->exit_state */
  170. #define EXIT_ZOMBIE 16
  171. #define EXIT_DEAD 32
  172. /* in tsk->state again */
  173. #define TASK_DEAD 64
  174. #define TASK_WAKEKILL 128
  175. #define TASK_WAKING 256
  176. #define TASK_STATE_MAX 512
  177. #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
  178. extern char ___assert_task_state[1 - 2*!!(
  179. sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
  180. /* Convenience macros for the sake of set_task_state */
  181. #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
  182. #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
  183. #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
  184. /* Convenience macros for the sake of wake_up */
  185. #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
  186. #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
  187. /* get_task_state() */
  188. #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
  189. TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
  190. __TASK_TRACED)
  191. #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
  192. #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
  193. #define task_is_dead(task) ((task)->exit_state != 0)
  194. #define task_is_stopped_or_traced(task) \
  195. ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
  196. #define task_contributes_to_load(task) \
  197. ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
  198. (task->flags & PF_FROZEN) == 0)
  199. #define __set_task_state(tsk, state_value) \
  200. do { (tsk)->state = (state_value); } while (0)
  201. #define set_task_state(tsk, state_value) \
  202. set_mb((tsk)->state, (state_value))
  203. /*
  204. * set_current_state() includes a barrier so that the write of current->state
  205. * is correctly serialised wrt the caller's subsequent test of whether to
  206. * actually sleep:
  207. *
  208. * set_current_state(TASK_UNINTERRUPTIBLE);
  209. * if (do_i_need_to_sleep())
  210. * schedule();
  211. *
  212. * If the caller does not need such serialisation then use __set_current_state()
  213. */
  214. #define __set_current_state(state_value) \
  215. do { current->state = (state_value); } while (0)
  216. #define set_current_state(state_value) \
  217. set_mb(current->state, (state_value))
  218. /* Task command name length */
  219. #define TASK_COMM_LEN 16
  220. #include <linux/spinlock.h>
  221. /*
  222. * This serializes "schedule()" and also protects
  223. * the run-queue from deletions/modifications (but
  224. * _adding_ to the beginning of the run-queue has
  225. * a separate lock).
  226. */
  227. extern rwlock_t tasklist_lock;
  228. extern spinlock_t mmlist_lock;
  229. struct task_struct;
  230. #ifdef CONFIG_PROVE_RCU
  231. extern int lockdep_tasklist_lock_is_held(void);
  232. #endif /* #ifdef CONFIG_PROVE_RCU */
  233. extern void sched_init(void);
  234. extern void sched_init_smp(void);
  235. extern asmlinkage void schedule_tail(struct task_struct *prev);
  236. extern void init_idle(struct task_struct *idle, int cpu);
  237. extern void init_idle_bootup_task(struct task_struct *idle);
  238. extern int runqueue_is_locked(int cpu);
  239. #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
  240. extern void select_nohz_load_balancer(int stop_tick);
  241. extern void set_cpu_sd_state_idle(void);
  242. extern int get_nohz_timer_target(void);
  243. #else
  244. static inline void select_nohz_load_balancer(int stop_tick) { }
  245. static inline void set_cpu_sd_state_idle(void) { }
  246. #endif
  247. /*
  248. * Only dump TASK_* tasks. (0 for all tasks)
  249. */
  250. extern void show_state_filter(unsigned long state_filter);
  251. static inline void show_state(void)
  252. {
  253. show_state_filter(0);
  254. }
  255. extern void show_regs(struct pt_regs *);
  256. /*
  257. * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
  258. * task), SP is the stack pointer of the first frame that should be shown in the back
  259. * trace (or NULL if the entire call-chain of the task should be shown).
  260. */
  261. extern void show_stack(struct task_struct *task, unsigned long *sp);
  262. void io_schedule(void);
  263. long io_schedule_timeout(long timeout);
  264. extern void cpu_init (void);
  265. extern void trap_init(void);
  266. extern void update_process_times(int user);
  267. extern void scheduler_tick(void);
  268. extern void sched_show_task(struct task_struct *p);
  269. #ifdef CONFIG_LOCKUP_DETECTOR
  270. extern void touch_softlockup_watchdog(void);
  271. extern void touch_softlockup_watchdog_sync(void);
  272. extern void touch_all_softlockup_watchdogs(void);
  273. extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
  274. void __user *buffer,
  275. size_t *lenp, loff_t *ppos);
  276. extern unsigned int softlockup_panic;
  277. void lockup_detector_init(void);
  278. #else
  279. static inline void touch_softlockup_watchdog(void)
  280. {
  281. }
  282. static inline void touch_softlockup_watchdog_sync(void)
  283. {
  284. }
  285. static inline void touch_all_softlockup_watchdogs(void)
  286. {
  287. }
  288. static inline void lockup_detector_init(void)
  289. {
  290. }
  291. #endif
  292. #ifdef CONFIG_DETECT_HUNG_TASK
  293. extern unsigned int sysctl_hung_task_panic;
  294. extern unsigned long sysctl_hung_task_check_count;
  295. extern unsigned long sysctl_hung_task_timeout_secs;
  296. extern unsigned long sysctl_hung_task_warnings;
  297. extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
  298. void __user *buffer,
  299. size_t *lenp, loff_t *ppos);
  300. #else
  301. /* Avoid need for ifdefs elsewhere in the code */
  302. enum { sysctl_hung_task_timeout_secs = 0 };
  303. #endif
  304. /* Attach to any functions which should be ignored in wchan output. */
  305. #define __sched __attribute__((__section__(".sched.text")))
  306. /* Linker adds these: start and end of __sched functions */
  307. extern char __sched_text_start[], __sched_text_end[];
  308. /* Is this address in the __sched functions? */
  309. extern int in_sched_functions(unsigned long addr);
  310. #define MAX_SCHEDULE_TIMEOUT LONG_MAX
  311. extern signed long schedule_timeout(signed long timeout);
  312. extern signed long schedule_timeout_interruptible(signed long timeout);
  313. extern signed long schedule_timeout_killable(signed long timeout);
  314. extern signed long schedule_timeout_uninterruptible(signed long timeout);
  315. asmlinkage void schedule(void);
  316. extern void schedule_preempt_disabled(void);
  317. extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
  318. struct nsproxy;
  319. struct user_namespace;
  320. /*
  321. * Default maximum number of active map areas, this limits the number of vmas
  322. * per mm struct. Users can overwrite this number by sysctl but there is a
  323. * problem.
  324. *
  325. * When a program's coredump is generated as ELF format, a section is created
  326. * per a vma. In ELF, the number of sections is represented in unsigned short.
  327. * This means the number of sections should be smaller than 65535 at coredump.
  328. * Because the kernel adds some informative sections to a image of program at
  329. * generating coredump, we need some margin. The number of extra sections is
  330. * 1-3 now and depends on arch. We use "5" as safe margin, here.
  331. */
  332. #define MAPCOUNT_ELF_CORE_MARGIN (5)
  333. #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
  334. extern int sysctl_max_map_count;
  335. #include <linux/aio.h>
  336. #ifdef CONFIG_MMU
  337. extern void arch_pick_mmap_layout(struct mm_struct *mm);
  338. extern unsigned long
  339. arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
  340. unsigned long, unsigned long);
  341. extern unsigned long
  342. arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
  343. unsigned long len, unsigned long pgoff,
  344. unsigned long flags);
  345. extern void arch_unmap_area(struct mm_struct *, unsigned long);
  346. extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
  347. #else
  348. static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
  349. #endif
  350. extern void set_dumpable(struct mm_struct *mm, int value);
  351. extern int get_dumpable(struct mm_struct *mm);
  352. /* mm flags */
  353. /* dumpable bits */
  354. #define MMF_DUMPABLE 0 /* core dump is permitted */
  355. #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
  356. #define MMF_DUMPABLE_BITS 2
  357. #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
  358. /* coredump filter bits */
  359. #define MMF_DUMP_ANON_PRIVATE 2
  360. #define MMF_DUMP_ANON_SHARED 3
  361. #define MMF_DUMP_MAPPED_PRIVATE 4
  362. #define MMF_DUMP_MAPPED_SHARED 5
  363. #define MMF_DUMP_ELF_HEADERS 6
  364. #define MMF_DUMP_HUGETLB_PRIVATE 7
  365. #define MMF_DUMP_HUGETLB_SHARED 8
  366. #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
  367. #define MMF_DUMP_FILTER_BITS 7
  368. #define MMF_DUMP_FILTER_MASK \
  369. (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
  370. #define MMF_DUMP_FILTER_DEFAULT \
  371. ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
  372. (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
  373. #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
  374. # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
  375. #else
  376. # define MMF_DUMP_MASK_DEFAULT_ELF 0
  377. #endif
  378. /* leave room for more dump flags */
  379. #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
  380. #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
  381. #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
  382. struct sighand_struct {
  383. atomic_t count;
  384. struct k_sigaction action[_NSIG];
  385. spinlock_t siglock;
  386. wait_queue_head_t signalfd_wqh;
  387. };
  388. struct pacct_struct {
  389. int ac_flag;
  390. long ac_exitcode;
  391. unsigned long ac_mem;
  392. cputime_t ac_utime, ac_stime;
  393. unsigned long ac_minflt, ac_majflt;
  394. };
  395. struct cpu_itimer {
  396. cputime_t expires;
  397. cputime_t incr;
  398. u32 error;
  399. u32 incr_error;
  400. };
  401. /**
  402. * struct task_cputime - collected CPU time counts
  403. * @utime: time spent in user mode, in &cputime_t units
  404. * @stime: time spent in kernel mode, in &cputime_t units
  405. * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
  406. *
  407. * This structure groups together three kinds of CPU time that are
  408. * tracked for threads and thread groups. Most things considering
  409. * CPU time want to group these counts together and treat all three
  410. * of them in parallel.
  411. */
  412. struct task_cputime {
  413. cputime_t utime;
  414. cputime_t stime;
  415. unsigned long long sum_exec_runtime;
  416. };
  417. /* Alternate field names when used to cache expirations. */
  418. #define prof_exp stime
  419. #define virt_exp utime
  420. #define sched_exp sum_exec_runtime
  421. #define INIT_CPUTIME \
  422. (struct task_cputime) { \
  423. .utime = 0, \
  424. .stime = 0, \
  425. .sum_exec_runtime = 0, \
  426. }
  427. /*
  428. * Disable preemption until the scheduler is running.
  429. * Reset by start_kernel()->sched_init()->init_idle().
  430. *
  431. * We include PREEMPT_ACTIVE to avoid cond_resched() from working
  432. * before the scheduler is active -- see should_resched().
  433. */
  434. #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
  435. /**
  436. * struct thread_group_cputimer - thread group interval timer counts
  437. * @cputime: thread group interval timers.
  438. * @running: non-zero when there are timers running and
  439. * @cputime receives updates.
  440. * @lock: lock for fields in this struct.
  441. *
  442. * This structure contains the version of task_cputime, above, that is
  443. * used for thread group CPU timer calculations.
  444. */
  445. struct thread_group_cputimer {
  446. struct task_cputime cputime;
  447. int running;
  448. raw_spinlock_t lock;
  449. };
  450. #include <linux/rwsem.h>
  451. struct autogroup;
  452. /*
  453. * NOTE! "signal_struct" does not have its own
  454. * locking, because a shared signal_struct always
  455. * implies a shared sighand_struct, so locking
  456. * sighand_struct is always a proper superset of
  457. * the locking of signal_struct.
  458. */
  459. struct signal_struct {
  460. atomic_t sigcnt;
  461. atomic_t live;
  462. int nr_threads;
  463. wait_queue_head_t wait_chldexit; /* for wait4() */
  464. /* current thread group signal load-balancing target: */
  465. struct task_struct *curr_target;
  466. /* shared signal handling: */
  467. struct sigpending shared_pending;
  468. /* thread group exit support */
  469. int group_exit_code;
  470. /* overloaded:
  471. * - notify group_exit_task when ->count is equal to notify_count
  472. * - everyone except group_exit_task is stopped during signal delivery
  473. * of fatal signals, group_exit_task processes the signal.
  474. */
  475. int notify_count;
  476. struct task_struct *group_exit_task;
  477. /* thread group stop support, overloads group_exit_code too */
  478. int group_stop_count;
  479. unsigned int flags; /* see SIGNAL_* flags below */
  480. /*
  481. * PR_SET_CHILD_SUBREAPER marks a process, like a service
  482. * manager, to re-parent orphan (double-forking) child processes
  483. * to this process instead of 'init'. The service manager is
  484. * able to receive SIGCHLD signals and is able to investigate
  485. * the process until it calls wait(). All children of this
  486. * process will inherit a flag if they should look for a
  487. * child_subreaper process at exit.
  488. */
  489. unsigned int is_child_subreaper:1;
  490. unsigned int has_child_subreaper:1;
  491. /* POSIX.1b Interval Timers */
  492. struct list_head posix_timers;
  493. /* ITIMER_REAL timer for the process */
  494. struct hrtimer real_timer;
  495. struct pid *leader_pid;
  496. ktime_t it_real_incr;
  497. /*
  498. * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
  499. * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
  500. * values are defined to 0 and 1 respectively
  501. */
  502. struct cpu_itimer it[2];
  503. /*
  504. * Thread group totals for process CPU timers.
  505. * See thread_group_cputimer(), et al, for details.
  506. */
  507. struct thread_group_cputimer cputimer;
  508. /* Earliest-expiration cache. */
  509. struct task_cputime cputime_expires;
  510. struct list_head cpu_timers[3];
  511. struct pid *tty_old_pgrp;
  512. /* boolean value for session group leader */
  513. int leader;
  514. struct tty_struct *tty; /* NULL if no tty */
  515. #ifdef CONFIG_SCHED_AUTOGROUP
  516. struct autogroup *autogroup;
  517. #endif
  518. /*
  519. * Cumulative resource counters for dead threads in the group,
  520. * and for reaped dead child processes forked by this group.
  521. * Live threads maintain their own counters and add to these
  522. * in __exit_signal, except for the group leader.
  523. */
  524. cputime_t utime, stime, cutime, cstime;
  525. cputime_t gtime;
  526. cputime_t cgtime;
  527. #ifndef CONFIG_VIRT_CPU_ACCOUNTING
  528. cputime_t prev_utime, prev_stime;
  529. #endif
  530. unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
  531. unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
  532. unsigned long inblock, oublock, cinblock, coublock;
  533. unsigned long maxrss, cmaxrss;
  534. struct task_io_accounting ioac;
  535. /*
  536. * Cumulative ns of schedule CPU time fo dead threads in the
  537. * group, not including a zombie group leader, (This only differs
  538. * from jiffies_to_ns(utime + stime) if sched_clock uses something
  539. * other than jiffies.)
  540. */
  541. unsigned long long sum_sched_runtime;
  542. /*
  543. * We don't bother to synchronize most readers of this at all,
  544. * because there is no reader checking a limit that actually needs
  545. * to get both rlim_cur and rlim_max atomically, and either one
  546. * alone is a single word that can safely be read normally.
  547. * getrlimit/setrlimit use task_lock(current->group_leader) to
  548. * protect this instead of the siglock, because they really
  549. * have no need to disable irqs.
  550. */
  551. struct rlimit rlim[RLIM_NLIMITS];
  552. #ifdef CONFIG_BSD_PROCESS_ACCT
  553. struct pacct_struct pacct; /* per-process accounting information */
  554. #endif
  555. #ifdef CONFIG_TASKSTATS
  556. struct taskstats *stats;
  557. #endif
  558. #ifdef CONFIG_AUDIT
  559. unsigned audit_tty;
  560. struct tty_audit_buf *tty_audit_buf;
  561. #endif
  562. #ifdef CONFIG_CGROUPS
  563. /*
  564. * group_rwsem prevents new tasks from entering the threadgroup and
  565. * member tasks from exiting,a more specifically, setting of
  566. * PF_EXITING. fork and exit paths are protected with this rwsem
  567. * using threadgroup_change_begin/end(). Users which require
  568. * threadgroup to remain stable should use threadgroup_[un]lock()
  569. * which also takes care of exec path. Currently, cgroup is the
  570. * only user.
  571. */
  572. struct rw_semaphore group_rwsem;
  573. #endif
  574. int oom_adj; /* OOM kill score adjustment (bit shift) */
  575. int oom_score_adj; /* OOM kill score adjustment */
  576. int oom_score_adj_min; /* OOM kill score adjustment minimum value.
  577. * Only settable by CAP_SYS_RESOURCE. */
  578. struct mutex cred_guard_mutex; /* guard against foreign influences on
  579. * credential calculations
  580. * (notably. ptrace) */
  581. };
  582. /* Context switch must be unlocked if interrupts are to be enabled */
  583. #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
  584. # define __ARCH_WANT_UNLOCKED_CTXSW
  585. #endif
  586. /*
  587. * Bits in flags field of signal_struct.
  588. */
  589. #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
  590. #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
  591. #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
  592. /*
  593. * Pending notifications to parent.
  594. */
  595. #define SIGNAL_CLD_STOPPED 0x00000010
  596. #define SIGNAL_CLD_CONTINUED 0x00000020
  597. #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
  598. #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
  599. /* If true, all threads except ->group_exit_task have pending SIGKILL */
  600. static inline int signal_group_exit(const struct signal_struct *sig)
  601. {
  602. return (sig->flags & SIGNAL_GROUP_EXIT) ||
  603. (sig->group_exit_task != NULL);
  604. }
  605. /*
  606. * Some day this will be a full-fledged user tracking system..
  607. */
  608. struct user_struct {
  609. atomic_t __count; /* reference count */
  610. atomic_t processes; /* How many processes does this user have? */
  611. atomic_t files; /* How many open files does this user have? */
  612. atomic_t sigpending; /* How many pending signals does this user have? */
  613. #ifdef CONFIG_INOTIFY_USER
  614. atomic_t inotify_watches; /* How many inotify watches does this user have? */
  615. atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
  616. #endif
  617. #ifdef CONFIG_FANOTIFY
  618. atomic_t fanotify_listeners;
  619. #endif
  620. #ifdef CONFIG_EPOLL
  621. atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
  622. #endif
  623. #ifdef CONFIG_POSIX_MQUEUE
  624. /* protected by mq_lock */
  625. unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
  626. #endif
  627. unsigned long locked_shm; /* How many pages of mlocked shm ? */
  628. #ifdef CONFIG_KEYS
  629. struct key *uid_keyring; /* UID specific keyring */
  630. struct key *session_keyring; /* UID's default session keyring */
  631. #endif
  632. /* Hash table maintenance information */
  633. struct hlist_node uidhash_node;
  634. uid_t uid;
  635. struct user_namespace *user_ns;
  636. #ifdef CONFIG_PERF_EVENTS
  637. atomic_long_t locked_vm;
  638. #endif
  639. };
  640. extern int uids_sysfs_init(void);
  641. extern struct user_struct *find_user(uid_t);
  642. extern struct user_struct root_user;
  643. #define INIT_USER (&root_user)
  644. struct backing_dev_info;
  645. struct reclaim_state;
  646. #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
  647. struct sched_info {
  648. /* cumulative counters */
  649. unsigned long pcount; /* # of times run on this cpu */
  650. unsigned long long run_delay; /* time spent waiting on a runqueue */
  651. /* timestamps */
  652. unsigned long long last_arrival,/* when we last ran on a cpu */
  653. last_queued; /* when we were last queued to run */
  654. };
  655. #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
  656. #ifdef CONFIG_TASK_DELAY_ACCT
  657. struct task_delay_info {
  658. spinlock_t lock;
  659. unsigned int flags; /* Private per-task flags */
  660. /* For each stat XXX, add following, aligned appropriately
  661. *
  662. * struct timespec XXX_start, XXX_end;
  663. * u64 XXX_delay;
  664. * u32 XXX_count;
  665. *
  666. * Atomicity of updates to XXX_delay, XXX_count protected by
  667. * single lock above (split into XXX_lock if contention is an issue).
  668. */
  669. /*
  670. * XXX_count is incremented on every XXX operation, the delay
  671. * associated with the operation is added to XXX_delay.
  672. * XXX_delay contains the accumulated delay time in nanoseconds.
  673. */
  674. struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
  675. u64 blkio_delay; /* wait for sync block io completion */
  676. u64 swapin_delay; /* wait for swapin block io completion */
  677. u32 blkio_count; /* total count of the number of sync block */
  678. /* io operations performed */
  679. u32 swapin_count; /* total count of the number of swapin block */
  680. /* io operations performed */
  681. struct timespec freepages_start, freepages_end;
  682. u64 freepages_delay; /* wait for memory reclaim */
  683. u32 freepages_count; /* total count of memory reclaim */
  684. };
  685. #endif /* CONFIG_TASK_DELAY_ACCT */
  686. static inline int sched_info_on(void)
  687. {
  688. #ifdef CONFIG_SCHEDSTATS
  689. return 1;
  690. #elif defined(CONFIG_TASK_DELAY_ACCT)
  691. extern int delayacct_on;
  692. return delayacct_on;
  693. #else
  694. return 0;
  695. #endif
  696. }
  697. enum cpu_idle_type {
  698. CPU_IDLE,
  699. CPU_NOT_IDLE,
  700. CPU_NEWLY_IDLE,
  701. CPU_MAX_IDLE_TYPES
  702. };
  703. /*
  704. * Increase resolution of nice-level calculations for 64-bit architectures.
  705. * The extra resolution improves shares distribution and load balancing of
  706. * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
  707. * hierarchies, especially on larger systems. This is not a user-visible change
  708. * and does not change the user-interface for setting shares/weights.
  709. *
  710. * We increase resolution only if we have enough bits to allow this increased
  711. * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
  712. * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
  713. * increased costs.
  714. */
  715. #if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load */
  716. # define SCHED_LOAD_RESOLUTION 10
  717. # define scale_load(w) ((w) << SCHED_LOAD_RESOLUTION)
  718. # define scale_load_down(w) ((w) >> SCHED_LOAD_RESOLUTION)
  719. #else
  720. # define SCHED_LOAD_RESOLUTION 0
  721. # define scale_load(w) (w)
  722. # define scale_load_down(w) (w)
  723. #endif
  724. #define SCHED_LOAD_SHIFT (10 + SCHED_LOAD_RESOLUTION)
  725. #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
  726. /*
  727. * Increase resolution of cpu_power calculations
  728. */
  729. #define SCHED_POWER_SHIFT 10
  730. #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
  731. /*
  732. * sched-domains (multiprocessor balancing) declarations:
  733. */
  734. #ifdef CONFIG_SMP
  735. #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
  736. #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
  737. #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
  738. #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
  739. #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
  740. #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
  741. #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
  742. #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
  743. #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
  744. #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
  745. #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
  746. #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
  747. #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
  748. #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
  749. enum powersavings_balance_level {
  750. POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
  751. POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
  752. * first for long running threads
  753. */
  754. POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
  755. * cpu package for power savings
  756. */
  757. MAX_POWERSAVINGS_BALANCE_LEVELS
  758. };
  759. extern int sched_mc_power_savings, sched_smt_power_savings;
  760. static inline int sd_balance_for_mc_power(void)
  761. {
  762. if (sched_smt_power_savings)
  763. return SD_POWERSAVINGS_BALANCE;
  764. if (!sched_mc_power_savings)
  765. return SD_PREFER_SIBLING;
  766. return 0;
  767. }
  768. static inline int sd_balance_for_package_power(void)
  769. {
  770. if (sched_mc_power_savings | sched_smt_power_savings)
  771. return SD_POWERSAVINGS_BALANCE;
  772. return SD_PREFER_SIBLING;
  773. }
  774. extern int __weak arch_sd_sibiling_asym_packing(void);
  775. /*
  776. * Optimise SD flags for power savings:
  777. * SD_BALANCE_NEWIDLE helps aggressive task consolidation and power savings.
  778. * Keep default SD flags if sched_{smt,mc}_power_saving=0
  779. */
  780. static inline int sd_power_saving_flags(void)
  781. {
  782. if (sched_mc_power_savings | sched_smt_power_savings)
  783. return SD_BALANCE_NEWIDLE;
  784. return 0;
  785. }
  786. struct sched_group_power {
  787. atomic_t ref;
  788. /*
  789. * CPU power of this group, SCHED_LOAD_SCALE being max power for a
  790. * single CPU.
  791. */
  792. unsigned int power, power_orig;
  793. unsigned long next_update;
  794. /*
  795. * Number of busy cpus in this group.
  796. */
  797. atomic_t nr_busy_cpus;
  798. };
  799. struct sched_group {
  800. struct sched_group *next; /* Must be a circular list */
  801. atomic_t ref;
  802. unsigned int group_weight;
  803. struct sched_group_power *sgp;
  804. /*
  805. * The CPUs this group covers.
  806. *
  807. * NOTE: this field is variable length. (Allocated dynamically
  808. * by attaching extra space to the end of the structure,
  809. * depending on how many CPUs the kernel has booted up with)
  810. */
  811. unsigned long cpumask[0];
  812. };
  813. static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
  814. {
  815. return to_cpumask(sg->cpumask);
  816. }
  817. /**
  818. * group_first_cpu - Returns the first cpu in the cpumask of a sched_group.
  819. * @group: The group whose first cpu is to be returned.
  820. */
  821. static inline unsigned int group_first_cpu(struct sched_group *group)
  822. {
  823. return cpumask_first(sched_group_cpus(group));
  824. }
  825. struct sched_domain_attr {
  826. int relax_domain_level;
  827. };
  828. #define SD_ATTR_INIT (struct sched_domain_attr) { \
  829. .relax_domain_level = -1, \
  830. }
  831. extern int sched_domain_level_max;
  832. struct sched_domain {
  833. /* These fields must be setup */
  834. struct sched_domain *parent; /* top domain must be null terminated */
  835. struct sched_domain *child; /* bottom domain must be null terminated */
  836. struct sched_group *groups; /* the balancing groups of the domain */
  837. unsigned long min_interval; /* Minimum balance interval ms */
  838. unsigned long max_interval; /* Maximum balance interval ms */
  839. unsigned int busy_factor; /* less balancing by factor if busy */
  840. unsigned int imbalance_pct; /* No balance until over watermark */
  841. unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
  842. unsigned int busy_idx;
  843. unsigned int idle_idx;
  844. unsigned int newidle_idx;
  845. unsigned int wake_idx;
  846. unsigned int forkexec_idx;
  847. unsigned int smt_gain;
  848. int flags; /* See SD_* */
  849. int level;
  850. /* Runtime fields. */
  851. unsigned long last_balance; /* init to jiffies. units in jiffies */
  852. unsigned int balance_interval; /* initialise to 1. units in ms. */
  853. unsigned int nr_balance_failed; /* initialise to 0 */
  854. u64 last_update;
  855. #ifdef CONFIG_SCHEDSTATS
  856. /* load_balance() stats */
  857. unsigned int lb_count[CPU_MAX_IDLE_TYPES];
  858. unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
  859. unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
  860. unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
  861. unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
  862. unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
  863. unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
  864. unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
  865. /* Active load balancing */
  866. unsigned int alb_count;
  867. unsigned int alb_failed;
  868. unsigned int alb_pushed;
  869. /* SD_BALANCE_EXEC stats */
  870. unsigned int sbe_count;
  871. unsigned int sbe_balanced;
  872. unsigned int sbe_pushed;
  873. /* SD_BALANCE_FORK stats */
  874. unsigned int sbf_count;
  875. unsigned int sbf_balanced;
  876. unsigned int sbf_pushed;
  877. /* try_to_wake_up() stats */
  878. unsigned int ttwu_wake_remote;
  879. unsigned int ttwu_move_affine;
  880. unsigned int ttwu_move_balance;
  881. #endif
  882. #ifdef CONFIG_SCHED_DEBUG
  883. char *name;
  884. #endif
  885. union {
  886. void *private; /* used during construction */
  887. struct rcu_head rcu; /* used during destruction */
  888. };
  889. unsigned int span_weight;
  890. /*
  891. * Span of all CPUs in this domain.
  892. *
  893. * NOTE: this field is variable length. (Allocated dynamically
  894. * by attaching extra space to the end of the structure,
  895. * depending on how many CPUs the kernel has booted up with)
  896. */
  897. unsigned long span[0];
  898. };
  899. static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
  900. {
  901. return to_cpumask(sd->span);
  902. }
  903. extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
  904. struct sched_domain_attr *dattr_new);
  905. /* Allocate an array of sched domains, for partition_sched_domains(). */
  906. cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
  907. void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
  908. /* Test a flag in parent sched domain */
  909. static inline int test_sd_parent(struct sched_domain *sd, int flag)
  910. {
  911. if (sd->parent && (sd->parent->flags & flag))
  912. return 1;
  913. return 0;
  914. }
  915. unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
  916. unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
  917. bool cpus_share_cache(int this_cpu, int that_cpu);
  918. #else /* CONFIG_SMP */
  919. struct sched_domain_attr;
  920. static inline void
  921. partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
  922. struct sched_domain_attr *dattr_new)
  923. {
  924. }
  925. static inline bool cpus_share_cache(int this_cpu, int that_cpu)
  926. {
  927. return true;
  928. }
  929. #endif /* !CONFIG_SMP */
  930. struct io_context; /* See blkdev.h */
  931. #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
  932. extern void prefetch_stack(struct task_struct *t);
  933. #else
  934. static inline void prefetch_stack(struct task_struct *t) { }
  935. #endif
  936. struct audit_context; /* See audit.c */
  937. struct mempolicy;
  938. struct pipe_inode_info;
  939. struct uts_namespace;
  940. struct rq;
  941. struct sched_domain;
  942. /*
  943. * wake flags
  944. */
  945. #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
  946. #define WF_FORK 0x02 /* child wakeup after fork */
  947. #define WF_MIGRATED 0x04 /* internal use, task got migrated */
  948. #define ENQUEUE_WAKEUP 1
  949. #define ENQUEUE_HEAD 2
  950. #ifdef CONFIG_SMP
  951. #define ENQUEUE_WAKING 4 /* sched_class::task_waking was called */
  952. #else
  953. #define ENQUEUE_WAKING 0
  954. #endif
  955. #define DEQUEUE_SLEEP 1
  956. struct sched_class {
  957. const struct sched_class *next;
  958. void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
  959. void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
  960. void (*yield_task) (struct rq *rq);
  961. bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
  962. void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
  963. struct task_struct * (*pick_next_task) (struct rq *rq);
  964. void (*put_prev_task) (struct rq *rq, struct task_struct *p);
  965. #ifdef CONFIG_SMP
  966. int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
  967. void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
  968. void (*post_schedule) (struct rq *this_rq);
  969. void (*task_waking) (struct task_struct *task);
  970. void (*task_woken) (struct rq *this_rq, struct task_struct *task);
  971. void (*set_cpus_allowed)(struct task_struct *p,
  972. const struct cpumask *newmask);
  973. void (*rq_online)(struct rq *rq);
  974. void (*rq_offline)(struct rq *rq);
  975. #endif
  976. void (*set_curr_task) (struct rq *rq);
  977. void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
  978. void (*task_fork) (struct task_struct *p);
  979. void (*switched_from) (struct rq *this_rq, struct task_struct *task);
  980. void (*switched_to) (struct rq *this_rq, struct task_struct *task);
  981. void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
  982. int oldprio);
  983. unsigned int (*get_rr_interval) (struct rq *rq,
  984. struct task_struct *task);
  985. #ifdef CONFIG_FAIR_GROUP_SCHED
  986. void (*task_move_group) (struct task_struct *p, int on_rq);
  987. #endif
  988. };
  989. struct load_weight {
  990. unsigned long weight, inv_weight;
  991. };
  992. #ifdef CONFIG_SCHEDSTATS
  993. struct sched_statistics {
  994. u64 wait_start;
  995. u64 wait_max;
  996. u64 wait_count;
  997. u64 wait_sum;
  998. u64 iowait_count;
  999. u64 iowait_sum;
  1000. u64 sleep_start;
  1001. u64 sleep_max;
  1002. s64 sum_sleep_runtime;
  1003. u64 block_start;
  1004. u64 block_max;
  1005. u64 exec_max;
  1006. u64 slice_max;
  1007. u64 nr_migrations_cold;
  1008. u64 nr_failed_migrations_affine;
  1009. u64 nr_failed_migrations_running;
  1010. u64 nr_failed_migrations_hot;
  1011. u64 nr_forced_migrations;
  1012. u64 nr_wakeups;
  1013. u64 nr_wakeups_sync;
  1014. u64 nr_wakeups_migrate;
  1015. u64 nr_wakeups_local;
  1016. u64 nr_wakeups_remote;
  1017. u64 nr_wakeups_affine;
  1018. u64 nr_wakeups_affine_attempts;
  1019. u64 nr_wakeups_passive;
  1020. u64 nr_wakeups_idle;
  1021. };
  1022. #endif
  1023. struct sched_entity {
  1024. struct load_weight load; /* for load-balancing */
  1025. struct rb_node run_node;
  1026. struct list_head group_node;
  1027. unsigned int on_rq;
  1028. u64 exec_start;
  1029. u64 sum_exec_runtime;
  1030. u64 vruntime;
  1031. u64 prev_sum_exec_runtime;
  1032. u64 nr_migrations;
  1033. #ifdef CONFIG_SCHEDSTATS
  1034. struct sched_statistics statistics;
  1035. #endif
  1036. #ifdef CONFIG_FAIR_GROUP_SCHED
  1037. struct sched_entity *parent;
  1038. /* rq on which this entity is (to be) queued: */
  1039. struct cfs_rq *cfs_rq;
  1040. /* rq "owned" by this entity/group: */
  1041. struct cfs_rq *my_q;
  1042. #endif
  1043. };
  1044. struct sched_rt_entity {
  1045. struct list_head run_list;
  1046. unsigned long timeout;
  1047. unsigned int time_slice;
  1048. int nr_cpus_allowed;
  1049. struct sched_rt_entity *back;
  1050. #ifdef CONFIG_RT_GROUP_SCHED
  1051. struct sched_rt_entity *parent;
  1052. /* rq on which this entity is (to be) queued: */
  1053. struct rt_rq *rt_rq;
  1054. /* rq "owned" by this entity/group: */
  1055. struct rt_rq *my_q;
  1056. #endif
  1057. };
  1058. /*
  1059. * default timeslice is 100 msecs (used only for SCHED_RR tasks).
  1060. * Timeslices get refilled after they expire.
  1061. */
  1062. #define RR_TIMESLICE (100 * HZ / 1000)
  1063. struct rcu_node;
  1064. enum perf_event_task_context {
  1065. perf_invalid_context = -1,
  1066. perf_hw_context = 0,
  1067. perf_sw_context,
  1068. perf_nr_task_contexts,
  1069. };
  1070. struct task_struct {
  1071. volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
  1072. void *stack;
  1073. atomic_t usage;
  1074. unsigned int flags; /* per process flags, defined below */
  1075. unsigned int ptrace;
  1076. #ifdef CONFIG_SMP
  1077. struct llist_node wake_entry;
  1078. int on_cpu;
  1079. #endif
  1080. int on_rq;
  1081. int prio, static_prio, normal_prio;
  1082. unsigned int rt_priority;
  1083. const struct sched_class *sched_class;
  1084. struct sched_entity se;
  1085. struct sched_rt_entity rt;
  1086. #ifdef CONFIG_PREEMPT_NOTIFIERS
  1087. /* list of struct preempt_notifier: */
  1088. struct hlist_head preempt_notifiers;
  1089. #endif
  1090. /*
  1091. * fpu_counter contains the number of consecutive context switches
  1092. * that the FPU is used. If this is over a threshold, the lazy fpu
  1093. * saving becomes unlazy to save the trap. This is an unsigned char
  1094. * so that after 256 times the counter wraps and the behavior turns
  1095. * lazy again; this to deal with bursty apps that only use FPU for
  1096. * a short time
  1097. */
  1098. unsigned char fpu_counter;
  1099. #ifdef CONFIG_BLK_DEV_IO_TRACE
  1100. unsigned int btrace_seq;
  1101. #endif
  1102. unsigned int policy;
  1103. cpumask_t cpus_allowed;
  1104. #ifdef CONFIG_PREEMPT_RCU
  1105. int rcu_read_lock_nesting;
  1106. char rcu_read_unlock_special;
  1107. struct list_head rcu_node_entry;
  1108. #endif /* #ifdef CONFIG_PREEMPT_RCU */
  1109. #ifdef CONFIG_TREE_PREEMPT_RCU
  1110. struct rcu_node *rcu_blocked_node;
  1111. #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
  1112. #ifdef CONFIG_RCU_BOOST
  1113. struct rt_mutex *rcu_boost_mutex;
  1114. #endif /* #ifdef CONFIG_RCU_BOOST */
  1115. #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
  1116. struct sched_info sched_info;
  1117. #endif
  1118. struct list_head tasks;
  1119. #ifdef CONFIG_SMP
  1120. struct plist_node pushable_tasks;
  1121. #endif
  1122. struct mm_struct *mm, *active_mm;
  1123. #ifdef CONFIG_COMPAT_BRK
  1124. unsigned brk_randomized:1;
  1125. #endif
  1126. #if defined(SPLIT_RSS_COUNTING)
  1127. struct task_rss_stat rss_stat;
  1128. #endif
  1129. /* task state */
  1130. int exit_state;
  1131. int exit_code, exit_signal;
  1132. int pdeath_signal; /* The signal sent when the parent dies */
  1133. unsigned int jobctl; /* JOBCTL_*, siglock protected */
  1134. /* ??? */
  1135. unsigned int personality;
  1136. unsigned did_exec:1;
  1137. unsigned in_execve:1; /* Tell the LSMs that the process is doing an
  1138. * execve */
  1139. unsigned in_iowait:1;
  1140. /* Revert to default priority/policy when forking */
  1141. unsigned sched_reset_on_fork:1;
  1142. unsigned sched_contributes_to_load:1;
  1143. #ifdef CONFIG_GENERIC_HARDIRQS
  1144. /* IRQ handler threads */
  1145. unsigned irq_thread:1;
  1146. #endif
  1147. pid_t pid;
  1148. pid_t tgid;
  1149. #ifdef CONFIG_CC_STACKPROTECTOR
  1150. /* Canary value for the -fstack-protector gcc feature */
  1151. unsigned long stack_canary;
  1152. #endif
  1153. /*
  1154. * pointers to (original) parent process, youngest child, younger sibling,
  1155. * older sibling, respectively. (p->father can be replaced with
  1156. * p->real_parent->pid)
  1157. */
  1158. struct task_struct __rcu *real_parent; /* real parent process */
  1159. struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
  1160. /*
  1161. * children/sibling forms the list of my natural children
  1162. */
  1163. struct list_head children; /* list of my children */
  1164. struct list_head sibling; /* linkage in my parent's children list */
  1165. struct task_struct *group_leader; /* threadgroup leader */
  1166. /*
  1167. * ptraced is the list of tasks this task is using ptrace on.
  1168. * This includes both natural children and PTRACE_ATTACH targets.
  1169. * p->ptrace_entry is p's link on the p->parent->ptraced list.
  1170. */
  1171. struct list_head ptraced;
  1172. struct list_head ptrace_entry;
  1173. /* PID/PID hash table linkage. */
  1174. struct pid_link pids[PIDTYPE_MAX];
  1175. struct list_head thread_group;
  1176. struct completion *vfork_done; /* for vfork() */
  1177. int __user *set_child_tid; /* CLONE_CHILD_SETTID */
  1178. int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
  1179. cputime_t utime, stime, utimescaled, stimescaled;
  1180. cputime_t gtime;
  1181. #ifndef CONFIG_VIRT_CPU_ACCOUNTING
  1182. cputime_t prev_utime, prev_stime;
  1183. #endif
  1184. unsigned long nvcsw, nivcsw; /* context switch counts */
  1185. struct timespec start_time; /* monotonic time */
  1186. struct timespec real_start_time; /* boot based time */
  1187. /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
  1188. unsigned long min_flt, maj_flt;
  1189. struct task_cputime cputime_expires;
  1190. struct list_head cpu_timers[3];
  1191. /* process credentials */
  1192. const struct cred __rcu *real_cred; /* objective and real subjective task
  1193. * credentials (COW) */
  1194. const struct cred __rcu *cred; /* effective (overridable) subjective task
  1195. * credentials (COW) */
  1196. struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
  1197. char comm[TASK_COMM_LEN]; /* executable name excluding path
  1198. - access with [gs]et_task_comm (which lock
  1199. it with task_lock())
  1200. - initialized normally by setup_new_exec */
  1201. /* file system info */
  1202. int link_count, total_link_count;
  1203. #ifdef CONFIG_SYSVIPC
  1204. /* ipc stuff */
  1205. struct sysv_sem sysvsem;
  1206. #endif
  1207. #ifdef CONFIG_DETECT_HUNG_TASK
  1208. /* hung task detection */
  1209. unsigned long last_switch_count;
  1210. #endif
  1211. /* CPU-specific state of this task */
  1212. struct thread_struct thread;
  1213. /* filesystem information */
  1214. struct fs_struct *fs;
  1215. /* open file information */
  1216. struct files_struct *files;
  1217. /* namespaces */
  1218. struct nsproxy *nsproxy;
  1219. /* signal handlers */
  1220. struct signal_struct *signal;
  1221. struct sighand_struct *sighand;
  1222. sigset_t blocked, real_blocked;
  1223. sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
  1224. struct sigpending pending;
  1225. unsigned long sas_ss_sp;
  1226. size_t sas_ss_size;
  1227. int (*notifier)(void *priv);
  1228. void *notifier_data;
  1229. sigset_t *notifier_mask;
  1230. struct audit_context *audit_context;
  1231. #ifdef CONFIG_AUDITSYSCALL
  1232. uid_t loginuid;
  1233. unsigned int sessionid;
  1234. #endif
  1235. seccomp_t seccomp;
  1236. /* Thread group tracking */
  1237. u32 parent_exec_id;
  1238. u32 self_exec_id;
  1239. /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
  1240. * mempolicy */
  1241. spinlock_t alloc_lock;
  1242. /* Protection of the PI data structures: */
  1243. raw_spinlock_t pi_lock;
  1244. #ifdef CONFIG_RT_MUTEXES
  1245. /* PI waiters blocked on a rt_mutex held by this task */
  1246. struct plist_head pi_waiters;
  1247. /* Deadlock detection and priority inheritance handling */
  1248. struct rt_mutex_waiter *pi_blocked_on;
  1249. #endif
  1250. #ifdef CONFIG_DEBUG_MUTEXES
  1251. /* mutex deadlock detection */
  1252. struct mutex_waiter *blocked_on;
  1253. #endif
  1254. #ifdef CONFIG_TRACE_IRQFLAGS
  1255. unsigned int irq_events;
  1256. unsigned long hardirq_enable_ip;
  1257. unsigned long hardirq_disable_ip;
  1258. unsigned int hardirq_enable_event;
  1259. unsigned int hardirq_disable_event;
  1260. int hardirqs_enabled;
  1261. int hardirq_context;
  1262. unsigned long softirq_disable_ip;
  1263. unsigned long softirq_enable_ip;
  1264. unsigned int softirq_disable_event;
  1265. unsigned int softirq_enable_event;
  1266. int softirqs_enabled;
  1267. int softirq_context;
  1268. #endif
  1269. #ifdef CONFIG_LOCKDEP
  1270. # define MAX_LOCK_DEPTH 48UL
  1271. u64 curr_chain_key;
  1272. int lockdep_depth;
  1273. unsigned int lockdep_recursion;
  1274. struct held_lock held_locks[MAX_LOCK_DEPTH];
  1275. gfp_t lockdep_reclaim_gfp;
  1276. #endif
  1277. /* journalling filesystem info */
  1278. void *journal_info;
  1279. /* stacked block device info */
  1280. struct bio_list *bio_list;
  1281. #ifdef CONFIG_BLOCK
  1282. /* stack plugging */
  1283. struct blk_plug *plug;
  1284. #endif
  1285. /* VM state */
  1286. struct reclaim_state *reclaim_state;
  1287. struct backing_dev_info *backing_dev_info;
  1288. struct io_context *io_context;
  1289. unsigned long ptrace_message;
  1290. siginfo_t *last_siginfo; /* For ptrace use. */
  1291. struct task_io_accounting ioac;
  1292. #if defined(CONFIG_TASK_XACCT)
  1293. u64 acct_rss_mem1; /* accumulated rss usage */
  1294. u64 acct_vm_mem1; /* accumulated virtual memory usage */
  1295. cputime_t acct_timexpd; /* stime + utime since last update */
  1296. #endif
  1297. #ifdef CONFIG_CPUSETS
  1298. nodemask_t mems_allowed; /* Protected by alloc_lock */
  1299. seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
  1300. int cpuset_mem_spread_rotor;
  1301. int cpuset_slab_spread_rotor;
  1302. #endif
  1303. #ifdef CONFIG_CGROUPS
  1304. /* Control Group info protected by css_set_lock */
  1305. struct css_set __rcu *cgroups;
  1306. /* cg_list protected by css_set_lock and tsk->alloc_lock */
  1307. struct list_head cg_list;
  1308. #endif
  1309. #ifdef CONFIG_FUTEX
  1310. struct robust_list_head __user *robust_list;
  1311. #ifdef CONFIG_COMPAT
  1312. struct compat_robust_list_head __user *compat_robust_list;
  1313. #endif
  1314. struct list_head pi_state_list;
  1315. struct futex_pi_state *pi_state_cache;
  1316. #endif
  1317. #ifdef CONFIG_PERF_EVENTS
  1318. struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
  1319. struct mutex perf_event_mutex;
  1320. struct list_head perf_event_list;
  1321. #endif
  1322. #ifdef CONFIG_NUMA
  1323. struct mempolicy *mempolicy; /* Protected by alloc_lock */
  1324. short il_next;
  1325. short pref_node_fork;
  1326. #endif
  1327. struct rcu_head rcu;
  1328. /*
  1329. * cache last used pipe for splice
  1330. */
  1331. struct pipe_inode_info *splice_pipe;
  1332. #ifdef CONFIG_TASK_DELAY_ACCT
  1333. struct task_delay_info *delays;
  1334. #endif
  1335. #ifdef CONFIG_FAULT_INJECTION
  1336. int make_it_fail;
  1337. #endif
  1338. /*
  1339. * when (nr_dirtied >= nr_dirtied_pause), it's time to call
  1340. * balance_dirty_pages() for some dirty throttling pause
  1341. */
  1342. int nr_dirtied;
  1343. int nr_dirtied_pause;
  1344. unsigned long dirty_paused_when; /* start of a write-and-pause period */
  1345. #ifdef CONFIG_LATENCYTOP
  1346. int latency_record_count;
  1347. struct latency_record latency_record[LT_SAVECOUNT];
  1348. #endif
  1349. /*
  1350. * time slack values; these are used to round up poll() and
  1351. * select() etc timeout values. These are in nanoseconds.
  1352. */
  1353. unsigned long timer_slack_ns;
  1354. unsigned long default_timer_slack_ns;
  1355. struct list_head *scm_work_list;
  1356. #ifdef CONFIG_FUNCTION_GRAPH_TRACER
  1357. /* Index of current stored address in ret_stack */
  1358. int curr_ret_stack;
  1359. /* Stack of return addresses for return function tracing */
  1360. struct ftrace_ret_stack *ret_stack;
  1361. /* time stamp for last schedule */
  1362. unsigned long long ftrace_timestamp;
  1363. /*
  1364. * Number of functions that haven't been traced
  1365. * because of depth overrun.
  1366. */
  1367. atomic_t trace_overrun;
  1368. /* Pause for the tracing */
  1369. atomic_t tracing_graph_pause;
  1370. #endif
  1371. #ifdef CONFIG_TRACING
  1372. /* state flags for use by tracers */
  1373. unsigned long trace;
  1374. /* bitmask and counter of trace recursion */
  1375. unsigned long trace_recursion;
  1376. #endif /* CONFIG_TRACING */
  1377. #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
  1378. struct memcg_batch_info {
  1379. int do_batch; /* incremented when batch uncharge started */
  1380. struct mem_cgroup *memcg; /* target memcg of uncharge */
  1381. unsigned long nr_pages; /* uncharged usage */
  1382. unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
  1383. } memcg_batch;
  1384. #endif
  1385. #ifdef CONFIG_HAVE_HW_BREAKPOINT
  1386. atomic_t ptrace_bp_refcnt;
  1387. #endif
  1388. };
  1389. /* Future-safe accessor for struct task_struct's cpus_allowed. */
  1390. #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
  1391. /*
  1392. * Priority of a process goes from 0..MAX_PRIO-1, valid RT
  1393. * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
  1394. * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
  1395. * values are inverted: lower p->prio value means higher priority.
  1396. *
  1397. * The MAX_USER_RT_PRIO value allows the actual maximum
  1398. * RT priority to be separate from the value exported to
  1399. * user-space. This allows kernel threads to set their
  1400. * priority to a value higher than any user task. Note:
  1401. * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
  1402. */
  1403. #define MAX_USER_RT_PRIO 100
  1404. #define MAX_RT_PRIO MAX_USER_RT_PRIO
  1405. #define MAX_PRIO (MAX_RT_PRIO + 40)
  1406. #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
  1407. static inline int rt_prio(int prio)
  1408. {
  1409. if (unlikely(prio < MAX_RT_PRIO))
  1410. return 1;
  1411. return 0;
  1412. }
  1413. static inline int rt_task(struct task_struct *p)
  1414. {
  1415. return rt_prio(p->prio);
  1416. }
  1417. static inline struct pid *task_pid(struct task_struct *task)
  1418. {
  1419. return task->pids[PIDTYPE_PID].pid;
  1420. }
  1421. static inline struct pid *task_tgid(struct task_struct *task)
  1422. {
  1423. return task->group_leader->pids[PIDTYPE_PID].pid;
  1424. }
  1425. /*
  1426. * Without tasklist or rcu lock it is not safe to dereference
  1427. * the result of task_pgrp/task_session even if task == current,
  1428. * we can race with another thread doing sys_setsid/sys_setpgid.
  1429. */
  1430. static inline struct pid *task_pgrp(struct task_struct *task)
  1431. {
  1432. return task->group_leader->pids[PIDTYPE_PGID].pid;
  1433. }
  1434. static inline struct pid *task_session(struct task_struct *task)
  1435. {
  1436. return task->group_leader->pids[PIDTYPE_SID].pid;
  1437. }
  1438. struct pid_namespace;
  1439. /*
  1440. * the helpers to get the task's different pids as they are seen
  1441. * from various namespaces
  1442. *
  1443. * task_xid_nr() : global id, i.e. the id seen from the init namespace;
  1444. * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
  1445. * current.
  1446. * task_xid_nr_ns() : id seen from the ns specified;
  1447. *
  1448. * set_task_vxid() : assigns a virtual id to a task;
  1449. *
  1450. * see also pid_nr() etc in include/linux/pid.h
  1451. */
  1452. pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
  1453. struct pid_namespace *ns);
  1454. static inline pid_t task_pid_nr(struct task_struct *tsk)
  1455. {
  1456. return tsk->pid;
  1457. }
  1458. static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
  1459. struct pid_namespace *ns)
  1460. {
  1461. return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
  1462. }
  1463. static inline pid_t task_pid_vnr(struct task_struct *tsk)
  1464. {
  1465. return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
  1466. }
  1467. static inline pid_t task_tgid_nr(struct task_struct *tsk)
  1468. {
  1469. return tsk->tgid;
  1470. }
  1471. pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
  1472. static inline pid_t task_tgid_vnr(struct task_struct *tsk)
  1473. {
  1474. return pid_vnr(task_tgid(tsk));
  1475. }
  1476. static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
  1477. struct pid_namespace *ns)
  1478. {
  1479. return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
  1480. }
  1481. static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
  1482. {
  1483. return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
  1484. }
  1485. static inline pid_t task_session_nr_ns(struct task_struct *tsk,
  1486. struct pid_namespace *ns)
  1487. {
  1488. return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
  1489. }
  1490. static inline pid_t task_session_vnr(struct task_struct *tsk)
  1491. {
  1492. return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
  1493. }
  1494. /* obsolete, do not use */
  1495. static inline pid_t task_pgrp_nr(struct task_struct *tsk)
  1496. {
  1497. return task_pgrp_nr_ns(tsk, &init_pid_ns);
  1498. }
  1499. /**
  1500. * pid_alive - check that a task structure is not stale
  1501. * @p: Task structure to be checked.
  1502. *
  1503. * Test if a process is not yet dead (at most zombie state)
  1504. * If pid_alive fails, then pointers within the task structure
  1505. * can be stale and must not be dereferenced.
  1506. */
  1507. static inline int pid_alive(struct task_struct *p)
  1508. {
  1509. return p->pids[PIDTYPE_PID].pid != NULL;
  1510. }
  1511. /**
  1512. * is_global_init - check if a task structure is init
  1513. * @tsk: Task structure to be checked.
  1514. *
  1515. * Check if a task structure is the first user space task the kernel created.
  1516. */
  1517. static inline int is_global_init(struct task_struct *tsk)
  1518. {
  1519. return tsk->pid == 1;
  1520. }
  1521. /*
  1522. * is_container_init:
  1523. * check whether in the task is init in its own pid namespace.
  1524. */
  1525. extern int is_container_init(struct task_struct *tsk);
  1526. extern struct pid *cad_pid;
  1527. extern void free_task(struct task_struct *tsk);
  1528. #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
  1529. extern void __put_task_struct(struct task_struct *t);
  1530. static inline void put_task_struct(struct task_struct *t)
  1531. {
  1532. if (atomic_dec_and_test(&t->usage))
  1533. __put_task_struct(t);
  1534. }
  1535. extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
  1536. extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
  1537. extern int task_free_register(struct notifier_block *n);
  1538. extern int task_free_unregister(struct notifier_block *n);
  1539. /*
  1540. * Per process flags
  1541. */
  1542. #define PF_EXITING 0x00000004 /* getting shut down */
  1543. #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
  1544. #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
  1545. #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
  1546. #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
  1547. #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
  1548. #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
  1549. #define PF_DUMPCORE 0x00000200 /* dumped core */
  1550. #define PF_SIGNALED 0x00000400 /* killed by a signal */
  1551. #define PF_MEMALLOC 0x00000800 /* Allocating memory */
  1552. #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
  1553. #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
  1554. #define PF_WAKE_UP_IDLE 0x00004000 /* try to wake up on an idle CPU */
  1555. #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
  1556. #define PF_FROZEN 0x00010000 /* frozen for system suspend */
  1557. #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
  1558. #define PF_KSWAPD 0x00040000 /* I am kswapd */
  1559. #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
  1560. #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
  1561. #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
  1562. #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
  1563. #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
  1564. #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
  1565. #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
  1566. #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
  1567. #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
  1568. #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
  1569. #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
  1570. /*
  1571. * Only the _current_ task can read/write to tsk->flags, but other
  1572. * tasks can access tsk->flags in readonly mode for example
  1573. * with tsk_used_math (like during threaded core dumping).
  1574. * There is however an exception to this rule during ptrace
  1575. * or during fork: the ptracer task is allowed to write to the
  1576. * child->flags of its traced child (same goes for fork, the parent
  1577. * can write to the child->flags), because we're guaranteed the
  1578. * child is not running and in turn not changing child->flags
  1579. * at the same time the parent does it.
  1580. */
  1581. #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
  1582. #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
  1583. #define clear_used_math() clear_stopped_child_used_math(current)
  1584. #define set_used_math() set_stopped_child_used_math(current)
  1585. #define conditional_stopped_child_used_math(condition, child) \
  1586. do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
  1587. #define conditional_used_math(condition) \
  1588. conditional_stopped_child_used_math(condition, current)
  1589. #define copy_to_stopped_child_used_math(child) \
  1590. do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
  1591. /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
  1592. #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
  1593. #define used_math() tsk_used_math(current)
  1594. /*
  1595. * task->jobctl flags
  1596. */
  1597. #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
  1598. #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
  1599. #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
  1600. #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
  1601. #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
  1602. #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
  1603. #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
  1604. #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
  1605. #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
  1606. #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
  1607. #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
  1608. #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
  1609. #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
  1610. #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
  1611. #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
  1612. #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
  1613. #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
  1614. extern bool task_set_jobctl_pending(struct task_struct *task,
  1615. unsigned int mask);
  1616. extern void task_clear_jobctl_trapping(struct task_struct *task);
  1617. extern void task_clear_jobctl_pending(struct task_struct *task,
  1618. unsigned int mask);
  1619. #ifdef CONFIG_PREEMPT_RCU
  1620. #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
  1621. #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
  1622. static inline void rcu_copy_process(struct task_struct *p)
  1623. {
  1624. p->rcu_read_lock_nesting = 0;
  1625. p->rcu_read_unlock_special = 0;
  1626. #ifdef CONFIG_TREE_PREEMPT_RCU
  1627. p->rcu_blocked_node = NULL;
  1628. #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
  1629. #ifdef CONFIG_RCU_BOOST
  1630. p->rcu_boost_mutex = NULL;
  1631. #endif /* #ifdef CONFIG_RCU_BOOST */
  1632. INIT_LIST_HEAD(&p->rcu_node_entry);
  1633. }
  1634. #else
  1635. static inline void rcu_copy_process(struct task_struct *p)
  1636. {
  1637. }
  1638. #endif
  1639. #ifdef CONFIG_SMP
  1640. extern void do_set_cpus_allowed(struct task_struct *p,
  1641. const struct cpumask *new_mask);
  1642. extern int set_cpus_allowed_ptr(struct task_struct *p,
  1643. const struct cpumask *new_mask);
  1644. #else
  1645. static inline void do_set_cpus_allowed(struct task_struct *p,
  1646. const struct cpumask *new_mask)
  1647. {
  1648. }
  1649. static inline int set_cpus_allowed_ptr(struct task_struct *p,
  1650. const struct cpumask *new_mask)
  1651. {
  1652. if (!cpumask_test_cpu(0, new_mask))
  1653. return -EINVAL;
  1654. return 0;
  1655. }
  1656. #endif
  1657. static inline void set_wake_up_idle(bool enabled)
  1658. {
  1659. if (enabled)
  1660. current->flags |= PF_WAKE_UP_IDLE;
  1661. else
  1662. current->flags &= ~PF_WAKE_UP_IDLE;
  1663. }
  1664. #ifndef CONFIG_CPUMASK_OFFSTACK
  1665. static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
  1666. {
  1667. return set_cpus_allowed_ptr(p, &new_mask);
  1668. }
  1669. #endif
  1670. /*
  1671. * Do not use outside of architecture code which knows its limitations.
  1672. *
  1673. * sched_clock() has no promise of monotonicity or bounded drift between
  1674. * CPUs, use (which you should not) requires disabling IRQs.
  1675. *
  1676. * Please use one of the three interfaces below.
  1677. */
  1678. extern unsigned long long notrace sched_clock(void);
  1679. /*
  1680. * See the comment in kernel/sched_clock.c
  1681. */
  1682. extern u64 cpu_clock(int cpu);
  1683. extern u64 local_clock(void);
  1684. extern u64 sched_clock_cpu(int cpu);
  1685. extern void sched_clock_init(void);
  1686. #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
  1687. static inline void sched_clock_tick(void)
  1688. {
  1689. }
  1690. static inline void sched_clock_idle_sleep_event(void)
  1691. {
  1692. }
  1693. static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
  1694. {
  1695. }
  1696. #else
  1697. /*
  1698. * Architectures can set this to 1 if they have specified
  1699. * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
  1700. * but then during bootup it turns out that sched_clock()
  1701. * is reliable after all:
  1702. */
  1703. extern int sched_clock_stable;
  1704. extern void sched_clock_tick(void);
  1705. extern void sched_clock_idle_sleep_event(void);
  1706. extern void sched_clock_idle_wakeup_event(u64 delta_ns);
  1707. #endif
  1708. #ifdef CONFIG_IRQ_TIME_ACCOUNTING
  1709. /*
  1710. * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
  1711. * The reason for this explicit opt-in is not to have perf penalty with
  1712. * slow sched_clocks.
  1713. */
  1714. extern void enable_sched_clock_irqtime(void);
  1715. extern void disable_sched_clock_irqtime(void);
  1716. #else
  1717. static inline void enable_sched_clock_irqtime(void) {}
  1718. static inline void disable_sched_clock_irqtime(void) {}
  1719. #endif
  1720. extern unsigned long long
  1721. task_sched_runtime(struct task_struct *task);
  1722. /* sched_exec is called by processes performing an exec */
  1723. #ifdef CONFIG_SMP
  1724. extern void sched_exec(void);
  1725. #else
  1726. #define sched_exec() {}
  1727. #endif
  1728. extern void sched_clock_idle_sleep_event(void);
  1729. extern void sched_clock_idle_wakeup_event(u64 delta_ns);
  1730. #ifdef CONFIG_HOTPLUG_CPU
  1731. extern void idle_task_exit(void);
  1732. #else
  1733. static inline void idle_task_exit(void) {}
  1734. #endif
  1735. #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
  1736. extern void wake_up_idle_cpu(int cpu);
  1737. #else
  1738. static inline void wake_up_idle_cpu(int cpu) { }
  1739. #endif
  1740. extern unsigned int sysctl_sched_latency;
  1741. extern unsigned int sysctl_sched_min_granularity;
  1742. extern unsigned int sysctl_sched_wakeup_granularity;
  1743. extern unsigned int sysctl_sched_child_runs_first;
  1744. extern unsigned int sysctl_sched_wake_to_idle;
  1745. enum sched_tunable_scaling {
  1746. SCHED_TUNABLESCALING_NONE,
  1747. SCHED_TUNABLESCALING_LOG,
  1748. SCHED_TUNABLESCALING_LINEAR,
  1749. SCHED_TUNABLESCALING_END,
  1750. };
  1751. extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
  1752. #ifdef CONFIG_SCHED_DEBUG
  1753. extern unsigned int sysctl_sched_migration_cost;
  1754. extern unsigned int sysctl_sched_nr_migrate;
  1755. extern unsigned int sysctl_sched_time_avg;
  1756. extern unsigned int sysctl_timer_migration;
  1757. extern unsigned int sysctl_sched_shares_window;
  1758. int sched_proc_update_handler(struct ctl_table *table, int write,
  1759. void __user *buffer, size_t *length,
  1760. loff_t *ppos);
  1761. #endif
  1762. #ifdef CONFIG_SCHED_DEBUG
  1763. static inline unsigned int get_sysctl_timer_migration(void)
  1764. {
  1765. return sysctl_timer_migration;
  1766. }
  1767. #else
  1768. static inline unsigned int get_sysctl_timer_migration(void)
  1769. {
  1770. return 1;
  1771. }
  1772. #endif
  1773. extern unsigned int sysctl_sched_rt_period;
  1774. extern int sysctl_sched_rt_runtime;
  1775. int sched_rt_handler(struct ctl_table *table, int write,
  1776. void __user *buffer, size_t *lenp,
  1777. loff_t *ppos);
  1778. #ifdef CONFIG_SCHED_AUTOGROUP
  1779. extern unsigned int sysctl_sched_autogroup_enabled;
  1780. extern void sched_autogroup_create_attach(struct task_struct *p);
  1781. extern void sched_autogroup_detach(struct task_struct *p);
  1782. extern void sched_autogroup_fork(struct signal_struct *sig);
  1783. extern void sched_autogroup_exit(struct signal_struct *sig);
  1784. #ifdef CONFIG_PROC_FS
  1785. extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
  1786. extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
  1787. #endif
  1788. #else
  1789. static inline void sched_autogroup_create_attach(struct task_struct *p) { }
  1790. static inline void sched_autogroup_detach(struct task_struct *p) { }
  1791. static inline void sched_autogroup_fork(struct signal_struct *sig) { }
  1792. static inline void sched_autogroup_exit(struct signal_struct *sig) { }
  1793. #endif
  1794. #ifdef CONFIG_CFS_BANDWIDTH
  1795. extern unsigned int sysctl_sched_cfs_bandwidth_slice;
  1796. #endif
  1797. #ifdef CONFIG_RT_MUTEXES
  1798. extern int rt_mutex_getprio(struct task_struct *p);
  1799. extern void rt_mutex_setprio(struct task_struct *p, int prio);
  1800. extern void rt_mutex_adjust_pi(struct task_struct *p);
  1801. static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
  1802. {
  1803. return tsk->pi_blocked_on != NULL;
  1804. }
  1805. #else
  1806. static inline int rt_mutex_getprio(struct task_struct *p)
  1807. {
  1808. return p->normal_prio;
  1809. }
  1810. # define rt_mutex_adjust_pi(p) do { } while (0)
  1811. static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
  1812. {
  1813. return false;
  1814. }
  1815. #endif
  1816. extern bool yield_to(struct task_struct *p, bool preempt);
  1817. extern void set_user_nice(struct task_struct *p, long nice);
  1818. extern int task_prio(const struct task_struct *p);
  1819. extern int task_nice(const struct task_struct *p);
  1820. extern int can_nice(const struct task_struct *p, const int nice);
  1821. extern int task_curr(const struct task_struct *p);
  1822. extern int idle_cpu(int cpu);
  1823. extern int sched_setscheduler(struct task_struct *, int,
  1824. const struct sched_param *);
  1825. extern int sched_setscheduler_nocheck(struct task_struct *, int,
  1826. const struct sched_param *);
  1827. extern struct task_struct *idle_task(int cpu);
  1828. /**
  1829. * is_idle_task - is the specified task an idle task?
  1830. * @p: the task in question.
  1831. */
  1832. static inline bool is_idle_task(const struct task_struct *p)
  1833. {
  1834. return p->pid == 0;
  1835. }
  1836. extern struct task_struct *curr_task(int cpu);
  1837. extern void set_curr_task(int cpu, struct task_struct *p);
  1838. void yield(void);
  1839. /*
  1840. * The default (Linux) execution domain.
  1841. */
  1842. extern struct exec_domain default_exec_domain;
  1843. union thread_union {
  1844. struct thread_info thread_info;
  1845. unsigned long stack[THREAD_SIZE/sizeof(long)];
  1846. };
  1847. #ifndef __HAVE_ARCH_KSTACK_END
  1848. static inline int kstack_end(void *addr)
  1849. {
  1850. /* Reliable end of stack detection:
  1851. * Some APM bios versions misalign the stack
  1852. */
  1853. return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
  1854. }
  1855. #endif
  1856. extern union thread_union init_thread_union;
  1857. extern struct task_struct init_task;
  1858. extern struct mm_struct init_mm;
  1859. extern struct pid_namespace init_pid_ns;
  1860. /*
  1861. * find a task by one of its numerical ids
  1862. *
  1863. * find_task_by_pid_ns():
  1864. * finds a task by its pid in the specified namespace
  1865. * find_task_by_vpid():
  1866. * finds a task by its virtual pid
  1867. *
  1868. * see also find_vpid() etc in include/linux/pid.h
  1869. */
  1870. extern struct task_struct *find_task_by_vpid(pid_t nr);
  1871. extern struct task_struct *find_task_by_pid_ns(pid_t nr,
  1872. struct pid_namespace *ns);
  1873. extern void __set_special_pids(struct pid *pid);
  1874. /* per-UID process charging. */
  1875. extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
  1876. static inline struct user_struct *get_uid(struct user_struct *u)
  1877. {
  1878. atomic_inc(&u->__count);
  1879. return u;
  1880. }
  1881. extern void free_uid(struct user_struct *);
  1882. extern void release_uids(struct user_namespace *ns);
  1883. #include <asm/current.h>
  1884. extern void xtime_update(unsigned long ticks);
  1885. extern int wake_up_state(struct task_struct *tsk, unsigned int state);
  1886. extern int wake_up_process(struct task_struct *tsk);
  1887. extern void wake_up_new_task(struct task_struct *tsk);
  1888. #ifdef CONFIG_SMP
  1889. extern void kick_process(struct task_struct *tsk);
  1890. #else
  1891. static inline void kick_process(struct task_struct *tsk) { }
  1892. #endif
  1893. extern void sched_fork(struct task_struct *p);
  1894. extern void sched_dead(struct task_struct *p);
  1895. extern void proc_caches_init(void);
  1896. extern void flush_signals(struct task_struct *);
  1897. extern void __flush_signals(struct task_struct *);
  1898. extern void ignore_signals(struct task_struct *);
  1899. extern void flush_signal_handlers(struct task_struct *, int force_default);
  1900. extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
  1901. static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
  1902. {
  1903. unsigned long flags;
  1904. int ret;
  1905. spin_lock_irqsave(&tsk->sighand->siglock, flags);
  1906. ret = dequeue_signal(tsk, mask, info);
  1907. spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
  1908. return ret;
  1909. }
  1910. extern void block_all_signals(int (*notifier)(void *priv), void *priv,
  1911. sigset_t *mask);
  1912. extern void unblock_all_signals(void);
  1913. extern void release_task(struct task_struct * p);
  1914. extern int send_sig_info(int, struct siginfo *, struct task_struct *);
  1915. extern int force_sigsegv(int, struct task_struct *);
  1916. extern int force_sig_info(int, struct siginfo *, struct task_struct *);
  1917. extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
  1918. extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
  1919. extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
  1920. const struct cred *, u32);
  1921. extern int kill_pgrp(struct pid *pid, int sig, int priv);
  1922. extern int kill_pid(struct pid *pid, int sig, int priv);
  1923. extern int kill_proc_info(int, struct siginfo *, pid_t);
  1924. extern __must_check bool do_notify_parent(struct task_struct *, int);
  1925. extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
  1926. extern void force_sig(int, struct task_struct *);
  1927. extern int send_sig(int, struct task_struct *, int);
  1928. extern int zap_other_threads(struct task_struct *p);
  1929. extern struct sigqueue *sigqueue_alloc(void);
  1930. extern void sigqueue_free(struct sigqueue *);
  1931. extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
  1932. extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
  1933. extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
  1934. static inline int kill_cad_pid(int sig, int priv)
  1935. {
  1936. return kill_pid(cad_pid, sig, priv);
  1937. }
  1938. /* These can be the second arg to send_sig_info/send_group_sig_info. */
  1939. #define SEND_SIG_NOINFO ((struct siginfo *) 0)
  1940. #define SEND_SIG_PRIV ((struct siginfo *) 1)
  1941. #define SEND_SIG_FORCED ((struct siginfo *) 2)
  1942. /*
  1943. * True if we are on the alternate signal stack.
  1944. */
  1945. static inline int on_sig_stack(unsigned long sp)
  1946. {
  1947. #ifdef CONFIG_STACK_GROWSUP
  1948. return sp >= current->sas_ss_sp &&
  1949. sp - current->sas_ss_sp < current->sas_ss_size;
  1950. #else
  1951. return sp > current->sas_ss_sp &&
  1952. sp - current->sas_ss_sp <= current->sas_ss_size;
  1953. #endif
  1954. }
  1955. static inline int sas_ss_flags(unsigned long sp)
  1956. {
  1957. return (current->sas_ss_size == 0 ? SS_DISABLE
  1958. : on_sig_stack(sp) ? SS_ONSTACK : 0);
  1959. }
  1960. /*
  1961. * Routines for handling mm_structs
  1962. */
  1963. extern struct mm_struct * mm_alloc(void);
  1964. /* mmdrop drops the mm and the page tables */
  1965. extern void __mmdrop(struct mm_struct *);
  1966. static inline void mmdrop(struct mm_struct * mm)
  1967. {
  1968. if (unlikely(atomic_dec_and_test(&mm->mm_count)))
  1969. __mmdrop(mm);
  1970. }
  1971. /* mmput gets rid of the mappings and all user-space */
  1972. extern int mmput(struct mm_struct *);
  1973. /* Grab a reference to a task's mm, if it is not already going away */
  1974. extern struct mm_struct *get_task_mm(struct task_struct *task);
  1975. /*
  1976. * Grab a reference to a task's mm, if it is not already going away
  1977. * and ptrace_may_access with the mode parameter passed to it
  1978. * succeeds.
  1979. */
  1980. extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
  1981. /* Remove the current tasks stale references to the old mm_struct */
  1982. extern void mm_release(struct task_struct *, struct mm_struct *);
  1983. /* Allocate a new mm structure and copy contents from tsk->mm */
  1984. extern struct mm_struct *dup_mm(struct task_struct *tsk);
  1985. extern int copy_thread(unsigned long, unsigned long, unsigned long,
  1986. struct task_struct *, struct pt_regs *);
  1987. extern void flush_thread(void);
  1988. extern void exit_thread(void);
  1989. extern void exit_files(struct task_struct *);
  1990. extern void __cleanup_sighand(struct sighand_struct *);
  1991. extern void exit_itimers(struct signal_struct *);
  1992. extern void flush_itimer_signals(void);
  1993. extern void do_group_exit(int);
  1994. extern void daemonize(const char *, ...);
  1995. extern int allow_signal(int);
  1996. extern int disallow_signal(int);
  1997. extern int do_execve(const char *,
  1998. const char __user * const __user *,
  1999. const char __user * const __user *, struct pt_regs *);
  2000. extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
  2001. struct task_struct *fork_idle(int);
  2002. extern void set_task_comm(struct task_struct *tsk, char *from);
  2003. extern char *get_task_comm(char *to, struct task_struct *tsk);
  2004. #ifdef CONFIG_SMP
  2005. void scheduler_ipi(void);
  2006. extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
  2007. #else
  2008. static inline void scheduler_ipi(void) { }
  2009. static inline unsigned long wait_task_inactive(struct task_struct *p,
  2010. long match_state)
  2011. {
  2012. return 1;
  2013. }
  2014. #endif
  2015. #define next_task(p) \
  2016. list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
  2017. #define for_each_process(p) \
  2018. for (p = &init_task ; (p = next_task(p)) != &init_task ; )
  2019. extern bool current_is_single_threaded(void);
  2020. /*
  2021. * Careful: do_each_thread/while_each_thread is a double loop so
  2022. * 'break' will not work as expected - use goto instead.
  2023. */
  2024. #define do_each_thread(g, t) \
  2025. for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
  2026. #define while_each_thread(g, t) \
  2027. while ((t = next_thread(t)) != g)
  2028. static inline int get_nr_threads(struct task_struct *tsk)
  2029. {
  2030. return tsk->signal->nr_threads;
  2031. }
  2032. static inline bool thread_group_leader(struct task_struct *p)
  2033. {
  2034. return p->exit_signal >= 0;
  2035. }
  2036. /* Do to the insanities of de_thread it is possible for a process
  2037. * to have the pid of the thread group leader without actually being
  2038. * the thread group leader. For iteration through the pids in proc
  2039. * all we care about is that we have a task with the appropriate
  2040. * pid, we don't actually care if we have the right task.
  2041. */
  2042. static inline int has_group_leader_pid(struct task_struct *p)
  2043. {
  2044. return p->pid == p->tgid;
  2045. }
  2046. static inline
  2047. int same_thread_group(struct task_struct *p1, struct task_struct *p2)
  2048. {
  2049. return p1->tgid == p2->tgid;
  2050. }
  2051. static inline struct task_struct *next_thread(const struct task_struct *p)
  2052. {
  2053. return list_entry_rcu(p->thread_group.next,
  2054. struct task_struct, thread_group);
  2055. }
  2056. static inline int thread_group_empty(struct task_struct *p)
  2057. {
  2058. return list_empty(&p->thread_group);
  2059. }
  2060. #define delay_group_leader(p) \
  2061. (thread_group_leader(p) && !thread_group_empty(p))
  2062. /*
  2063. * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
  2064. * subscriptions and synchronises with wait4(). Also used in procfs. Also
  2065. * pins the final release of task.io_context. Also protects ->cpuset and
  2066. * ->cgroup.subsys[]. And ->vfork_done.
  2067. *
  2068. * Nests both inside and outside of read_lock(&tasklist_lock).
  2069. * It must not be nested with write_lock_irq(&tasklist_lock),
  2070. * neither inside nor outside.
  2071. */
  2072. static inline void task_lock(struct task_struct *p)
  2073. {
  2074. spin_lock(&p->alloc_lock);
  2075. }
  2076. static inline void task_unlock(struct task_struct *p)
  2077. {
  2078. spin_unlock(&p->alloc_lock);
  2079. }
  2080. extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
  2081. unsigned long *flags);
  2082. static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
  2083. unsigned long *flags)
  2084. {
  2085. struct sighand_struct *ret;
  2086. ret = __lock_task_sighand(tsk, flags);
  2087. (void)__cond_lock(&tsk->sighand->siglock, ret);
  2088. return ret;
  2089. }
  2090. static inline void unlock_task_sighand(struct task_struct *tsk,
  2091. unsigned long *flags)
  2092. {
  2093. spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
  2094. }
  2095. #ifdef CONFIG_CGROUPS
  2096. static inline void threadgroup_change_begin(struct task_struct *tsk)
  2097. {
  2098. down_read(&tsk->signal->group_rwsem);
  2099. }
  2100. static inline void threadgroup_change_end(struct task_struct *tsk)
  2101. {
  2102. up_read(&tsk->signal->group_rwsem);
  2103. }
  2104. /**
  2105. * threadgroup_lock - lock threadgroup
  2106. * @tsk: member task of the threadgroup to lock
  2107. *
  2108. * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
  2109. * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
  2110. * perform exec. This is useful for cases where the threadgroup needs to
  2111. * stay stable across blockable operations.
  2112. *
  2113. * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
  2114. * synchronization. While held, no new task will be added to threadgroup
  2115. * and no existing live task will have its PF_EXITING set.
  2116. *
  2117. * During exec, a task goes and puts its thread group through unusual
  2118. * changes. After de-threading, exclusive access is assumed to resources
  2119. * which are usually shared by tasks in the same group - e.g. sighand may
  2120. * be replaced with a new one. Also, the exec'ing task takes over group
  2121. * leader role including its pid. Exclude these changes while locked by
  2122. * grabbing cred_guard_mutex which is used to synchronize exec path.
  2123. */
  2124. static inline void threadgroup_lock(struct task_struct *tsk)
  2125. {
  2126. /*
  2127. * exec uses exit for de-threading nesting group_rwsem inside
  2128. * cred_guard_mutex. Grab cred_guard_mutex first.
  2129. */
  2130. mutex_lock(&tsk->signal->cred_guard_mutex);
  2131. down_write(&tsk->signal->group_rwsem);
  2132. }
  2133. /**
  2134. * threadgroup_unlock - unlock threadgroup
  2135. * @tsk: member task of the threadgroup to unlock
  2136. *
  2137. * Reverse threadgroup_lock().
  2138. */
  2139. static inline void threadgroup_unlock(struct task_struct *tsk)
  2140. {
  2141. up_write(&tsk->signal->group_rwsem);
  2142. mutex_unlock(&tsk->signal->cred_guard_mutex);
  2143. }
  2144. #else
  2145. static inline void threadgroup_change_begin(struct task_struct *tsk) {}
  2146. static inline void threadgroup_change_end(struct task_struct *tsk) {}
  2147. static inline void threadgroup_lock(struct task_struct *tsk) {}
  2148. static inline void threadgroup_unlock(struct task_struct *tsk) {}
  2149. #endif
  2150. #ifndef __HAVE_THREAD_FUNCTIONS
  2151. #define task_thread_info(task) ((struct thread_info *)(task)->stack)
  2152. #define task_stack_page(task) ((task)->stack)
  2153. static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
  2154. {
  2155. *task_thread_info(p) = *task_thread_info(org);
  2156. task_thread_info(p)->task = p;
  2157. }
  2158. static inline unsigned long *end_of_stack(struct task_struct *p)
  2159. {
  2160. return (unsigned long *)(task_thread_info(p) + 1);
  2161. }
  2162. #endif
  2163. static inline int object_is_on_stack(void *obj)
  2164. {
  2165. void *stack = task_stack_page(current);
  2166. return (obj >= stack) && (obj < (stack + THREAD_SIZE));
  2167. }
  2168. extern void thread_info_cache_init(void);
  2169. #ifdef CONFIG_DEBUG_STACK_USAGE
  2170. static inline unsigned long stack_not_used(struct task_struct *p)
  2171. {
  2172. unsigned long *n = end_of_stack(p);
  2173. do { /* Skip over canary */
  2174. n++;
  2175. } while (!*n);
  2176. return (unsigned long)n - (unsigned long)end_of_stack(p);
  2177. }
  2178. #endif
  2179. /* set thread flags in other task's structures
  2180. * - see asm/thread_info.h for TIF_xxxx flags available
  2181. */
  2182. static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
  2183. {
  2184. set_ti_thread_flag(task_thread_info(tsk), flag);
  2185. }
  2186. static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
  2187. {
  2188. clear_ti_thread_flag(task_thread_info(tsk), flag);
  2189. }
  2190. static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
  2191. {
  2192. return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
  2193. }
  2194. static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
  2195. {
  2196. return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
  2197. }
  2198. static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
  2199. {
  2200. return test_ti_thread_flag(task_thread_info(tsk), flag);
  2201. }
  2202. static inline void set_tsk_need_resched(struct task_struct *tsk)
  2203. {
  2204. set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
  2205. }
  2206. static inline void clear_tsk_need_resched(struct task_struct *tsk)
  2207. {
  2208. clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
  2209. }
  2210. static inline int test_tsk_need_resched(struct task_struct *tsk)
  2211. {
  2212. return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
  2213. }
  2214. static inline int restart_syscall(void)
  2215. {
  2216. set_tsk_thread_flag(current, TIF_SIGPENDING);
  2217. return -ERESTARTNOINTR;
  2218. }
  2219. static inline int signal_pending(struct task_struct *p)
  2220. {
  2221. return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
  2222. }
  2223. static inline int __fatal_signal_pending(struct task_struct *p)
  2224. {
  2225. return unlikely(sigismember(&p->pending.signal, SIGKILL));
  2226. }
  2227. static inline int fatal_signal_pending(struct task_struct *p)
  2228. {
  2229. return signal_pending(p) && __fatal_signal_pending(p);
  2230. }
  2231. static inline int signal_pending_state(long state, struct task_struct *p)
  2232. {
  2233. if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
  2234. return 0;
  2235. if (!signal_pending(p))
  2236. return 0;
  2237. return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
  2238. }
  2239. static inline int need_resched(void)
  2240. {
  2241. return unlikely(test_thread_flag(TIF_NEED_RESCHED));
  2242. }
  2243. /*
  2244. * cond_resched() and cond_resched_lock(): latency reduction via
  2245. * explicit rescheduling in places that are safe. The return
  2246. * value indicates whether a reschedule was done in fact.
  2247. * cond_resched_lock() will drop the spinlock before scheduling,
  2248. * cond_resched_softirq() will enable bhs before scheduling.
  2249. */
  2250. extern int _cond_resched(void);
  2251. #define cond_resched() ({ \
  2252. __might_sleep(__FILE__, __LINE__, 0); \
  2253. _cond_resched(); \
  2254. })
  2255. extern int __cond_resched_lock(spinlock_t *lock);
  2256. #ifdef CONFIG_PREEMPT_COUNT
  2257. #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
  2258. #else
  2259. #define PREEMPT_LOCK_OFFSET 0
  2260. #endif
  2261. #define cond_resched_lock(lock) ({ \
  2262. __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
  2263. __cond_resched_lock(lock); \
  2264. })
  2265. extern int __cond_resched_softirq(void);
  2266. #define cond_resched_softirq() ({ \
  2267. __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
  2268. __cond_resched_softirq(); \
  2269. })
  2270. /*
  2271. * Does a critical section need to be broken due to another
  2272. * task waiting?: (technically does not depend on CONFIG_PREEMPT,
  2273. * but a general need for low latency)
  2274. */
  2275. static inline int spin_needbreak(spinlock_t *lock)
  2276. {
  2277. #ifdef CONFIG_PREEMPT
  2278. return spin_is_contended(lock);
  2279. #else
  2280. return 0;
  2281. #endif
  2282. }
  2283. /*
  2284. * Thread group CPU time accounting.
  2285. */
  2286. void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
  2287. void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
  2288. static inline void thread_group_cputime_init(struct signal_struct *sig)
  2289. {
  2290. raw_spin_lock_init(&sig->cputimer.lock);
  2291. }
  2292. /*
  2293. * Reevaluate whether the task has signals pending delivery.
  2294. * Wake the task if so.
  2295. * This is required every time the blocked sigset_t changes.
  2296. * callers must hold sighand->siglock.
  2297. */
  2298. extern void recalc_sigpending_and_wake(struct task_struct *t);
  2299. extern void recalc_sigpending(void);
  2300. extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
  2301. static inline void signal_wake_up(struct task_struct *t, bool resume)
  2302. {
  2303. signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
  2304. }
  2305. static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
  2306. {
  2307. signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
  2308. }
  2309. /*
  2310. * Wrappers for p->thread_info->cpu access. No-op on UP.
  2311. */
  2312. #ifdef CONFIG_SMP
  2313. static inline unsigned int task_cpu(const struct task_struct *p)
  2314. {
  2315. return task_thread_info(p)->cpu;
  2316. }
  2317. extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
  2318. #else
  2319. static inline unsigned int task_cpu(const struct task_struct *p)
  2320. {
  2321. return 0;
  2322. }
  2323. static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
  2324. {
  2325. }
  2326. #endif /* CONFIG_SMP */
  2327. extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
  2328. extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
  2329. extern void normalize_rt_tasks(void);
  2330. #ifdef CONFIG_CGROUP_SCHED
  2331. extern struct task_group root_task_group;
  2332. extern struct task_group *sched_create_group(struct task_group *parent);
  2333. extern void sched_destroy_group(struct task_group *tg);
  2334. extern void sched_move_task(struct task_struct *tsk);
  2335. #ifdef CONFIG_FAIR_GROUP_SCHED
  2336. extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
  2337. extern unsigned long sched_group_shares(struct task_group *tg);
  2338. #endif
  2339. #ifdef CONFIG_RT_GROUP_SCHED
  2340. extern int sched_group_set_rt_runtime(struct task_group *tg,
  2341. long rt_runtime_us);
  2342. extern long sched_group_rt_runtime(struct task_group *tg);
  2343. extern int sched_group_set_rt_period(struct task_group *tg,
  2344. long rt_period_us);
  2345. extern long sched_group_rt_period(struct task_group *tg);
  2346. extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
  2347. #endif
  2348. #endif
  2349. extern int task_can_switch_user(struct user_struct *up,
  2350. struct task_struct *tsk);
  2351. #ifdef CONFIG_TASK_XACCT
  2352. static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
  2353. {
  2354. tsk->ioac.rchar += amt;
  2355. }
  2356. static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
  2357. {
  2358. tsk->ioac.wchar += amt;
  2359. }
  2360. static inline void inc_syscr(struct task_struct *tsk)
  2361. {
  2362. tsk->ioac.syscr++;
  2363. }
  2364. static inline void inc_syscw(struct task_struct *tsk)
  2365. {
  2366. tsk->ioac.syscw++;
  2367. }
  2368. #else
  2369. static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
  2370. {
  2371. }
  2372. static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
  2373. {
  2374. }
  2375. static inline void inc_syscr(struct task_struct *tsk)
  2376. {
  2377. }
  2378. static inline void inc_syscw(struct task_struct *tsk)
  2379. {
  2380. }
  2381. #endif
  2382. #ifndef TASK_SIZE_OF
  2383. #define TASK_SIZE_OF(tsk) TASK_SIZE
  2384. #endif
  2385. #ifdef CONFIG_MM_OWNER
  2386. extern void mm_update_next_owner(struct mm_struct *mm);
  2387. extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
  2388. #else
  2389. static inline void mm_update_next_owner(struct mm_struct *mm)
  2390. {
  2391. }
  2392. static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
  2393. {
  2394. }
  2395. #endif /* CONFIG_MM_OWNER */
  2396. static inline unsigned long task_rlimit(const struct task_struct *tsk,
  2397. unsigned int limit)
  2398. {
  2399. return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
  2400. }
  2401. static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
  2402. unsigned int limit)
  2403. {
  2404. return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
  2405. }
  2406. static inline unsigned long rlimit(unsigned int limit)
  2407. {
  2408. return task_rlimit(current, limit);
  2409. }
  2410. static inline unsigned long rlimit_max(unsigned int limit)
  2411. {
  2412. return task_rlimit_max(current, limit);
  2413. }
  2414. #endif /* __KERNEL__ */
  2415. #endif