sched.h revision 1d3504fcf5606579d60b649d19f44b3871c1ddae
1#ifndef _LINUX_SCHED_H 2#define _LINUX_SCHED_H 3 4/* 5 * cloning flags: 6 */ 7#define CSIGNAL 0x000000ff /* signal mask to be sent at exit */ 8#define CLONE_VM 0x00000100 /* set if VM shared between processes */ 9#define CLONE_FS 0x00000200 /* set if fs info shared between processes */ 10#define CLONE_FILES 0x00000400 /* set if open files shared between processes */ 11#define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */ 12#define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */ 13#define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */ 14#define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */ 15#define CLONE_THREAD 0x00010000 /* Same thread group? */ 16#define CLONE_NEWNS 0x00020000 /* New namespace group? */ 17#define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */ 18#define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */ 19#define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */ 20#define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */ 21#define CLONE_DETACHED 0x00400000 /* Unused, ignored */ 22#define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */ 23#define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */ 24#define CLONE_STOPPED 0x02000000 /* Start in stopped state */ 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/* 33 * Scheduling policies 34 */ 35#define SCHED_NORMAL 0 36#define SCHED_FIFO 1 37#define SCHED_RR 2 38#define SCHED_BATCH 3 39/* SCHED_ISO: reserved but not implemented yet */ 40#define SCHED_IDLE 5 41 42#ifdef __KERNEL__ 43 44struct sched_param { 45 int sched_priority; 46}; 47 48#include <asm/param.h> /* for HZ */ 49 50#include <linux/capability.h> 51#include <linux/threads.h> 52#include <linux/kernel.h> 53#include <linux/types.h> 54#include <linux/timex.h> 55#include <linux/jiffies.h> 56#include <linux/rbtree.h> 57#include <linux/thread_info.h> 58#include <linux/cpumask.h> 59#include <linux/errno.h> 60#include <linux/nodemask.h> 61#include <linux/mm_types.h> 62 63#include <asm/system.h> 64#include <asm/semaphore.h> 65#include <asm/page.h> 66#include <asm/ptrace.h> 67#include <asm/cputime.h> 68 69#include <linux/smp.h> 70#include <linux/sem.h> 71#include <linux/signal.h> 72#include <linux/securebits.h> 73#include <linux/fs_struct.h> 74#include <linux/compiler.h> 75#include <linux/completion.h> 76#include <linux/pid.h> 77#include <linux/percpu.h> 78#include <linux/topology.h> 79#include <linux/proportions.h> 80#include <linux/seccomp.h> 81#include <linux/rcupdate.h> 82#include <linux/rtmutex.h> 83 84#include <linux/time.h> 85#include <linux/param.h> 86#include <linux/resource.h> 87#include <linux/timer.h> 88#include <linux/hrtimer.h> 89#include <linux/task_io_accounting.h> 90#include <linux/kobject.h> 91#include <linux/latencytop.h> 92 93#include <asm/processor.h> 94 95struct mem_cgroup; 96struct exec_domain; 97struct futex_pi_state; 98struct robust_list_head; 99struct bio; 100 101/* 102 * List of flags we want to share for kernel threads, 103 * if only because they are not used by them anyway. 104 */ 105#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND) 106 107/* 108 * These are the constant used to fake the fixed-point load-average 109 * counting. Some notes: 110 * - 11 bit fractions expand to 22 bits by the multiplies: this gives 111 * a load-average precision of 10 bits integer + 11 bits fractional 112 * - if you want to count load-averages more often, you need more 113 * precision, or rounding will get you. With 2-second counting freq, 114 * the EXP_n values would be 1981, 2034 and 2043 if still using only 115 * 11 bit fractions. 116 */ 117extern unsigned long avenrun[]; /* Load averages */ 118 119#define FSHIFT 11 /* nr of bits of precision */ 120#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */ 121#define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */ 122#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */ 123#define EXP_5 2014 /* 1/exp(5sec/5min) */ 124#define EXP_15 2037 /* 1/exp(5sec/15min) */ 125 126#define CALC_LOAD(load,exp,n) \ 127 load *= exp; \ 128 load += n*(FIXED_1-exp); \ 129 load >>= FSHIFT; 130 131extern unsigned long total_forks; 132extern int nr_threads; 133DECLARE_PER_CPU(unsigned long, process_counts); 134extern int nr_processes(void); 135extern unsigned long nr_running(void); 136extern unsigned long nr_uninterruptible(void); 137extern unsigned long nr_active(void); 138extern unsigned long nr_iowait(void); 139extern unsigned long weighted_cpuload(const int cpu); 140 141struct seq_file; 142struct cfs_rq; 143struct task_group; 144#ifdef CONFIG_SCHED_DEBUG 145extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m); 146extern void proc_sched_set_task(struct task_struct *p); 147extern void 148print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq); 149#else 150static inline void 151proc_sched_show_task(struct task_struct *p, struct seq_file *m) 152{ 153} 154static inline void proc_sched_set_task(struct task_struct *p) 155{ 156} 157static inline void 158print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) 159{ 160} 161#endif 162 163/* 164 * Task state bitmask. NOTE! These bits are also 165 * encoded in fs/proc/array.c: get_task_state(). 166 * 167 * We have two separate sets of flags: task->state 168 * is about runnability, while task->exit_state are 169 * about the task exiting. Confusing, but this way 170 * modifying one set can't modify the other one by 171 * mistake. 172 */ 173#define TASK_RUNNING 0 174#define TASK_INTERRUPTIBLE 1 175#define TASK_UNINTERRUPTIBLE 2 176#define __TASK_STOPPED 4 177#define __TASK_TRACED 8 178/* in tsk->exit_state */ 179#define EXIT_ZOMBIE 16 180#define EXIT_DEAD 32 181/* in tsk->state again */ 182#define TASK_DEAD 64 183#define TASK_WAKEKILL 128 184 185/* Convenience macros for the sake of set_task_state */ 186#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) 187#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) 188#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED) 189 190/* Convenience macros for the sake of wake_up */ 191#define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) 192#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED) 193 194/* get_task_state() */ 195#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ 196 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ 197 __TASK_TRACED) 198 199#define task_is_traced(task) ((task->state & __TASK_TRACED) != 0) 200#define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0) 201#define task_is_stopped_or_traced(task) \ 202 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0) 203#define task_contributes_to_load(task) \ 204 ((task->state & TASK_UNINTERRUPTIBLE) != 0) 205 206#define __set_task_state(tsk, state_value) \ 207 do { (tsk)->state = (state_value); } while (0) 208#define set_task_state(tsk, state_value) \ 209 set_mb((tsk)->state, (state_value)) 210 211/* 212 * set_current_state() includes a barrier so that the write of current->state 213 * is correctly serialised wrt the caller's subsequent test of whether to 214 * actually sleep: 215 * 216 * set_current_state(TASK_UNINTERRUPTIBLE); 217 * if (do_i_need_to_sleep()) 218 * schedule(); 219 * 220 * If the caller does not need such serialisation then use __set_current_state() 221 */ 222#define __set_current_state(state_value) \ 223 do { current->state = (state_value); } while (0) 224#define set_current_state(state_value) \ 225 set_mb(current->state, (state_value)) 226 227/* Task command name length */ 228#define TASK_COMM_LEN 16 229 230#include <linux/spinlock.h> 231 232/* 233 * This serializes "schedule()" and also protects 234 * the run-queue from deletions/modifications (but 235 * _adding_ to the beginning of the run-queue has 236 * a separate lock). 237 */ 238extern rwlock_t tasklist_lock; 239extern spinlock_t mmlist_lock; 240 241struct task_struct; 242 243extern void sched_init(void); 244extern void sched_init_smp(void); 245extern asmlinkage void schedule_tail(struct task_struct *prev); 246extern void init_idle(struct task_struct *idle, int cpu); 247extern void init_idle_bootup_task(struct task_struct *idle); 248 249extern cpumask_t nohz_cpu_mask; 250#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ) 251extern int select_nohz_load_balancer(int cpu); 252#else 253static inline int select_nohz_load_balancer(int cpu) 254{ 255 return 0; 256} 257#endif 258 259extern unsigned long rt_needs_cpu(int cpu); 260 261/* 262 * Only dump TASK_* tasks. (0 for all tasks) 263 */ 264extern void show_state_filter(unsigned long state_filter); 265 266static inline void show_state(void) 267{ 268 show_state_filter(0); 269} 270 271extern void show_regs(struct pt_regs *); 272 273/* 274 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current 275 * task), SP is the stack pointer of the first frame that should be shown in the back 276 * trace (or NULL if the entire call-chain of the task should be shown). 277 */ 278extern void show_stack(struct task_struct *task, unsigned long *sp); 279 280void io_schedule(void); 281long io_schedule_timeout(long timeout); 282 283extern void cpu_init (void); 284extern void trap_init(void); 285extern void account_process_tick(struct task_struct *task, int user); 286extern void update_process_times(int user); 287extern void scheduler_tick(void); 288extern void hrtick_resched(void); 289 290extern void sched_show_task(struct task_struct *p); 291 292#ifdef CONFIG_DETECT_SOFTLOCKUP 293extern void softlockup_tick(void); 294extern void spawn_softlockup_task(void); 295extern void touch_softlockup_watchdog(void); 296extern void touch_all_softlockup_watchdogs(void); 297extern unsigned long softlockup_thresh; 298extern unsigned long sysctl_hung_task_check_count; 299extern unsigned long sysctl_hung_task_timeout_secs; 300extern unsigned long sysctl_hung_task_warnings; 301#else 302static inline void softlockup_tick(void) 303{ 304} 305static inline void spawn_softlockup_task(void) 306{ 307} 308static inline void touch_softlockup_watchdog(void) 309{ 310} 311static inline void touch_all_softlockup_watchdogs(void) 312{ 313} 314#endif 315 316 317/* Attach to any functions which should be ignored in wchan output. */ 318#define __sched __attribute__((__section__(".sched.text"))) 319 320/* Linker adds these: start and end of __sched functions */ 321extern char __sched_text_start[], __sched_text_end[]; 322 323/* Is this address in the __sched functions? */ 324extern int in_sched_functions(unsigned long addr); 325 326#define MAX_SCHEDULE_TIMEOUT LONG_MAX 327extern signed long schedule_timeout(signed long timeout); 328extern signed long schedule_timeout_interruptible(signed long timeout); 329extern signed long schedule_timeout_killable(signed long timeout); 330extern signed long schedule_timeout_uninterruptible(signed long timeout); 331asmlinkage void schedule(void); 332 333struct nsproxy; 334struct user_namespace; 335 336/* Maximum number of active map areas.. This is a random (large) number */ 337#define DEFAULT_MAX_MAP_COUNT 65536 338 339extern int sysctl_max_map_count; 340 341#include <linux/aio.h> 342 343extern unsigned long 344arch_get_unmapped_area(struct file *, unsigned long, unsigned long, 345 unsigned long, unsigned long); 346extern unsigned long 347arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr, 348 unsigned long len, unsigned long pgoff, 349 unsigned long flags); 350extern void arch_unmap_area(struct mm_struct *, unsigned long); 351extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long); 352 353#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS 354/* 355 * The mm counters are not protected by its page_table_lock, 356 * so must be incremented atomically. 357 */ 358#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value) 359#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member)) 360#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member) 361#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member) 362#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member) 363 364#else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ 365/* 366 * The mm counters are protected by its page_table_lock, 367 * so can be incremented directly. 368 */ 369#define set_mm_counter(mm, member, value) (mm)->_##member = (value) 370#define get_mm_counter(mm, member) ((mm)->_##member) 371#define add_mm_counter(mm, member, value) (mm)->_##member += (value) 372#define inc_mm_counter(mm, member) (mm)->_##member++ 373#define dec_mm_counter(mm, member) (mm)->_##member-- 374 375#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ 376 377#define get_mm_rss(mm) \ 378 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss)) 379#define update_hiwater_rss(mm) do { \ 380 unsigned long _rss = get_mm_rss(mm); \ 381 if ((mm)->hiwater_rss < _rss) \ 382 (mm)->hiwater_rss = _rss; \ 383} while (0) 384#define update_hiwater_vm(mm) do { \ 385 if ((mm)->hiwater_vm < (mm)->total_vm) \ 386 (mm)->hiwater_vm = (mm)->total_vm; \ 387} while (0) 388 389extern void set_dumpable(struct mm_struct *mm, int value); 390extern int get_dumpable(struct mm_struct *mm); 391 392/* mm flags */ 393/* dumpable bits */ 394#define MMF_DUMPABLE 0 /* core dump is permitted */ 395#define MMF_DUMP_SECURELY 1 /* core file is readable only by root */ 396#define MMF_DUMPABLE_BITS 2 397 398/* coredump filter bits */ 399#define MMF_DUMP_ANON_PRIVATE 2 400#define MMF_DUMP_ANON_SHARED 3 401#define MMF_DUMP_MAPPED_PRIVATE 4 402#define MMF_DUMP_MAPPED_SHARED 5 403#define MMF_DUMP_ELF_HEADERS 6 404#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS 405#define MMF_DUMP_FILTER_BITS 5 406#define MMF_DUMP_FILTER_MASK \ 407 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT) 408#define MMF_DUMP_FILTER_DEFAULT \ 409 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED)) 410 411struct sighand_struct { 412 atomic_t count; 413 struct k_sigaction action[_NSIG]; 414 spinlock_t siglock; 415 wait_queue_head_t signalfd_wqh; 416}; 417 418struct pacct_struct { 419 int ac_flag; 420 long ac_exitcode; 421 unsigned long ac_mem; 422 cputime_t ac_utime, ac_stime; 423 unsigned long ac_minflt, ac_majflt; 424}; 425 426/* 427 * NOTE! "signal_struct" does not have it's own 428 * locking, because a shared signal_struct always 429 * implies a shared sighand_struct, so locking 430 * sighand_struct is always a proper superset of 431 * the locking of signal_struct. 432 */ 433struct signal_struct { 434 atomic_t count; 435 atomic_t live; 436 437 wait_queue_head_t wait_chldexit; /* for wait4() */ 438 439 /* current thread group signal load-balancing target: */ 440 struct task_struct *curr_target; 441 442 /* shared signal handling: */ 443 struct sigpending shared_pending; 444 445 /* thread group exit support */ 446 int group_exit_code; 447 /* overloaded: 448 * - notify group_exit_task when ->count is equal to notify_count 449 * - everyone except group_exit_task is stopped during signal delivery 450 * of fatal signals, group_exit_task processes the signal. 451 */ 452 struct task_struct *group_exit_task; 453 int notify_count; 454 455 /* thread group stop support, overloads group_exit_code too */ 456 int group_stop_count; 457 unsigned int flags; /* see SIGNAL_* flags below */ 458 459 /* POSIX.1b Interval Timers */ 460 struct list_head posix_timers; 461 462 /* ITIMER_REAL timer for the process */ 463 struct hrtimer real_timer; 464 struct pid *leader_pid; 465 ktime_t it_real_incr; 466 467 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */ 468 cputime_t it_prof_expires, it_virt_expires; 469 cputime_t it_prof_incr, it_virt_incr; 470 471 /* job control IDs */ 472 473 /* 474 * pgrp and session fields are deprecated. 475 * use the task_session_Xnr and task_pgrp_Xnr routines below 476 */ 477 478 union { 479 pid_t pgrp __deprecated; 480 pid_t __pgrp; 481 }; 482 483 struct pid *tty_old_pgrp; 484 485 union { 486 pid_t session __deprecated; 487 pid_t __session; 488 }; 489 490 /* boolean value for session group leader */ 491 int leader; 492 493 struct tty_struct *tty; /* NULL if no tty */ 494 495 /* 496 * Cumulative resource counters for dead threads in the group, 497 * and for reaped dead child processes forked by this group. 498 * Live threads maintain their own counters and add to these 499 * in __exit_signal, except for the group leader. 500 */ 501 cputime_t utime, stime, cutime, cstime; 502 cputime_t gtime; 503 cputime_t cgtime; 504 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw; 505 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt; 506 unsigned long inblock, oublock, cinblock, coublock; 507 508 /* 509 * Cumulative ns of scheduled CPU time for dead threads in the 510 * group, not including a zombie group leader. (This only differs 511 * from jiffies_to_ns(utime + stime) if sched_clock uses something 512 * other than jiffies.) 513 */ 514 unsigned long long sum_sched_runtime; 515 516 /* 517 * We don't bother to synchronize most readers of this at all, 518 * because there is no reader checking a limit that actually needs 519 * to get both rlim_cur and rlim_max atomically, and either one 520 * alone is a single word that can safely be read normally. 521 * getrlimit/setrlimit use task_lock(current->group_leader) to 522 * protect this instead of the siglock, because they really 523 * have no need to disable irqs. 524 */ 525 struct rlimit rlim[RLIM_NLIMITS]; 526 527 struct list_head cpu_timers[3]; 528 529 /* keep the process-shared keyrings here so that they do the right 530 * thing in threads created with CLONE_THREAD */ 531#ifdef CONFIG_KEYS 532 struct key *session_keyring; /* keyring inherited over fork */ 533 struct key *process_keyring; /* keyring private to this process */ 534#endif 535#ifdef CONFIG_BSD_PROCESS_ACCT 536 struct pacct_struct pacct; /* per-process accounting information */ 537#endif 538#ifdef CONFIG_TASKSTATS 539 struct taskstats *stats; 540#endif 541#ifdef CONFIG_AUDIT 542 unsigned audit_tty; 543 struct tty_audit_buf *tty_audit_buf; 544#endif 545}; 546 547/* Context switch must be unlocked if interrupts are to be enabled */ 548#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW 549# define __ARCH_WANT_UNLOCKED_CTXSW 550#endif 551 552/* 553 * Bits in flags field of signal_struct. 554 */ 555#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */ 556#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */ 557#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */ 558#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */ 559 560/* If true, all threads except ->group_exit_task have pending SIGKILL */ 561static inline int signal_group_exit(const struct signal_struct *sig) 562{ 563 return (sig->flags & SIGNAL_GROUP_EXIT) || 564 (sig->group_exit_task != NULL); 565} 566 567/* 568 * Some day this will be a full-fledged user tracking system.. 569 */ 570struct user_struct { 571 atomic_t __count; /* reference count */ 572 atomic_t processes; /* How many processes does this user have? */ 573 atomic_t files; /* How many open files does this user have? */ 574 atomic_t sigpending; /* How many pending signals does this user have? */ 575#ifdef CONFIG_INOTIFY_USER 576 atomic_t inotify_watches; /* How many inotify watches does this user have? */ 577 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */ 578#endif 579#ifdef CONFIG_POSIX_MQUEUE 580 /* protected by mq_lock */ 581 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */ 582#endif 583 unsigned long locked_shm; /* How many pages of mlocked shm ? */ 584 585#ifdef CONFIG_KEYS 586 struct key *uid_keyring; /* UID specific keyring */ 587 struct key *session_keyring; /* UID's default session keyring */ 588#endif 589 590 /* Hash table maintenance information */ 591 struct hlist_node uidhash_node; 592 uid_t uid; 593 594#ifdef CONFIG_USER_SCHED 595 struct task_group *tg; 596#ifdef CONFIG_SYSFS 597 struct kobject kobj; 598 struct work_struct work; 599#endif 600#endif 601}; 602 603extern int uids_sysfs_init(void); 604 605extern struct user_struct *find_user(uid_t); 606 607extern struct user_struct root_user; 608#define INIT_USER (&root_user) 609 610struct backing_dev_info; 611struct reclaim_state; 612 613#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 614struct sched_info { 615 /* cumulative counters */ 616 unsigned long pcount; /* # of times run on this cpu */ 617 unsigned long long cpu_time, /* time spent on the cpu */ 618 run_delay; /* time spent waiting on a runqueue */ 619 620 /* timestamps */ 621 unsigned long long last_arrival,/* when we last ran on a cpu */ 622 last_queued; /* when we were last queued to run */ 623#ifdef CONFIG_SCHEDSTATS 624 /* BKL stats */ 625 unsigned int bkl_count; 626#endif 627}; 628#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */ 629 630#ifdef CONFIG_SCHEDSTATS 631extern const struct file_operations proc_schedstat_operations; 632#endif /* CONFIG_SCHEDSTATS */ 633 634#ifdef CONFIG_TASK_DELAY_ACCT 635struct task_delay_info { 636 spinlock_t lock; 637 unsigned int flags; /* Private per-task flags */ 638 639 /* For each stat XXX, add following, aligned appropriately 640 * 641 * struct timespec XXX_start, XXX_end; 642 * u64 XXX_delay; 643 * u32 XXX_count; 644 * 645 * Atomicity of updates to XXX_delay, XXX_count protected by 646 * single lock above (split into XXX_lock if contention is an issue). 647 */ 648 649 /* 650 * XXX_count is incremented on every XXX operation, the delay 651 * associated with the operation is added to XXX_delay. 652 * XXX_delay contains the accumulated delay time in nanoseconds. 653 */ 654 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */ 655 u64 blkio_delay; /* wait for sync block io completion */ 656 u64 swapin_delay; /* wait for swapin block io completion */ 657 u32 blkio_count; /* total count of the number of sync block */ 658 /* io operations performed */ 659 u32 swapin_count; /* total count of the number of swapin block */ 660 /* io operations performed */ 661}; 662#endif /* CONFIG_TASK_DELAY_ACCT */ 663 664static inline int sched_info_on(void) 665{ 666#ifdef CONFIG_SCHEDSTATS 667 return 1; 668#elif defined(CONFIG_TASK_DELAY_ACCT) 669 extern int delayacct_on; 670 return delayacct_on; 671#else 672 return 0; 673#endif 674} 675 676enum cpu_idle_type { 677 CPU_IDLE, 678 CPU_NOT_IDLE, 679 CPU_NEWLY_IDLE, 680 CPU_MAX_IDLE_TYPES 681}; 682 683/* 684 * sched-domains (multiprocessor balancing) declarations: 685 */ 686 687/* 688 * Increase resolution of nice-level calculations: 689 */ 690#define SCHED_LOAD_SHIFT 10 691#define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT) 692 693#define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE 694 695#ifdef CONFIG_SMP 696#define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */ 697#define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */ 698#define SD_BALANCE_EXEC 4 /* Balance on exec */ 699#define SD_BALANCE_FORK 8 /* Balance on fork, clone */ 700#define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */ 701#define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */ 702#define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */ 703#define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */ 704#define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */ 705#define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */ 706#define SD_SERIALIZE 1024 /* Only a single load balancing instance */ 707#define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */ 708 709#define BALANCE_FOR_MC_POWER \ 710 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0) 711 712#define BALANCE_FOR_PKG_POWER \ 713 ((sched_mc_power_savings || sched_smt_power_savings) ? \ 714 SD_POWERSAVINGS_BALANCE : 0) 715 716#define test_sd_parent(sd, flag) ((sd->parent && \ 717 (sd->parent->flags & flag)) ? 1 : 0) 718 719 720struct sched_group { 721 struct sched_group *next; /* Must be a circular list */ 722 cpumask_t cpumask; 723 724 /* 725 * CPU power of this group, SCHED_LOAD_SCALE being max power for a 726 * single CPU. This is read only (except for setup, hotplug CPU). 727 * Note : Never change cpu_power without recompute its reciprocal 728 */ 729 unsigned int __cpu_power; 730 /* 731 * reciprocal value of cpu_power to avoid expensive divides 732 * (see include/linux/reciprocal_div.h) 733 */ 734 u32 reciprocal_cpu_power; 735}; 736 737enum sched_domain_level { 738 SD_LV_NONE = 0, 739 SD_LV_SIBLING, 740 SD_LV_MC, 741 SD_LV_CPU, 742 SD_LV_NODE, 743 SD_LV_ALLNODES, 744 SD_LV_MAX 745}; 746 747struct sched_domain_attr { 748 int relax_domain_level; 749}; 750 751#define SD_ATTR_INIT (struct sched_domain_attr) { \ 752 .relax_domain_level = -1, \ 753} 754 755struct sched_domain { 756 /* These fields must be setup */ 757 struct sched_domain *parent; /* top domain must be null terminated */ 758 struct sched_domain *child; /* bottom domain must be null terminated */ 759 struct sched_group *groups; /* the balancing groups of the domain */ 760 cpumask_t span; /* span of all CPUs in this domain */ 761 unsigned long min_interval; /* Minimum balance interval ms */ 762 unsigned long max_interval; /* Maximum balance interval ms */ 763 unsigned int busy_factor; /* less balancing by factor if busy */ 764 unsigned int imbalance_pct; /* No balance until over watermark */ 765 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */ 766 unsigned int busy_idx; 767 unsigned int idle_idx; 768 unsigned int newidle_idx; 769 unsigned int wake_idx; 770 unsigned int forkexec_idx; 771 int flags; /* See SD_* */ 772 enum sched_domain_level level; 773 774 /* Runtime fields. */ 775 unsigned long last_balance; /* init to jiffies. units in jiffies */ 776 unsigned int balance_interval; /* initialise to 1. units in ms. */ 777 unsigned int nr_balance_failed; /* initialise to 0 */ 778 779#ifdef CONFIG_SCHEDSTATS 780 /* load_balance() stats */ 781 unsigned int lb_count[CPU_MAX_IDLE_TYPES]; 782 unsigned int lb_failed[CPU_MAX_IDLE_TYPES]; 783 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES]; 784 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES]; 785 unsigned int lb_gained[CPU_MAX_IDLE_TYPES]; 786 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES]; 787 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES]; 788 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES]; 789 790 /* Active load balancing */ 791 unsigned int alb_count; 792 unsigned int alb_failed; 793 unsigned int alb_pushed; 794 795 /* SD_BALANCE_EXEC stats */ 796 unsigned int sbe_count; 797 unsigned int sbe_balanced; 798 unsigned int sbe_pushed; 799 800 /* SD_BALANCE_FORK stats */ 801 unsigned int sbf_count; 802 unsigned int sbf_balanced; 803 unsigned int sbf_pushed; 804 805 /* try_to_wake_up() stats */ 806 unsigned int ttwu_wake_remote; 807 unsigned int ttwu_move_affine; 808 unsigned int ttwu_move_balance; 809#endif 810}; 811 812extern void partition_sched_domains(int ndoms_new, cpumask_t *doms_new, 813 struct sched_domain_attr *dattr_new); 814extern int arch_reinit_sched_domains(void); 815 816#endif /* CONFIG_SMP */ 817 818/* 819 * A runqueue laden with a single nice 0 task scores a weighted_cpuload of 820 * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a 821 * task of nice 0 or enough lower priority tasks to bring up the 822 * weighted_cpuload 823 */ 824static inline int above_background_load(void) 825{ 826 unsigned long cpu; 827 828 for_each_online_cpu(cpu) { 829 if (weighted_cpuload(cpu) >= SCHED_LOAD_SCALE) 830 return 1; 831 } 832 return 0; 833} 834 835struct io_context; /* See blkdev.h */ 836#define NGROUPS_SMALL 32 837#define NGROUPS_PER_BLOCK ((unsigned int)(PAGE_SIZE / sizeof(gid_t))) 838struct group_info { 839 int ngroups; 840 atomic_t usage; 841 gid_t small_block[NGROUPS_SMALL]; 842 int nblocks; 843 gid_t *blocks[0]; 844}; 845 846/* 847 * get_group_info() must be called with the owning task locked (via task_lock()) 848 * when task != current. The reason being that the vast majority of callers are 849 * looking at current->group_info, which can not be changed except by the 850 * current task. Changing current->group_info requires the task lock, too. 851 */ 852#define get_group_info(group_info) do { \ 853 atomic_inc(&(group_info)->usage); \ 854} while (0) 855 856#define put_group_info(group_info) do { \ 857 if (atomic_dec_and_test(&(group_info)->usage)) \ 858 groups_free(group_info); \ 859} while (0) 860 861extern struct group_info *groups_alloc(int gidsetsize); 862extern void groups_free(struct group_info *group_info); 863extern int set_current_groups(struct group_info *group_info); 864extern int groups_search(struct group_info *group_info, gid_t grp); 865/* access the groups "array" with this macro */ 866#define GROUP_AT(gi, i) \ 867 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK]) 868 869#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK 870extern void prefetch_stack(struct task_struct *t); 871#else 872static inline void prefetch_stack(struct task_struct *t) { } 873#endif 874 875struct audit_context; /* See audit.c */ 876struct mempolicy; 877struct pipe_inode_info; 878struct uts_namespace; 879 880struct rq; 881struct sched_domain; 882 883struct sched_class { 884 const struct sched_class *next; 885 886 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup); 887 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep); 888 void (*yield_task) (struct rq *rq); 889 int (*select_task_rq)(struct task_struct *p, int sync); 890 891 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p); 892 893 struct task_struct * (*pick_next_task) (struct rq *rq); 894 void (*put_prev_task) (struct rq *rq, struct task_struct *p); 895 896#ifdef CONFIG_SMP 897 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu, 898 struct rq *busiest, unsigned long max_load_move, 899 struct sched_domain *sd, enum cpu_idle_type idle, 900 int *all_pinned, int *this_best_prio); 901 902 int (*move_one_task) (struct rq *this_rq, int this_cpu, 903 struct rq *busiest, struct sched_domain *sd, 904 enum cpu_idle_type idle); 905 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task); 906 void (*post_schedule) (struct rq *this_rq); 907 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task); 908#endif 909 910 void (*set_curr_task) (struct rq *rq); 911 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued); 912 void (*task_new) (struct rq *rq, struct task_struct *p); 913 void (*set_cpus_allowed)(struct task_struct *p, 914 const cpumask_t *newmask); 915 916 void (*join_domain)(struct rq *rq); 917 void (*leave_domain)(struct rq *rq); 918 919 void (*switched_from) (struct rq *this_rq, struct task_struct *task, 920 int running); 921 void (*switched_to) (struct rq *this_rq, struct task_struct *task, 922 int running); 923 void (*prio_changed) (struct rq *this_rq, struct task_struct *task, 924 int oldprio, int running); 925 926#ifdef CONFIG_FAIR_GROUP_SCHED 927 void (*moved_group) (struct task_struct *p); 928#endif 929}; 930 931struct load_weight { 932 unsigned long weight, inv_weight; 933}; 934 935/* 936 * CFS stats for a schedulable entity (task, task-group etc) 937 * 938 * Current field usage histogram: 939 * 940 * 4 se->block_start 941 * 4 se->run_node 942 * 4 se->sleep_start 943 * 6 se->load.weight 944 */ 945struct sched_entity { 946 struct load_weight load; /* for load-balancing */ 947 struct rb_node run_node; 948 unsigned int on_rq; 949 950 u64 exec_start; 951 u64 sum_exec_runtime; 952 u64 vruntime; 953 u64 prev_sum_exec_runtime; 954 955 u64 last_wakeup; 956 u64 avg_overlap; 957 958#ifdef CONFIG_SCHEDSTATS 959 u64 wait_start; 960 u64 wait_max; 961 u64 wait_count; 962 u64 wait_sum; 963 964 u64 sleep_start; 965 u64 sleep_max; 966 s64 sum_sleep_runtime; 967 968 u64 block_start; 969 u64 block_max; 970 u64 exec_max; 971 u64 slice_max; 972 973 u64 nr_migrations; 974 u64 nr_migrations_cold; 975 u64 nr_failed_migrations_affine; 976 u64 nr_failed_migrations_running; 977 u64 nr_failed_migrations_hot; 978 u64 nr_forced_migrations; 979 u64 nr_forced2_migrations; 980 981 u64 nr_wakeups; 982 u64 nr_wakeups_sync; 983 u64 nr_wakeups_migrate; 984 u64 nr_wakeups_local; 985 u64 nr_wakeups_remote; 986 u64 nr_wakeups_affine; 987 u64 nr_wakeups_affine_attempts; 988 u64 nr_wakeups_passive; 989 u64 nr_wakeups_idle; 990#endif 991 992#ifdef CONFIG_FAIR_GROUP_SCHED 993 struct sched_entity *parent; 994 /* rq on which this entity is (to be) queued: */ 995 struct cfs_rq *cfs_rq; 996 /* rq "owned" by this entity/group: */ 997 struct cfs_rq *my_q; 998#endif 999}; 1000 1001struct sched_rt_entity { 1002 struct list_head run_list; 1003 unsigned int time_slice; 1004 unsigned long timeout; 1005 int nr_cpus_allowed; 1006 1007#ifdef CONFIG_RT_GROUP_SCHED 1008 struct sched_rt_entity *parent; 1009 /* rq on which this entity is (to be) queued: */ 1010 struct rt_rq *rt_rq; 1011 /* rq "owned" by this entity/group: */ 1012 struct rt_rq *my_q; 1013#endif 1014}; 1015 1016struct task_struct { 1017 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */ 1018 void *stack; 1019 atomic_t usage; 1020 unsigned int flags; /* per process flags, defined below */ 1021 unsigned int ptrace; 1022 1023 int lock_depth; /* BKL lock depth */ 1024 1025#ifdef CONFIG_SMP 1026#ifdef __ARCH_WANT_UNLOCKED_CTXSW 1027 int oncpu; 1028#endif 1029#endif 1030 1031 int prio, static_prio, normal_prio; 1032 const struct sched_class *sched_class; 1033 struct sched_entity se; 1034 struct sched_rt_entity rt; 1035 1036#ifdef CONFIG_PREEMPT_NOTIFIERS 1037 /* list of struct preempt_notifier: */ 1038 struct hlist_head preempt_notifiers; 1039#endif 1040 1041 /* 1042 * fpu_counter contains the number of consecutive context switches 1043 * that the FPU is used. If this is over a threshold, the lazy fpu 1044 * saving becomes unlazy to save the trap. This is an unsigned char 1045 * so that after 256 times the counter wraps and the behavior turns 1046 * lazy again; this to deal with bursty apps that only use FPU for 1047 * a short time 1048 */ 1049 unsigned char fpu_counter; 1050 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */ 1051#ifdef CONFIG_BLK_DEV_IO_TRACE 1052 unsigned int btrace_seq; 1053#endif 1054 1055 unsigned int policy; 1056 cpumask_t cpus_allowed; 1057 1058#ifdef CONFIG_PREEMPT_RCU 1059 int rcu_read_lock_nesting; 1060 int rcu_flipctr_idx; 1061#endif /* #ifdef CONFIG_PREEMPT_RCU */ 1062 1063#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 1064 struct sched_info sched_info; 1065#endif 1066 1067 struct list_head tasks; 1068 /* 1069 * ptrace_list/ptrace_children forms the list of my children 1070 * that were stolen by a ptracer. 1071 */ 1072 struct list_head ptrace_children; 1073 struct list_head ptrace_list; 1074 1075 struct mm_struct *mm, *active_mm; 1076 1077/* task state */ 1078 struct linux_binfmt *binfmt; 1079 int exit_state; 1080 int exit_code, exit_signal; 1081 int pdeath_signal; /* The signal sent when the parent dies */ 1082 /* ??? */ 1083 unsigned int personality; 1084 unsigned did_exec:1; 1085 pid_t pid; 1086 pid_t tgid; 1087 1088#ifdef CONFIG_CC_STACKPROTECTOR 1089 /* Canary value for the -fstack-protector gcc feature */ 1090 unsigned long stack_canary; 1091#endif 1092 /* 1093 * pointers to (original) parent process, youngest child, younger sibling, 1094 * older sibling, respectively. (p->father can be replaced with 1095 * p->parent->pid) 1096 */ 1097 struct task_struct *real_parent; /* real parent process (when being debugged) */ 1098 struct task_struct *parent; /* parent process */ 1099 /* 1100 * children/sibling forms the list of my children plus the 1101 * tasks I'm ptracing. 1102 */ 1103 struct list_head children; /* list of my children */ 1104 struct list_head sibling; /* linkage in my parent's children list */ 1105 struct task_struct *group_leader; /* threadgroup leader */ 1106 1107 /* PID/PID hash table linkage. */ 1108 struct pid_link pids[PIDTYPE_MAX]; 1109 struct list_head thread_group; 1110 1111 struct completion *vfork_done; /* for vfork() */ 1112 int __user *set_child_tid; /* CLONE_CHILD_SETTID */ 1113 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ 1114 1115 unsigned int rt_priority; 1116 cputime_t utime, stime, utimescaled, stimescaled; 1117 cputime_t gtime; 1118 cputime_t prev_utime, prev_stime; 1119 unsigned long nvcsw, nivcsw; /* context switch counts */ 1120 struct timespec start_time; /* monotonic time */ 1121 struct timespec real_start_time; /* boot based time */ 1122/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ 1123 unsigned long min_flt, maj_flt; 1124 1125 cputime_t it_prof_expires, it_virt_expires; 1126 unsigned long long it_sched_expires; 1127 struct list_head cpu_timers[3]; 1128 1129/* process credentials */ 1130 uid_t uid,euid,suid,fsuid; 1131 gid_t gid,egid,sgid,fsgid; 1132 struct group_info *group_info; 1133 kernel_cap_t cap_effective, cap_inheritable, cap_permitted, cap_bset; 1134 unsigned keep_capabilities:1; 1135 struct user_struct *user; 1136#ifdef CONFIG_KEYS 1137 struct key *request_key_auth; /* assumed request_key authority */ 1138 struct key *thread_keyring; /* keyring private to this thread */ 1139 unsigned char jit_keyring; /* default keyring to attach requested keys to */ 1140#endif 1141 char comm[TASK_COMM_LEN]; /* executable name excluding path 1142 - access with [gs]et_task_comm (which lock 1143 it with task_lock()) 1144 - initialized normally by flush_old_exec */ 1145/* file system info */ 1146 int link_count, total_link_count; 1147#ifdef CONFIG_SYSVIPC 1148/* ipc stuff */ 1149 struct sysv_sem sysvsem; 1150#endif 1151#ifdef CONFIG_DETECT_SOFTLOCKUP 1152/* hung task detection */ 1153 unsigned long last_switch_timestamp; 1154 unsigned long last_switch_count; 1155#endif 1156/* CPU-specific state of this task */ 1157 struct thread_struct thread; 1158/* filesystem information */ 1159 struct fs_struct *fs; 1160/* open file information */ 1161 struct files_struct *files; 1162/* namespaces */ 1163 struct nsproxy *nsproxy; 1164/* signal handlers */ 1165 struct signal_struct *signal; 1166 struct sighand_struct *sighand; 1167 1168 sigset_t blocked, real_blocked; 1169 sigset_t saved_sigmask; /* To be restored with TIF_RESTORE_SIGMASK */ 1170 struct sigpending pending; 1171 1172 unsigned long sas_ss_sp; 1173 size_t sas_ss_size; 1174 int (*notifier)(void *priv); 1175 void *notifier_data; 1176 sigset_t *notifier_mask; 1177#ifdef CONFIG_SECURITY 1178 void *security; 1179#endif 1180 struct audit_context *audit_context; 1181#ifdef CONFIG_AUDITSYSCALL 1182 uid_t loginuid; 1183 unsigned int sessionid; 1184#endif 1185 seccomp_t seccomp; 1186 1187/* Thread group tracking */ 1188 u32 parent_exec_id; 1189 u32 self_exec_id; 1190/* Protection of (de-)allocation: mm, files, fs, tty, keyrings */ 1191 spinlock_t alloc_lock; 1192 1193 /* Protection of the PI data structures: */ 1194 spinlock_t pi_lock; 1195 1196#ifdef CONFIG_RT_MUTEXES 1197 /* PI waiters blocked on a rt_mutex held by this task */ 1198 struct plist_head pi_waiters; 1199 /* Deadlock detection and priority inheritance handling */ 1200 struct rt_mutex_waiter *pi_blocked_on; 1201#endif 1202 1203#ifdef CONFIG_DEBUG_MUTEXES 1204 /* mutex deadlock detection */ 1205 struct mutex_waiter *blocked_on; 1206#endif 1207#ifdef CONFIG_TRACE_IRQFLAGS 1208 unsigned int irq_events; 1209 int hardirqs_enabled; 1210 unsigned long hardirq_enable_ip; 1211 unsigned int hardirq_enable_event; 1212 unsigned long hardirq_disable_ip; 1213 unsigned int hardirq_disable_event; 1214 int softirqs_enabled; 1215 unsigned long softirq_disable_ip; 1216 unsigned int softirq_disable_event; 1217 unsigned long softirq_enable_ip; 1218 unsigned int softirq_enable_event; 1219 int hardirq_context; 1220 int softirq_context; 1221#endif 1222#ifdef CONFIG_LOCKDEP 1223# define MAX_LOCK_DEPTH 48UL 1224 u64 curr_chain_key; 1225 int lockdep_depth; 1226 struct held_lock held_locks[MAX_LOCK_DEPTH]; 1227 unsigned int lockdep_recursion; 1228#endif 1229 1230/* journalling filesystem info */ 1231 void *journal_info; 1232 1233/* stacked block device info */ 1234 struct bio *bio_list, **bio_tail; 1235 1236/* VM state */ 1237 struct reclaim_state *reclaim_state; 1238 1239 struct backing_dev_info *backing_dev_info; 1240 1241 struct io_context *io_context; 1242 1243 unsigned long ptrace_message; 1244 siginfo_t *last_siginfo; /* For ptrace use. */ 1245#ifdef CONFIG_TASK_XACCT 1246/* i/o counters(bytes read/written, #syscalls */ 1247 u64 rchar, wchar, syscr, syscw; 1248#endif 1249 struct task_io_accounting ioac; 1250#if defined(CONFIG_TASK_XACCT) 1251 u64 acct_rss_mem1; /* accumulated rss usage */ 1252 u64 acct_vm_mem1; /* accumulated virtual memory usage */ 1253 cputime_t acct_stimexpd;/* stime since last update */ 1254#endif 1255#ifdef CONFIG_NUMA 1256 struct mempolicy *mempolicy; 1257 short il_next; 1258#endif 1259#ifdef CONFIG_CPUSETS 1260 nodemask_t mems_allowed; 1261 int cpuset_mems_generation; 1262 int cpuset_mem_spread_rotor; 1263#endif 1264#ifdef CONFIG_CGROUPS 1265 /* Control Group info protected by css_set_lock */ 1266 struct css_set *cgroups; 1267 /* cg_list protected by css_set_lock and tsk->alloc_lock */ 1268 struct list_head cg_list; 1269#endif 1270#ifdef CONFIG_FUTEX 1271 struct robust_list_head __user *robust_list; 1272#ifdef CONFIG_COMPAT 1273 struct compat_robust_list_head __user *compat_robust_list; 1274#endif 1275 struct list_head pi_state_list; 1276 struct futex_pi_state *pi_state_cache; 1277#endif 1278 atomic_t fs_excl; /* holding fs exclusive resources */ 1279 struct rcu_head rcu; 1280 1281 /* 1282 * cache last used pipe for splice 1283 */ 1284 struct pipe_inode_info *splice_pipe; 1285#ifdef CONFIG_TASK_DELAY_ACCT 1286 struct task_delay_info *delays; 1287#endif 1288#ifdef CONFIG_FAULT_INJECTION 1289 int make_it_fail; 1290#endif 1291 struct prop_local_single dirties; 1292#ifdef CONFIG_LATENCYTOP 1293 int latency_record_count; 1294 struct latency_record latency_record[LT_SAVECOUNT]; 1295#endif 1296}; 1297 1298/* 1299 * Priority of a process goes from 0..MAX_PRIO-1, valid RT 1300 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH 1301 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority 1302 * values are inverted: lower p->prio value means higher priority. 1303 * 1304 * The MAX_USER_RT_PRIO value allows the actual maximum 1305 * RT priority to be separate from the value exported to 1306 * user-space. This allows kernel threads to set their 1307 * priority to a value higher than any user task. Note: 1308 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO. 1309 */ 1310 1311#define MAX_USER_RT_PRIO 100 1312#define MAX_RT_PRIO MAX_USER_RT_PRIO 1313 1314#define MAX_PRIO (MAX_RT_PRIO + 40) 1315#define DEFAULT_PRIO (MAX_RT_PRIO + 20) 1316 1317static inline int rt_prio(int prio) 1318{ 1319 if (unlikely(prio < MAX_RT_PRIO)) 1320 return 1; 1321 return 0; 1322} 1323 1324static inline int rt_task(struct task_struct *p) 1325{ 1326 return rt_prio(p->prio); 1327} 1328 1329static inline void set_task_session(struct task_struct *tsk, pid_t session) 1330{ 1331 tsk->signal->__session = session; 1332} 1333 1334static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp) 1335{ 1336 tsk->signal->__pgrp = pgrp; 1337} 1338 1339static inline struct pid *task_pid(struct task_struct *task) 1340{ 1341 return task->pids[PIDTYPE_PID].pid; 1342} 1343 1344static inline struct pid *task_tgid(struct task_struct *task) 1345{ 1346 return task->group_leader->pids[PIDTYPE_PID].pid; 1347} 1348 1349static inline struct pid *task_pgrp(struct task_struct *task) 1350{ 1351 return task->group_leader->pids[PIDTYPE_PGID].pid; 1352} 1353 1354static inline struct pid *task_session(struct task_struct *task) 1355{ 1356 return task->group_leader->pids[PIDTYPE_SID].pid; 1357} 1358 1359struct pid_namespace; 1360 1361/* 1362 * the helpers to get the task's different pids as they are seen 1363 * from various namespaces 1364 * 1365 * task_xid_nr() : global id, i.e. the id seen from the init namespace; 1366 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of 1367 * current. 1368 * task_xid_nr_ns() : id seen from the ns specified; 1369 * 1370 * set_task_vxid() : assigns a virtual id to a task; 1371 * 1372 * see also pid_nr() etc in include/linux/pid.h 1373 */ 1374 1375static inline pid_t task_pid_nr(struct task_struct *tsk) 1376{ 1377 return tsk->pid; 1378} 1379 1380pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1381 1382static inline pid_t task_pid_vnr(struct task_struct *tsk) 1383{ 1384 return pid_vnr(task_pid(tsk)); 1385} 1386 1387 1388static inline pid_t task_tgid_nr(struct task_struct *tsk) 1389{ 1390 return tsk->tgid; 1391} 1392 1393pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1394 1395static inline pid_t task_tgid_vnr(struct task_struct *tsk) 1396{ 1397 return pid_vnr(task_tgid(tsk)); 1398} 1399 1400 1401static inline pid_t task_pgrp_nr(struct task_struct *tsk) 1402{ 1403 return tsk->signal->__pgrp; 1404} 1405 1406pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1407 1408static inline pid_t task_pgrp_vnr(struct task_struct *tsk) 1409{ 1410 return pid_vnr(task_pgrp(tsk)); 1411} 1412 1413 1414static inline pid_t task_session_nr(struct task_struct *tsk) 1415{ 1416 return tsk->signal->__session; 1417} 1418 1419pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1420 1421static inline pid_t task_session_vnr(struct task_struct *tsk) 1422{ 1423 return pid_vnr(task_session(tsk)); 1424} 1425 1426 1427/** 1428 * pid_alive - check that a task structure is not stale 1429 * @p: Task structure to be checked. 1430 * 1431 * Test if a process is not yet dead (at most zombie state) 1432 * If pid_alive fails, then pointers within the task structure 1433 * can be stale and must not be dereferenced. 1434 */ 1435static inline int pid_alive(struct task_struct *p) 1436{ 1437 return p->pids[PIDTYPE_PID].pid != NULL; 1438} 1439 1440/** 1441 * is_global_init - check if a task structure is init 1442 * @tsk: Task structure to be checked. 1443 * 1444 * Check if a task structure is the first user space task the kernel created. 1445 */ 1446static inline int is_global_init(struct task_struct *tsk) 1447{ 1448 return tsk->pid == 1; 1449} 1450 1451/* 1452 * is_container_init: 1453 * check whether in the task is init in its own pid namespace. 1454 */ 1455extern int is_container_init(struct task_struct *tsk); 1456 1457extern struct pid *cad_pid; 1458 1459extern void free_task(struct task_struct *tsk); 1460#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0) 1461 1462extern void __put_task_struct(struct task_struct *t); 1463 1464static inline void put_task_struct(struct task_struct *t) 1465{ 1466 if (atomic_dec_and_test(&t->usage)) 1467 __put_task_struct(t); 1468} 1469 1470/* 1471 * Per process flags 1472 */ 1473#define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */ 1474 /* Not implemented yet, only for 486*/ 1475#define PF_STARTING 0x00000002 /* being created */ 1476#define PF_EXITING 0x00000004 /* getting shut down */ 1477#define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */ 1478#define PF_VCPU 0x00000010 /* I'm a virtual CPU */ 1479#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ 1480#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ 1481#define PF_DUMPCORE 0x00000200 /* dumped core */ 1482#define PF_SIGNALED 0x00000400 /* killed by a signal */ 1483#define PF_MEMALLOC 0x00000800 /* Allocating memory */ 1484#define PF_FLUSHER 0x00001000 /* responsible for disk writeback */ 1485#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */ 1486#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */ 1487#define PF_FROZEN 0x00010000 /* frozen for system suspend */ 1488#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ 1489#define PF_KSWAPD 0x00040000 /* I am kswapd */ 1490#define PF_SWAPOFF 0x00080000 /* I am in swapoff */ 1491#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ 1492#define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */ 1493#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */ 1494#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ 1495#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */ 1496#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */ 1497#define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */ 1498#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ 1499#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */ 1500 1501/* 1502 * Only the _current_ task can read/write to tsk->flags, but other 1503 * tasks can access tsk->flags in readonly mode for example 1504 * with tsk_used_math (like during threaded core dumping). 1505 * There is however an exception to this rule during ptrace 1506 * or during fork: the ptracer task is allowed to write to the 1507 * child->flags of its traced child (same goes for fork, the parent 1508 * can write to the child->flags), because we're guaranteed the 1509 * child is not running and in turn not changing child->flags 1510 * at the same time the parent does it. 1511 */ 1512#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) 1513#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) 1514#define clear_used_math() clear_stopped_child_used_math(current) 1515#define set_used_math() set_stopped_child_used_math(current) 1516#define conditional_stopped_child_used_math(condition, child) \ 1517 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) 1518#define conditional_used_math(condition) \ 1519 conditional_stopped_child_used_math(condition, current) 1520#define copy_to_stopped_child_used_math(child) \ 1521 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) 1522/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ 1523#define tsk_used_math(p) ((p)->flags & PF_USED_MATH) 1524#define used_math() tsk_used_math(current) 1525 1526#ifdef CONFIG_SMP 1527extern int set_cpus_allowed_ptr(struct task_struct *p, 1528 const cpumask_t *new_mask); 1529#else 1530static inline int set_cpus_allowed_ptr(struct task_struct *p, 1531 const cpumask_t *new_mask) 1532{ 1533 if (!cpu_isset(0, *new_mask)) 1534 return -EINVAL; 1535 return 0; 1536} 1537#endif 1538static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask) 1539{ 1540 return set_cpus_allowed_ptr(p, &new_mask); 1541} 1542 1543extern unsigned long long sched_clock(void); 1544 1545/* 1546 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu 1547 * clock constructed from sched_clock(): 1548 */ 1549extern unsigned long long cpu_clock(int cpu); 1550 1551extern unsigned long long 1552task_sched_runtime(struct task_struct *task); 1553 1554/* sched_exec is called by processes performing an exec */ 1555#ifdef CONFIG_SMP 1556extern void sched_exec(void); 1557#else 1558#define sched_exec() {} 1559#endif 1560 1561extern void sched_clock_idle_sleep_event(void); 1562extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1563 1564#ifdef CONFIG_HOTPLUG_CPU 1565extern void idle_task_exit(void); 1566#else 1567static inline void idle_task_exit(void) {} 1568#endif 1569 1570extern void sched_idle_next(void); 1571 1572#if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP) 1573extern void wake_up_idle_cpu(int cpu); 1574#else 1575static inline void wake_up_idle_cpu(int cpu) { } 1576#endif 1577 1578#ifdef CONFIG_SCHED_DEBUG 1579extern unsigned int sysctl_sched_latency; 1580extern unsigned int sysctl_sched_min_granularity; 1581extern unsigned int sysctl_sched_wakeup_granularity; 1582extern unsigned int sysctl_sched_child_runs_first; 1583extern unsigned int sysctl_sched_features; 1584extern unsigned int sysctl_sched_migration_cost; 1585extern unsigned int sysctl_sched_nr_migrate; 1586 1587int sched_nr_latency_handler(struct ctl_table *table, int write, 1588 struct file *file, void __user *buffer, size_t *length, 1589 loff_t *ppos); 1590#endif 1591extern unsigned int sysctl_sched_rt_period; 1592extern int sysctl_sched_rt_runtime; 1593 1594int sched_rt_handler(struct ctl_table *table, int write, 1595 struct file *filp, void __user *buffer, size_t *lenp, 1596 loff_t *ppos); 1597 1598extern unsigned int sysctl_sched_compat_yield; 1599 1600#ifdef CONFIG_RT_MUTEXES 1601extern int rt_mutex_getprio(struct task_struct *p); 1602extern void rt_mutex_setprio(struct task_struct *p, int prio); 1603extern void rt_mutex_adjust_pi(struct task_struct *p); 1604#else 1605static inline int rt_mutex_getprio(struct task_struct *p) 1606{ 1607 return p->normal_prio; 1608} 1609# define rt_mutex_adjust_pi(p) do { } while (0) 1610#endif 1611 1612extern void set_user_nice(struct task_struct *p, long nice); 1613extern int task_prio(const struct task_struct *p); 1614extern int task_nice(const struct task_struct *p); 1615extern int can_nice(const struct task_struct *p, const int nice); 1616extern int task_curr(const struct task_struct *p); 1617extern int idle_cpu(int cpu); 1618extern int sched_setscheduler(struct task_struct *, int, struct sched_param *); 1619extern struct task_struct *idle_task(int cpu); 1620extern struct task_struct *curr_task(int cpu); 1621extern void set_curr_task(int cpu, struct task_struct *p); 1622 1623void yield(void); 1624 1625/* 1626 * The default (Linux) execution domain. 1627 */ 1628extern struct exec_domain default_exec_domain; 1629 1630union thread_union { 1631 struct thread_info thread_info; 1632 unsigned long stack[THREAD_SIZE/sizeof(long)]; 1633}; 1634 1635#ifndef __HAVE_ARCH_KSTACK_END 1636static inline int kstack_end(void *addr) 1637{ 1638 /* Reliable end of stack detection: 1639 * Some APM bios versions misalign the stack 1640 */ 1641 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*))); 1642} 1643#endif 1644 1645extern union thread_union init_thread_union; 1646extern struct task_struct init_task; 1647 1648extern struct mm_struct init_mm; 1649 1650extern struct pid_namespace init_pid_ns; 1651 1652/* 1653 * find a task by one of its numerical ids 1654 * 1655 * find_task_by_pid_type_ns(): 1656 * it is the most generic call - it finds a task by all id, 1657 * type and namespace specified 1658 * find_task_by_pid_ns(): 1659 * finds a task by its pid in the specified namespace 1660 * find_task_by_vpid(): 1661 * finds a task by its virtual pid 1662 * find_task_by_pid(): 1663 * finds a task by its global pid 1664 * 1665 * see also find_pid() etc in include/linux/pid.h 1666 */ 1667 1668extern struct task_struct *find_task_by_pid_type_ns(int type, int pid, 1669 struct pid_namespace *ns); 1670 1671extern struct task_struct *find_task_by_pid(pid_t nr); 1672extern struct task_struct *find_task_by_vpid(pid_t nr); 1673extern struct task_struct *find_task_by_pid_ns(pid_t nr, 1674 struct pid_namespace *ns); 1675 1676extern void __set_special_pids(struct pid *pid); 1677 1678/* per-UID process charging. */ 1679extern struct user_struct * alloc_uid(struct user_namespace *, uid_t); 1680static inline struct user_struct *get_uid(struct user_struct *u) 1681{ 1682 atomic_inc(&u->__count); 1683 return u; 1684} 1685extern void free_uid(struct user_struct *); 1686extern void switch_uid(struct user_struct *); 1687extern void release_uids(struct user_namespace *ns); 1688 1689#include <asm/current.h> 1690 1691extern void do_timer(unsigned long ticks); 1692 1693extern int wake_up_state(struct task_struct *tsk, unsigned int state); 1694extern int wake_up_process(struct task_struct *tsk); 1695extern void wake_up_new_task(struct task_struct *tsk, 1696 unsigned long clone_flags); 1697#ifdef CONFIG_SMP 1698 extern void kick_process(struct task_struct *tsk); 1699#else 1700 static inline void kick_process(struct task_struct *tsk) { } 1701#endif 1702extern void sched_fork(struct task_struct *p, int clone_flags); 1703extern void sched_dead(struct task_struct *p); 1704 1705extern int in_group_p(gid_t); 1706extern int in_egroup_p(gid_t); 1707 1708extern void proc_caches_init(void); 1709extern void flush_signals(struct task_struct *); 1710extern void ignore_signals(struct task_struct *); 1711extern void flush_signal_handlers(struct task_struct *, int force_default); 1712extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info); 1713 1714static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info) 1715{ 1716 unsigned long flags; 1717 int ret; 1718 1719 spin_lock_irqsave(&tsk->sighand->siglock, flags); 1720 ret = dequeue_signal(tsk, mask, info); 1721 spin_unlock_irqrestore(&tsk->sighand->siglock, flags); 1722 1723 return ret; 1724} 1725 1726extern void block_all_signals(int (*notifier)(void *priv), void *priv, 1727 sigset_t *mask); 1728extern void unblock_all_signals(void); 1729extern void release_task(struct task_struct * p); 1730extern int send_sig_info(int, struct siginfo *, struct task_struct *); 1731extern int force_sigsegv(int, struct task_struct *); 1732extern int force_sig_info(int, struct siginfo *, struct task_struct *); 1733extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); 1734extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid); 1735extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32); 1736extern int kill_pgrp(struct pid *pid, int sig, int priv); 1737extern int kill_pid(struct pid *pid, int sig, int priv); 1738extern int kill_proc_info(int, struct siginfo *, pid_t); 1739extern void do_notify_parent(struct task_struct *, int); 1740extern void force_sig(int, struct task_struct *); 1741extern void force_sig_specific(int, struct task_struct *); 1742extern int send_sig(int, struct task_struct *, int); 1743extern void zap_other_threads(struct task_struct *p); 1744extern int kill_proc(pid_t, int, int); 1745extern struct sigqueue *sigqueue_alloc(void); 1746extern void sigqueue_free(struct sigqueue *); 1747extern int send_sigqueue(int, struct sigqueue *, struct task_struct *); 1748extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *); 1749extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *); 1750extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long); 1751 1752static inline int kill_cad_pid(int sig, int priv) 1753{ 1754 return kill_pid(cad_pid, sig, priv); 1755} 1756 1757/* These can be the second arg to send_sig_info/send_group_sig_info. */ 1758#define SEND_SIG_NOINFO ((struct siginfo *) 0) 1759#define SEND_SIG_PRIV ((struct siginfo *) 1) 1760#define SEND_SIG_FORCED ((struct siginfo *) 2) 1761 1762static inline int is_si_special(const struct siginfo *info) 1763{ 1764 return info <= SEND_SIG_FORCED; 1765} 1766 1767/* True if we are on the alternate signal stack. */ 1768 1769static inline int on_sig_stack(unsigned long sp) 1770{ 1771 return (sp - current->sas_ss_sp < current->sas_ss_size); 1772} 1773 1774static inline int sas_ss_flags(unsigned long sp) 1775{ 1776 return (current->sas_ss_size == 0 ? SS_DISABLE 1777 : on_sig_stack(sp) ? SS_ONSTACK : 0); 1778} 1779 1780/* 1781 * Routines for handling mm_structs 1782 */ 1783extern struct mm_struct * mm_alloc(void); 1784 1785/* mmdrop drops the mm and the page tables */ 1786extern void __mmdrop(struct mm_struct *); 1787static inline void mmdrop(struct mm_struct * mm) 1788{ 1789 if (unlikely(atomic_dec_and_test(&mm->mm_count))) 1790 __mmdrop(mm); 1791} 1792 1793/* mmput gets rid of the mappings and all user-space */ 1794extern void mmput(struct mm_struct *); 1795/* Grab a reference to a task's mm, if it is not already going away */ 1796extern struct mm_struct *get_task_mm(struct task_struct *task); 1797/* Remove the current tasks stale references to the old mm_struct */ 1798extern void mm_release(struct task_struct *, struct mm_struct *); 1799 1800extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *); 1801extern void flush_thread(void); 1802extern void exit_thread(void); 1803 1804extern void exit_files(struct task_struct *); 1805extern void __cleanup_signal(struct signal_struct *); 1806extern void __cleanup_sighand(struct sighand_struct *); 1807extern void exit_itimers(struct signal_struct *); 1808 1809extern NORET_TYPE void do_group_exit(int); 1810 1811extern void daemonize(const char *, ...); 1812extern int allow_signal(int); 1813extern int disallow_signal(int); 1814 1815extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *); 1816extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *); 1817struct task_struct *fork_idle(int); 1818 1819extern void set_task_comm(struct task_struct *tsk, char *from); 1820extern char *get_task_comm(char *to, struct task_struct *tsk); 1821 1822#ifdef CONFIG_SMP 1823extern void wait_task_inactive(struct task_struct * p); 1824#else 1825#define wait_task_inactive(p) do { } while (0) 1826#endif 1827 1828#define remove_parent(p) list_del_init(&(p)->sibling) 1829#define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children) 1830 1831#define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks) 1832 1833#define for_each_process(p) \ 1834 for (p = &init_task ; (p = next_task(p)) != &init_task ; ) 1835 1836/* 1837 * Careful: do_each_thread/while_each_thread is a double loop so 1838 * 'break' will not work as expected - use goto instead. 1839 */ 1840#define do_each_thread(g, t) \ 1841 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do 1842 1843#define while_each_thread(g, t) \ 1844 while ((t = next_thread(t)) != g) 1845 1846/* de_thread depends on thread_group_leader not being a pid based check */ 1847#define thread_group_leader(p) (p == p->group_leader) 1848 1849/* Do to the insanities of de_thread it is possible for a process 1850 * to have the pid of the thread group leader without actually being 1851 * the thread group leader. For iteration through the pids in proc 1852 * all we care about is that we have a task with the appropriate 1853 * pid, we don't actually care if we have the right task. 1854 */ 1855static inline int has_group_leader_pid(struct task_struct *p) 1856{ 1857 return p->pid == p->tgid; 1858} 1859 1860static inline 1861int same_thread_group(struct task_struct *p1, struct task_struct *p2) 1862{ 1863 return p1->tgid == p2->tgid; 1864} 1865 1866static inline struct task_struct *next_thread(const struct task_struct *p) 1867{ 1868 return list_entry(rcu_dereference(p->thread_group.next), 1869 struct task_struct, thread_group); 1870} 1871 1872static inline int thread_group_empty(struct task_struct *p) 1873{ 1874 return list_empty(&p->thread_group); 1875} 1876 1877#define delay_group_leader(p) \ 1878 (thread_group_leader(p) && !thread_group_empty(p)) 1879 1880/* 1881 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring 1882 * subscriptions and synchronises with wait4(). Also used in procfs. Also 1883 * pins the final release of task.io_context. Also protects ->cpuset and 1884 * ->cgroup.subsys[]. 1885 * 1886 * Nests both inside and outside of read_lock(&tasklist_lock). 1887 * It must not be nested with write_lock_irq(&tasklist_lock), 1888 * neither inside nor outside. 1889 */ 1890static inline void task_lock(struct task_struct *p) 1891{ 1892 spin_lock(&p->alloc_lock); 1893} 1894 1895static inline void task_unlock(struct task_struct *p) 1896{ 1897 spin_unlock(&p->alloc_lock); 1898} 1899 1900extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk, 1901 unsigned long *flags); 1902 1903static inline void unlock_task_sighand(struct task_struct *tsk, 1904 unsigned long *flags) 1905{ 1906 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags); 1907} 1908 1909#ifndef __HAVE_THREAD_FUNCTIONS 1910 1911#define task_thread_info(task) ((struct thread_info *)(task)->stack) 1912#define task_stack_page(task) ((task)->stack) 1913 1914static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org) 1915{ 1916 *task_thread_info(p) = *task_thread_info(org); 1917 task_thread_info(p)->task = p; 1918} 1919 1920static inline unsigned long *end_of_stack(struct task_struct *p) 1921{ 1922 return (unsigned long *)(task_thread_info(p) + 1); 1923} 1924 1925#endif 1926 1927/* set thread flags in other task's structures 1928 * - see asm/thread_info.h for TIF_xxxx flags available 1929 */ 1930static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) 1931{ 1932 set_ti_thread_flag(task_thread_info(tsk), flag); 1933} 1934 1935static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) 1936{ 1937 clear_ti_thread_flag(task_thread_info(tsk), flag); 1938} 1939 1940static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) 1941{ 1942 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); 1943} 1944 1945static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) 1946{ 1947 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); 1948} 1949 1950static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) 1951{ 1952 return test_ti_thread_flag(task_thread_info(tsk), flag); 1953} 1954 1955static inline void set_tsk_need_resched(struct task_struct *tsk) 1956{ 1957 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 1958} 1959 1960static inline void clear_tsk_need_resched(struct task_struct *tsk) 1961{ 1962 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 1963} 1964 1965static inline int signal_pending(struct task_struct *p) 1966{ 1967 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING)); 1968} 1969 1970extern int __fatal_signal_pending(struct task_struct *p); 1971 1972static inline int fatal_signal_pending(struct task_struct *p) 1973{ 1974 return signal_pending(p) && __fatal_signal_pending(p); 1975} 1976 1977static inline int need_resched(void) 1978{ 1979 return unlikely(test_thread_flag(TIF_NEED_RESCHED)); 1980} 1981 1982/* 1983 * cond_resched() and cond_resched_lock(): latency reduction via 1984 * explicit rescheduling in places that are safe. The return 1985 * value indicates whether a reschedule was done in fact. 1986 * cond_resched_lock() will drop the spinlock before scheduling, 1987 * cond_resched_softirq() will enable bhs before scheduling. 1988 */ 1989#ifdef CONFIG_PREEMPT 1990static inline int cond_resched(void) 1991{ 1992 return 0; 1993} 1994#else 1995extern int _cond_resched(void); 1996static inline int cond_resched(void) 1997{ 1998 return _cond_resched(); 1999} 2000#endif 2001extern int cond_resched_lock(spinlock_t * lock); 2002extern int cond_resched_softirq(void); 2003 2004/* 2005 * Does a critical section need to be broken due to another 2006 * task waiting?: (technically does not depend on CONFIG_PREEMPT, 2007 * but a general need for low latency) 2008 */ 2009static inline int spin_needbreak(spinlock_t *lock) 2010{ 2011#ifdef CONFIG_PREEMPT 2012 return spin_is_contended(lock); 2013#else 2014 return 0; 2015#endif 2016} 2017 2018/* 2019 * Reevaluate whether the task has signals pending delivery. 2020 * Wake the task if so. 2021 * This is required every time the blocked sigset_t changes. 2022 * callers must hold sighand->siglock. 2023 */ 2024extern void recalc_sigpending_and_wake(struct task_struct *t); 2025extern void recalc_sigpending(void); 2026 2027extern void signal_wake_up(struct task_struct *t, int resume_stopped); 2028 2029/* 2030 * Wrappers for p->thread_info->cpu access. No-op on UP. 2031 */ 2032#ifdef CONFIG_SMP 2033 2034static inline unsigned int task_cpu(const struct task_struct *p) 2035{ 2036 return task_thread_info(p)->cpu; 2037} 2038 2039extern void set_task_cpu(struct task_struct *p, unsigned int cpu); 2040 2041#else 2042 2043static inline unsigned int task_cpu(const struct task_struct *p) 2044{ 2045 return 0; 2046} 2047 2048static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 2049{ 2050} 2051 2052#endif /* CONFIG_SMP */ 2053 2054#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 2055extern void arch_pick_mmap_layout(struct mm_struct *mm); 2056#else 2057static inline void arch_pick_mmap_layout(struct mm_struct *mm) 2058{ 2059 mm->mmap_base = TASK_UNMAPPED_BASE; 2060 mm->get_unmapped_area = arch_get_unmapped_area; 2061 mm->unmap_area = arch_unmap_area; 2062} 2063#endif 2064 2065extern long sched_setaffinity(pid_t pid, const cpumask_t *new_mask); 2066extern long sched_getaffinity(pid_t pid, cpumask_t *mask); 2067 2068extern int sched_mc_power_savings, sched_smt_power_savings; 2069 2070extern void normalize_rt_tasks(void); 2071 2072#ifdef CONFIG_GROUP_SCHED 2073 2074extern struct task_group init_task_group; 2075#ifdef CONFIG_USER_SCHED 2076extern struct task_group root_task_group; 2077#endif 2078 2079extern struct task_group *sched_create_group(struct task_group *parent); 2080extern void sched_destroy_group(struct task_group *tg); 2081extern void sched_move_task(struct task_struct *tsk); 2082#ifdef CONFIG_FAIR_GROUP_SCHED 2083extern int sched_group_set_shares(struct task_group *tg, unsigned long shares); 2084extern unsigned long sched_group_shares(struct task_group *tg); 2085#endif 2086#ifdef CONFIG_RT_GROUP_SCHED 2087extern int sched_group_set_rt_runtime(struct task_group *tg, 2088 long rt_runtime_us); 2089extern long sched_group_rt_runtime(struct task_group *tg); 2090extern int sched_group_set_rt_period(struct task_group *tg, 2091 long rt_period_us); 2092extern long sched_group_rt_period(struct task_group *tg); 2093#endif 2094#endif 2095 2096#ifdef CONFIG_TASK_XACCT 2097static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2098{ 2099 tsk->rchar += amt; 2100} 2101 2102static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2103{ 2104 tsk->wchar += amt; 2105} 2106 2107static inline void inc_syscr(struct task_struct *tsk) 2108{ 2109 tsk->syscr++; 2110} 2111 2112static inline void inc_syscw(struct task_struct *tsk) 2113{ 2114 tsk->syscw++; 2115} 2116#else 2117static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2118{ 2119} 2120 2121static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2122{ 2123} 2124 2125static inline void inc_syscr(struct task_struct *tsk) 2126{ 2127} 2128 2129static inline void inc_syscw(struct task_struct *tsk) 2130{ 2131} 2132#endif 2133 2134#ifdef CONFIG_SMP 2135void migration_init(void); 2136#else 2137static inline void migration_init(void) 2138{ 2139} 2140#endif 2141 2142#ifndef TASK_SIZE_OF 2143#define TASK_SIZE_OF(tsk) TASK_SIZE 2144#endif 2145 2146#endif /* __KERNEL__ */ 2147 2148#endif 2149