sched.h revision 246bb0b1deb29726990620d8b5e55ca29f331362
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/page.h> 65#include <asm/ptrace.h> 66#include <asm/cputime.h> 67 68#include <linux/smp.h> 69#include <linux/sem.h> 70#include <linux/signal.h> 71#include <linux/fs_struct.h> 72#include <linux/compiler.h> 73#include <linux/completion.h> 74#include <linux/pid.h> 75#include <linux/percpu.h> 76#include <linux/topology.h> 77#include <linux/proportions.h> 78#include <linux/seccomp.h> 79#include <linux/rcupdate.h> 80#include <linux/rtmutex.h> 81 82#include <linux/time.h> 83#include <linux/param.h> 84#include <linux/resource.h> 85#include <linux/timer.h> 86#include <linux/hrtimer.h> 87#include <linux/task_io_accounting.h> 88#include <linux/kobject.h> 89#include <linux/latencytop.h> 90 91#include <asm/processor.h> 92 93struct mem_cgroup; 94struct exec_domain; 95struct futex_pi_state; 96struct robust_list_head; 97struct bio; 98 99/* 100 * List of flags we want to share for kernel threads, 101 * if only because they are not used by them anyway. 102 */ 103#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND) 104 105/* 106 * These are the constant used to fake the fixed-point load-average 107 * counting. Some notes: 108 * - 11 bit fractions expand to 22 bits by the multiplies: this gives 109 * a load-average precision of 10 bits integer + 11 bits fractional 110 * - if you want to count load-averages more often, you need more 111 * precision, or rounding will get you. With 2-second counting freq, 112 * the EXP_n values would be 1981, 2034 and 2043 if still using only 113 * 11 bit fractions. 114 */ 115extern unsigned long avenrun[]; /* Load averages */ 116 117#define FSHIFT 11 /* nr of bits of precision */ 118#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */ 119#define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */ 120#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */ 121#define EXP_5 2014 /* 1/exp(5sec/5min) */ 122#define EXP_15 2037 /* 1/exp(5sec/15min) */ 123 124#define CALC_LOAD(load,exp,n) \ 125 load *= exp; \ 126 load += n*(FIXED_1-exp); \ 127 load >>= FSHIFT; 128 129extern unsigned long total_forks; 130extern int nr_threads; 131DECLARE_PER_CPU(unsigned long, process_counts); 132extern int nr_processes(void); 133extern unsigned long nr_running(void); 134extern unsigned long nr_uninterruptible(void); 135extern unsigned long nr_active(void); 136extern unsigned long nr_iowait(void); 137 138struct seq_file; 139struct cfs_rq; 140struct task_group; 141#ifdef CONFIG_SCHED_DEBUG 142extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m); 143extern void proc_sched_set_task(struct task_struct *p); 144extern void 145print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq); 146#else 147static inline void 148proc_sched_show_task(struct task_struct *p, struct seq_file *m) 149{ 150} 151static inline void proc_sched_set_task(struct task_struct *p) 152{ 153} 154static inline void 155print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) 156{ 157} 158#endif 159 160extern unsigned long long time_sync_thresh; 161 162/* 163 * Task state bitmask. NOTE! These bits are also 164 * encoded in fs/proc/array.c: get_task_state(). 165 * 166 * We have two separate sets of flags: task->state 167 * is about runnability, while task->exit_state are 168 * about the task exiting. Confusing, but this way 169 * modifying one set can't modify the other one by 170 * mistake. 171 */ 172#define TASK_RUNNING 0 173#define TASK_INTERRUPTIBLE 1 174#define TASK_UNINTERRUPTIBLE 2 175#define __TASK_STOPPED 4 176#define __TASK_TRACED 8 177/* in tsk->exit_state */ 178#define EXIT_ZOMBIE 16 179#define EXIT_DEAD 32 180/* in tsk->state again */ 181#define TASK_DEAD 64 182#define TASK_WAKEKILL 128 183 184/* Convenience macros for the sake of set_task_state */ 185#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) 186#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) 187#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED) 188 189/* Convenience macros for the sake of wake_up */ 190#define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) 191#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED) 192 193/* get_task_state() */ 194#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ 195 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ 196 __TASK_TRACED) 197 198#define task_is_traced(task) ((task->state & __TASK_TRACED) != 0) 199#define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0) 200#define task_is_stopped_or_traced(task) \ 201 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0) 202#define task_contributes_to_load(task) \ 203 ((task->state & TASK_UNINTERRUPTIBLE) != 0) 204 205#define __set_task_state(tsk, state_value) \ 206 do { (tsk)->state = (state_value); } while (0) 207#define set_task_state(tsk, state_value) \ 208 set_mb((tsk)->state, (state_value)) 209 210/* 211 * set_current_state() includes a barrier so that the write of current->state 212 * is correctly serialised wrt the caller's subsequent test of whether to 213 * actually sleep: 214 * 215 * set_current_state(TASK_UNINTERRUPTIBLE); 216 * if (do_i_need_to_sleep()) 217 * schedule(); 218 * 219 * If the caller does not need such serialisation then use __set_current_state() 220 */ 221#define __set_current_state(state_value) \ 222 do { current->state = (state_value); } while (0) 223#define set_current_state(state_value) \ 224 set_mb(current->state, (state_value)) 225 226/* Task command name length */ 227#define TASK_COMM_LEN 16 228 229#include <linux/spinlock.h> 230 231/* 232 * This serializes "schedule()" and also protects 233 * the run-queue from deletions/modifications (but 234 * _adding_ to the beginning of the run-queue has 235 * a separate lock). 236 */ 237extern rwlock_t tasklist_lock; 238extern spinlock_t mmlist_lock; 239 240struct task_struct; 241 242extern void sched_init(void); 243extern void sched_init_smp(void); 244extern asmlinkage void schedule_tail(struct task_struct *prev); 245extern void init_idle(struct task_struct *idle, int cpu); 246extern void init_idle_bootup_task(struct task_struct *idle); 247 248extern int runqueue_is_locked(void); 249 250extern cpumask_t nohz_cpu_mask; 251#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ) 252extern int select_nohz_load_balancer(int cpu); 253#else 254static inline int select_nohz_load_balancer(int cpu) 255{ 256 return 0; 257} 258#endif 259 260extern unsigned long rt_needs_cpu(int cpu); 261 262/* 263 * Only dump TASK_* tasks. (0 for all tasks) 264 */ 265extern void show_state_filter(unsigned long state_filter); 266 267static inline void show_state(void) 268{ 269 show_state_filter(0); 270} 271 272extern void show_regs(struct pt_regs *); 273 274/* 275 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current 276 * task), SP is the stack pointer of the first frame that should be shown in the back 277 * trace (or NULL if the entire call-chain of the task should be shown). 278 */ 279extern void show_stack(struct task_struct *task, unsigned long *sp); 280 281void io_schedule(void); 282long io_schedule_timeout(long timeout); 283 284extern void cpu_init (void); 285extern void trap_init(void); 286extern void account_process_tick(struct task_struct *task, int user); 287extern void update_process_times(int user); 288extern void scheduler_tick(void); 289extern void hrtick_resched(void); 290 291extern void sched_show_task(struct task_struct *p); 292 293#ifdef CONFIG_DETECT_SOFTLOCKUP 294extern void softlockup_tick(void); 295extern void spawn_softlockup_task(void); 296extern void touch_softlockup_watchdog(void); 297extern void touch_all_softlockup_watchdogs(void); 298extern unsigned int softlockup_panic; 299extern unsigned long sysctl_hung_task_check_count; 300extern unsigned long sysctl_hung_task_timeout_secs; 301extern unsigned long sysctl_hung_task_warnings; 302extern int softlockup_thresh; 303#else 304static inline void softlockup_tick(void) 305{ 306} 307static inline void spawn_softlockup_task(void) 308{ 309} 310static inline void touch_softlockup_watchdog(void) 311{ 312} 313static inline void touch_all_softlockup_watchdogs(void) 314{ 315} 316#endif 317 318 319/* Attach to any functions which should be ignored in wchan output. */ 320#define __sched __attribute__((__section__(".sched.text"))) 321 322/* Linker adds these: start and end of __sched functions */ 323extern char __sched_text_start[], __sched_text_end[]; 324 325/* Is this address in the __sched functions? */ 326extern int in_sched_functions(unsigned long addr); 327 328#define MAX_SCHEDULE_TIMEOUT LONG_MAX 329extern signed long schedule_timeout(signed long timeout); 330extern signed long schedule_timeout_interruptible(signed long timeout); 331extern signed long schedule_timeout_killable(signed long timeout); 332extern signed long schedule_timeout_uninterruptible(signed long timeout); 333asmlinkage void schedule(void); 334 335struct nsproxy; 336struct user_namespace; 337 338/* Maximum number of active map areas.. This is a random (large) number */ 339#define DEFAULT_MAX_MAP_COUNT 65536 340 341extern int sysctl_max_map_count; 342 343#include <linux/aio.h> 344 345extern unsigned long 346arch_get_unmapped_area(struct file *, unsigned long, unsigned long, 347 unsigned long, unsigned long); 348extern unsigned long 349arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr, 350 unsigned long len, unsigned long pgoff, 351 unsigned long flags); 352extern void arch_unmap_area(struct mm_struct *, unsigned long); 353extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long); 354 355#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS 356/* 357 * The mm counters are not protected by its page_table_lock, 358 * so must be incremented atomically. 359 */ 360#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value) 361#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member)) 362#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member) 363#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member) 364#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member) 365 366#else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ 367/* 368 * The mm counters are protected by its page_table_lock, 369 * so can be incremented directly. 370 */ 371#define set_mm_counter(mm, member, value) (mm)->_##member = (value) 372#define get_mm_counter(mm, member) ((mm)->_##member) 373#define add_mm_counter(mm, member, value) (mm)->_##member += (value) 374#define inc_mm_counter(mm, member) (mm)->_##member++ 375#define dec_mm_counter(mm, member) (mm)->_##member-- 376 377#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ 378 379#define get_mm_rss(mm) \ 380 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss)) 381#define update_hiwater_rss(mm) do { \ 382 unsigned long _rss = get_mm_rss(mm); \ 383 if ((mm)->hiwater_rss < _rss) \ 384 (mm)->hiwater_rss = _rss; \ 385} while (0) 386#define update_hiwater_vm(mm) do { \ 387 if ((mm)->hiwater_vm < (mm)->total_vm) \ 388 (mm)->hiwater_vm = (mm)->total_vm; \ 389} while (0) 390 391extern void set_dumpable(struct mm_struct *mm, int value); 392extern int get_dumpable(struct mm_struct *mm); 393 394/* mm flags */ 395/* dumpable bits */ 396#define MMF_DUMPABLE 0 /* core dump is permitted */ 397#define MMF_DUMP_SECURELY 1 /* core file is readable only by root */ 398#define MMF_DUMPABLE_BITS 2 399 400/* coredump filter bits */ 401#define MMF_DUMP_ANON_PRIVATE 2 402#define MMF_DUMP_ANON_SHARED 3 403#define MMF_DUMP_MAPPED_PRIVATE 4 404#define MMF_DUMP_MAPPED_SHARED 5 405#define MMF_DUMP_ELF_HEADERS 6 406#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS 407#define MMF_DUMP_FILTER_BITS 5 408#define MMF_DUMP_FILTER_MASK \ 409 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT) 410#define MMF_DUMP_FILTER_DEFAULT \ 411 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED)) 412 413struct sighand_struct { 414 atomic_t count; 415 struct k_sigaction action[_NSIG]; 416 spinlock_t siglock; 417 wait_queue_head_t signalfd_wqh; 418}; 419 420struct pacct_struct { 421 int ac_flag; 422 long ac_exitcode; 423 unsigned long ac_mem; 424 cputime_t ac_utime, ac_stime; 425 unsigned long ac_minflt, ac_majflt; 426}; 427 428/* 429 * NOTE! "signal_struct" does not have it's own 430 * locking, because a shared signal_struct always 431 * implies a shared sighand_struct, so locking 432 * sighand_struct is always a proper superset of 433 * the locking of signal_struct. 434 */ 435struct signal_struct { 436 atomic_t count; 437 atomic_t live; 438 439 wait_queue_head_t wait_chldexit; /* for wait4() */ 440 441 /* current thread group signal load-balancing target: */ 442 struct task_struct *curr_target; 443 444 /* shared signal handling: */ 445 struct sigpending shared_pending; 446 447 /* thread group exit support */ 448 int group_exit_code; 449 /* overloaded: 450 * - notify group_exit_task when ->count is equal to notify_count 451 * - everyone except group_exit_task is stopped during signal delivery 452 * of fatal signals, group_exit_task processes the signal. 453 */ 454 struct task_struct *group_exit_task; 455 int notify_count; 456 457 /* thread group stop support, overloads group_exit_code too */ 458 int group_stop_count; 459 unsigned int flags; /* see SIGNAL_* flags below */ 460 461 /* POSIX.1b Interval Timers */ 462 struct list_head posix_timers; 463 464 /* ITIMER_REAL timer for the process */ 465 struct hrtimer real_timer; 466 struct pid *leader_pid; 467 ktime_t it_real_incr; 468 469 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */ 470 cputime_t it_prof_expires, it_virt_expires; 471 cputime_t it_prof_incr, it_virt_incr; 472 473 /* job control IDs */ 474 475 /* 476 * pgrp and session fields are deprecated. 477 * use the task_session_Xnr and task_pgrp_Xnr routines below 478 */ 479 480 union { 481 pid_t pgrp __deprecated; 482 pid_t __pgrp; 483 }; 484 485 struct pid *tty_old_pgrp; 486 487 union { 488 pid_t session __deprecated; 489 pid_t __session; 490 }; 491 492 /* boolean value for session group leader */ 493 int leader; 494 495 struct tty_struct *tty; /* NULL if no tty */ 496 497 /* 498 * Cumulative resource counters for dead threads in the group, 499 * and for reaped dead child processes forked by this group. 500 * Live threads maintain their own counters and add to these 501 * in __exit_signal, except for the group leader. 502 */ 503 cputime_t utime, stime, cutime, cstime; 504 cputime_t gtime; 505 cputime_t cgtime; 506 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw; 507 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt; 508 unsigned long inblock, oublock, cinblock, coublock; 509 510 /* 511 * Cumulative ns of scheduled CPU time for dead threads in the 512 * group, not including a zombie group leader. (This only differs 513 * from jiffies_to_ns(utime + stime) if sched_clock uses something 514 * other than jiffies.) 515 */ 516 unsigned long long sum_sched_runtime; 517 518 /* 519 * We don't bother to synchronize most readers of this at all, 520 * because there is no reader checking a limit that actually needs 521 * to get both rlim_cur and rlim_max atomically, and either one 522 * alone is a single word that can safely be read normally. 523 * getrlimit/setrlimit use task_lock(current->group_leader) to 524 * protect this instead of the siglock, because they really 525 * have no need to disable irqs. 526 */ 527 struct rlimit rlim[RLIM_NLIMITS]; 528 529 struct list_head cpu_timers[3]; 530 531 /* keep the process-shared keyrings here so that they do the right 532 * thing in threads created with CLONE_THREAD */ 533#ifdef CONFIG_KEYS 534 struct key *session_keyring; /* keyring inherited over fork */ 535 struct key *process_keyring; /* keyring private to this process */ 536#endif 537#ifdef CONFIG_BSD_PROCESS_ACCT 538 struct pacct_struct pacct; /* per-process accounting information */ 539#endif 540#ifdef CONFIG_TASKSTATS 541 struct taskstats *stats; 542#endif 543#ifdef CONFIG_AUDIT 544 unsigned audit_tty; 545 struct tty_audit_buf *tty_audit_buf; 546#endif 547}; 548 549/* Context switch must be unlocked if interrupts are to be enabled */ 550#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW 551# define __ARCH_WANT_UNLOCKED_CTXSW 552#endif 553 554/* 555 * Bits in flags field of signal_struct. 556 */ 557#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */ 558#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */ 559#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */ 560#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */ 561/* 562 * Pending notifications to parent. 563 */ 564#define SIGNAL_CLD_STOPPED 0x00000010 565#define SIGNAL_CLD_CONTINUED 0x00000020 566#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED) 567 568#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */ 569 570/* If true, all threads except ->group_exit_task have pending SIGKILL */ 571static inline int signal_group_exit(const struct signal_struct *sig) 572{ 573 return (sig->flags & SIGNAL_GROUP_EXIT) || 574 (sig->group_exit_task != NULL); 575} 576 577/* 578 * Some day this will be a full-fledged user tracking system.. 579 */ 580struct user_struct { 581 atomic_t __count; /* reference count */ 582 atomic_t processes; /* How many processes does this user have? */ 583 atomic_t files; /* How many open files does this user have? */ 584 atomic_t sigpending; /* How many pending signals does this user have? */ 585#ifdef CONFIG_INOTIFY_USER 586 atomic_t inotify_watches; /* How many inotify watches does this user have? */ 587 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */ 588#endif 589#ifdef CONFIG_POSIX_MQUEUE 590 /* protected by mq_lock */ 591 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */ 592#endif 593 unsigned long locked_shm; /* How many pages of mlocked shm ? */ 594 595#ifdef CONFIG_KEYS 596 struct key *uid_keyring; /* UID specific keyring */ 597 struct key *session_keyring; /* UID's default session keyring */ 598#endif 599 600 /* Hash table maintenance information */ 601 struct hlist_node uidhash_node; 602 uid_t uid; 603 604#ifdef CONFIG_USER_SCHED 605 struct task_group *tg; 606#ifdef CONFIG_SYSFS 607 struct kobject kobj; 608 struct work_struct work; 609#endif 610#endif 611}; 612 613extern int uids_sysfs_init(void); 614 615extern struct user_struct *find_user(uid_t); 616 617extern struct user_struct root_user; 618#define INIT_USER (&root_user) 619 620struct backing_dev_info; 621struct reclaim_state; 622 623#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 624struct sched_info { 625 /* cumulative counters */ 626 unsigned long pcount; /* # of times run on this cpu */ 627 unsigned long long cpu_time, /* time spent on the cpu */ 628 run_delay; /* time spent waiting on a runqueue */ 629 630 /* timestamps */ 631 unsigned long long last_arrival,/* when we last ran on a cpu */ 632 last_queued; /* when we were last queued to run */ 633#ifdef CONFIG_SCHEDSTATS 634 /* BKL stats */ 635 unsigned int bkl_count; 636#endif 637}; 638#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */ 639 640#ifdef CONFIG_SCHEDSTATS 641extern const struct file_operations proc_schedstat_operations; 642#endif /* CONFIG_SCHEDSTATS */ 643 644#ifdef CONFIG_TASK_DELAY_ACCT 645struct task_delay_info { 646 spinlock_t lock; 647 unsigned int flags; /* Private per-task flags */ 648 649 /* For each stat XXX, add following, aligned appropriately 650 * 651 * struct timespec XXX_start, XXX_end; 652 * u64 XXX_delay; 653 * u32 XXX_count; 654 * 655 * Atomicity of updates to XXX_delay, XXX_count protected by 656 * single lock above (split into XXX_lock if contention is an issue). 657 */ 658 659 /* 660 * XXX_count is incremented on every XXX operation, the delay 661 * associated with the operation is added to XXX_delay. 662 * XXX_delay contains the accumulated delay time in nanoseconds. 663 */ 664 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */ 665 u64 blkio_delay; /* wait for sync block io completion */ 666 u64 swapin_delay; /* wait for swapin block io completion */ 667 u32 blkio_count; /* total count of the number of sync block */ 668 /* io operations performed */ 669 u32 swapin_count; /* total count of the number of swapin block */ 670 /* io operations performed */ 671}; 672#endif /* CONFIG_TASK_DELAY_ACCT */ 673 674static inline int sched_info_on(void) 675{ 676#ifdef CONFIG_SCHEDSTATS 677 return 1; 678#elif defined(CONFIG_TASK_DELAY_ACCT) 679 extern int delayacct_on; 680 return delayacct_on; 681#else 682 return 0; 683#endif 684} 685 686enum cpu_idle_type { 687 CPU_IDLE, 688 CPU_NOT_IDLE, 689 CPU_NEWLY_IDLE, 690 CPU_MAX_IDLE_TYPES 691}; 692 693/* 694 * sched-domains (multiprocessor balancing) declarations: 695 */ 696 697/* 698 * Increase resolution of nice-level calculations: 699 */ 700#define SCHED_LOAD_SHIFT 10 701#define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT) 702 703#define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE 704 705#ifdef CONFIG_SMP 706#define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */ 707#define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */ 708#define SD_BALANCE_EXEC 4 /* Balance on exec */ 709#define SD_BALANCE_FORK 8 /* Balance on fork, clone */ 710#define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */ 711#define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */ 712#define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */ 713#define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */ 714#define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */ 715#define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */ 716#define SD_SERIALIZE 1024 /* Only a single load balancing instance */ 717#define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */ 718 719#define BALANCE_FOR_MC_POWER \ 720 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0) 721 722#define BALANCE_FOR_PKG_POWER \ 723 ((sched_mc_power_savings || sched_smt_power_savings) ? \ 724 SD_POWERSAVINGS_BALANCE : 0) 725 726#define test_sd_parent(sd, flag) ((sd->parent && \ 727 (sd->parent->flags & flag)) ? 1 : 0) 728 729 730struct sched_group { 731 struct sched_group *next; /* Must be a circular list */ 732 cpumask_t cpumask; 733 734 /* 735 * CPU power of this group, SCHED_LOAD_SCALE being max power for a 736 * single CPU. This is read only (except for setup, hotplug CPU). 737 * Note : Never change cpu_power without recompute its reciprocal 738 */ 739 unsigned int __cpu_power; 740 /* 741 * reciprocal value of cpu_power to avoid expensive divides 742 * (see include/linux/reciprocal_div.h) 743 */ 744 u32 reciprocal_cpu_power; 745}; 746 747enum sched_domain_level { 748 SD_LV_NONE = 0, 749 SD_LV_SIBLING, 750 SD_LV_MC, 751 SD_LV_CPU, 752 SD_LV_NODE, 753 SD_LV_ALLNODES, 754 SD_LV_MAX 755}; 756 757struct sched_domain_attr { 758 int relax_domain_level; 759}; 760 761#define SD_ATTR_INIT (struct sched_domain_attr) { \ 762 .relax_domain_level = -1, \ 763} 764 765struct sched_domain { 766 /* These fields must be setup */ 767 struct sched_domain *parent; /* top domain must be null terminated */ 768 struct sched_domain *child; /* bottom domain must be null terminated */ 769 struct sched_group *groups; /* the balancing groups of the domain */ 770 cpumask_t span; /* span of all CPUs in this domain */ 771 unsigned long min_interval; /* Minimum balance interval ms */ 772 unsigned long max_interval; /* Maximum balance interval ms */ 773 unsigned int busy_factor; /* less balancing by factor if busy */ 774 unsigned int imbalance_pct; /* No balance until over watermark */ 775 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */ 776 unsigned int busy_idx; 777 unsigned int idle_idx; 778 unsigned int newidle_idx; 779 unsigned int wake_idx; 780 unsigned int forkexec_idx; 781 int flags; /* See SD_* */ 782 enum sched_domain_level level; 783 784 /* Runtime fields. */ 785 unsigned long last_balance; /* init to jiffies. units in jiffies */ 786 unsigned int balance_interval; /* initialise to 1. units in ms. */ 787 unsigned int nr_balance_failed; /* initialise to 0 */ 788 789 u64 last_update; 790 791#ifdef CONFIG_SCHEDSTATS 792 /* load_balance() stats */ 793 unsigned int lb_count[CPU_MAX_IDLE_TYPES]; 794 unsigned int lb_failed[CPU_MAX_IDLE_TYPES]; 795 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES]; 796 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES]; 797 unsigned int lb_gained[CPU_MAX_IDLE_TYPES]; 798 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES]; 799 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES]; 800 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES]; 801 802 /* Active load balancing */ 803 unsigned int alb_count; 804 unsigned int alb_failed; 805 unsigned int alb_pushed; 806 807 /* SD_BALANCE_EXEC stats */ 808 unsigned int sbe_count; 809 unsigned int sbe_balanced; 810 unsigned int sbe_pushed; 811 812 /* SD_BALANCE_FORK stats */ 813 unsigned int sbf_count; 814 unsigned int sbf_balanced; 815 unsigned int sbf_pushed; 816 817 /* try_to_wake_up() stats */ 818 unsigned int ttwu_wake_remote; 819 unsigned int ttwu_move_affine; 820 unsigned int ttwu_move_balance; 821#endif 822}; 823 824extern void partition_sched_domains(int ndoms_new, cpumask_t *doms_new, 825 struct sched_domain_attr *dattr_new); 826extern int arch_reinit_sched_domains(void); 827 828#else /* CONFIG_SMP */ 829 830struct sched_domain_attr; 831 832static inline void 833partition_sched_domains(int ndoms_new, cpumask_t *doms_new, 834 struct sched_domain_attr *dattr_new) 835{ 836} 837#endif /* !CONFIG_SMP */ 838 839struct io_context; /* See blkdev.h */ 840#define NGROUPS_SMALL 32 841#define NGROUPS_PER_BLOCK ((unsigned int)(PAGE_SIZE / sizeof(gid_t))) 842struct group_info { 843 int ngroups; 844 atomic_t usage; 845 gid_t small_block[NGROUPS_SMALL]; 846 int nblocks; 847 gid_t *blocks[0]; 848}; 849 850/* 851 * get_group_info() must be called with the owning task locked (via task_lock()) 852 * when task != current. The reason being that the vast majority of callers are 853 * looking at current->group_info, which can not be changed except by the 854 * current task. Changing current->group_info requires the task lock, too. 855 */ 856#define get_group_info(group_info) do { \ 857 atomic_inc(&(group_info)->usage); \ 858} while (0) 859 860#define put_group_info(group_info) do { \ 861 if (atomic_dec_and_test(&(group_info)->usage)) \ 862 groups_free(group_info); \ 863} while (0) 864 865extern struct group_info *groups_alloc(int gidsetsize); 866extern void groups_free(struct group_info *group_info); 867extern int set_current_groups(struct group_info *group_info); 868extern int groups_search(struct group_info *group_info, gid_t grp); 869/* access the groups "array" with this macro */ 870#define GROUP_AT(gi, i) \ 871 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK]) 872 873#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK 874extern void prefetch_stack(struct task_struct *t); 875#else 876static inline void prefetch_stack(struct task_struct *t) { } 877#endif 878 879struct audit_context; /* See audit.c */ 880struct mempolicy; 881struct pipe_inode_info; 882struct uts_namespace; 883 884struct rq; 885struct sched_domain; 886 887struct sched_class { 888 const struct sched_class *next; 889 890 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup); 891 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep); 892 void (*yield_task) (struct rq *rq); 893 int (*select_task_rq)(struct task_struct *p, int sync); 894 895 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p); 896 897 struct task_struct * (*pick_next_task) (struct rq *rq); 898 void (*put_prev_task) (struct rq *rq, struct task_struct *p); 899 900#ifdef CONFIG_SMP 901 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu, 902 struct rq *busiest, unsigned long max_load_move, 903 struct sched_domain *sd, enum cpu_idle_type idle, 904 int *all_pinned, int *this_best_prio); 905 906 int (*move_one_task) (struct rq *this_rq, int this_cpu, 907 struct rq *busiest, struct sched_domain *sd, 908 enum cpu_idle_type idle); 909 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task); 910 void (*post_schedule) (struct rq *this_rq); 911 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task); 912#endif 913 914 void (*set_curr_task) (struct rq *rq); 915 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued); 916 void (*task_new) (struct rq *rq, struct task_struct *p); 917 void (*set_cpus_allowed)(struct task_struct *p, 918 const cpumask_t *newmask); 919 920 void (*rq_online)(struct rq *rq); 921 void (*rq_offline)(struct rq *rq); 922 923 void (*switched_from) (struct rq *this_rq, struct task_struct *task, 924 int running); 925 void (*switched_to) (struct rq *this_rq, struct task_struct *task, 926 int running); 927 void (*prio_changed) (struct rq *this_rq, struct task_struct *task, 928 int oldprio, int running); 929 930#ifdef CONFIG_FAIR_GROUP_SCHED 931 void (*moved_group) (struct task_struct *p); 932#endif 933}; 934 935struct load_weight { 936 unsigned long weight, inv_weight; 937}; 938 939/* 940 * CFS stats for a schedulable entity (task, task-group etc) 941 * 942 * Current field usage histogram: 943 * 944 * 4 se->block_start 945 * 4 se->run_node 946 * 4 se->sleep_start 947 * 6 se->load.weight 948 */ 949struct sched_entity { 950 struct load_weight load; /* for load-balancing */ 951 struct rb_node run_node; 952 struct list_head group_node; 953 unsigned int on_rq; 954 955 u64 exec_start; 956 u64 sum_exec_runtime; 957 u64 vruntime; 958 u64 prev_sum_exec_runtime; 959 960 u64 last_wakeup; 961 u64 avg_overlap; 962 963#ifdef CONFIG_SCHEDSTATS 964 u64 wait_start; 965 u64 wait_max; 966 u64 wait_count; 967 u64 wait_sum; 968 969 u64 sleep_start; 970 u64 sleep_max; 971 s64 sum_sleep_runtime; 972 973 u64 block_start; 974 u64 block_max; 975 u64 exec_max; 976 u64 slice_max; 977 978 u64 nr_migrations; 979 u64 nr_migrations_cold; 980 u64 nr_failed_migrations_affine; 981 u64 nr_failed_migrations_running; 982 u64 nr_failed_migrations_hot; 983 u64 nr_forced_migrations; 984 u64 nr_forced2_migrations; 985 986 u64 nr_wakeups; 987 u64 nr_wakeups_sync; 988 u64 nr_wakeups_migrate; 989 u64 nr_wakeups_local; 990 u64 nr_wakeups_remote; 991 u64 nr_wakeups_affine; 992 u64 nr_wakeups_affine_attempts; 993 u64 nr_wakeups_passive; 994 u64 nr_wakeups_idle; 995#endif 996 997#ifdef CONFIG_FAIR_GROUP_SCHED 998 struct sched_entity *parent; 999 /* rq on which this entity is (to be) queued: */ 1000 struct cfs_rq *cfs_rq; 1001 /* rq "owned" by this entity/group: */ 1002 struct cfs_rq *my_q; 1003#endif 1004}; 1005 1006struct sched_rt_entity { 1007 struct list_head run_list; 1008 unsigned int time_slice; 1009 unsigned long timeout; 1010 int nr_cpus_allowed; 1011 1012 struct sched_rt_entity *back; 1013#ifdef CONFIG_RT_GROUP_SCHED 1014 struct sched_rt_entity *parent; 1015 /* rq on which this entity is (to be) queued: */ 1016 struct rt_rq *rt_rq; 1017 /* rq "owned" by this entity/group: */ 1018 struct rt_rq *my_q; 1019#endif 1020}; 1021 1022struct task_struct { 1023 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */ 1024 void *stack; 1025 atomic_t usage; 1026 unsigned int flags; /* per process flags, defined below */ 1027 unsigned int ptrace; 1028 1029 int lock_depth; /* BKL lock depth */ 1030 1031#ifdef CONFIG_SMP 1032#ifdef __ARCH_WANT_UNLOCKED_CTXSW 1033 int oncpu; 1034#endif 1035#endif 1036 1037 int prio, static_prio, normal_prio; 1038 unsigned int rt_priority; 1039 const struct sched_class *sched_class; 1040 struct sched_entity se; 1041 struct sched_rt_entity rt; 1042 1043#ifdef CONFIG_PREEMPT_NOTIFIERS 1044 /* list of struct preempt_notifier: */ 1045 struct hlist_head preempt_notifiers; 1046#endif 1047 1048 /* 1049 * fpu_counter contains the number of consecutive context switches 1050 * that the FPU is used. If this is over a threshold, the lazy fpu 1051 * saving becomes unlazy to save the trap. This is an unsigned char 1052 * so that after 256 times the counter wraps and the behavior turns 1053 * lazy again; this to deal with bursty apps that only use FPU for 1054 * a short time 1055 */ 1056 unsigned char fpu_counter; 1057 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */ 1058#ifdef CONFIG_BLK_DEV_IO_TRACE 1059 unsigned int btrace_seq; 1060#endif 1061 1062 unsigned int policy; 1063 cpumask_t cpus_allowed; 1064 1065#ifdef CONFIG_PREEMPT_RCU 1066 int rcu_read_lock_nesting; 1067 int rcu_flipctr_idx; 1068#endif /* #ifdef CONFIG_PREEMPT_RCU */ 1069 1070#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 1071 struct sched_info sched_info; 1072#endif 1073 1074 struct list_head tasks; 1075 1076 struct mm_struct *mm, *active_mm; 1077 1078/* task state */ 1079 struct linux_binfmt *binfmt; 1080 int exit_state; 1081 int exit_code, exit_signal; 1082 int pdeath_signal; /* The signal sent when the parent dies */ 1083 /* ??? */ 1084 unsigned int personality; 1085 unsigned did_exec:1; 1086 pid_t pid; 1087 pid_t tgid; 1088 1089#ifdef CONFIG_CC_STACKPROTECTOR 1090 /* Canary value for the -fstack-protector gcc feature */ 1091 unsigned long stack_canary; 1092#endif 1093 /* 1094 * pointers to (original) parent process, youngest child, younger sibling, 1095 * older sibling, respectively. (p->father can be replaced with 1096 * p->real_parent->pid) 1097 */ 1098 struct task_struct *real_parent; /* real parent process */ 1099 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */ 1100 /* 1101 * children/sibling forms the list of my natural children 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 /* 1108 * ptraced is the list of tasks this task is using ptrace on. 1109 * This includes both natural children and PTRACE_ATTACH targets. 1110 * p->ptrace_entry is p's link on the p->parent->ptraced list. 1111 */ 1112 struct list_head ptraced; 1113 struct list_head ptrace_entry; 1114 1115 /* PID/PID hash table linkage. */ 1116 struct pid_link pids[PIDTYPE_MAX]; 1117 struct list_head thread_group; 1118 1119 struct completion *vfork_done; /* for vfork() */ 1120 int __user *set_child_tid; /* CLONE_CHILD_SETTID */ 1121 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ 1122 1123 cputime_t utime, stime, utimescaled, stimescaled; 1124 cputime_t gtime; 1125 cputime_t prev_utime, prev_stime; 1126 unsigned long nvcsw, nivcsw; /* context switch counts */ 1127 struct timespec start_time; /* monotonic time */ 1128 struct timespec real_start_time; /* boot based time */ 1129/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ 1130 unsigned long min_flt, maj_flt; 1131 1132 cputime_t it_prof_expires, it_virt_expires; 1133 unsigned long long it_sched_expires; 1134 struct list_head cpu_timers[3]; 1135 1136/* process credentials */ 1137 uid_t uid,euid,suid,fsuid; 1138 gid_t gid,egid,sgid,fsgid; 1139 struct group_info *group_info; 1140 kernel_cap_t cap_effective, cap_inheritable, cap_permitted, cap_bset; 1141 struct user_struct *user; 1142 unsigned securebits; 1143#ifdef CONFIG_KEYS 1144 unsigned char jit_keyring; /* default keyring to attach requested keys to */ 1145 struct key *request_key_auth; /* assumed request_key authority */ 1146 struct key *thread_keyring; /* keyring private to this thread */ 1147#endif 1148 char comm[TASK_COMM_LEN]; /* executable name excluding path 1149 - access with [gs]et_task_comm (which lock 1150 it with task_lock()) 1151 - initialized normally by flush_old_exec */ 1152/* file system info */ 1153 int link_count, total_link_count; 1154#ifdef CONFIG_SYSVIPC 1155/* ipc stuff */ 1156 struct sysv_sem sysvsem; 1157#endif 1158#ifdef CONFIG_DETECT_SOFTLOCKUP 1159/* hung task detection */ 1160 unsigned long last_switch_timestamp; 1161 unsigned long last_switch_count; 1162#endif 1163/* CPU-specific state of this task */ 1164 struct thread_struct thread; 1165/* filesystem information */ 1166 struct fs_struct *fs; 1167/* open file information */ 1168 struct files_struct *files; 1169/* namespaces */ 1170 struct nsproxy *nsproxy; 1171/* signal handlers */ 1172 struct signal_struct *signal; 1173 struct sighand_struct *sighand; 1174 1175 sigset_t blocked, real_blocked; 1176 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */ 1177 struct sigpending pending; 1178 1179 unsigned long sas_ss_sp; 1180 size_t sas_ss_size; 1181 int (*notifier)(void *priv); 1182 void *notifier_data; 1183 sigset_t *notifier_mask; 1184#ifdef CONFIG_SECURITY 1185 void *security; 1186#endif 1187 struct audit_context *audit_context; 1188#ifdef CONFIG_AUDITSYSCALL 1189 uid_t loginuid; 1190 unsigned int sessionid; 1191#endif 1192 seccomp_t seccomp; 1193 1194/* Thread group tracking */ 1195 u32 parent_exec_id; 1196 u32 self_exec_id; 1197/* Protection of (de-)allocation: mm, files, fs, tty, keyrings */ 1198 spinlock_t alloc_lock; 1199 1200 /* Protection of the PI data structures: */ 1201 spinlock_t pi_lock; 1202 1203#ifdef CONFIG_RT_MUTEXES 1204 /* PI waiters blocked on a rt_mutex held by this task */ 1205 struct plist_head pi_waiters; 1206 /* Deadlock detection and priority inheritance handling */ 1207 struct rt_mutex_waiter *pi_blocked_on; 1208#endif 1209 1210#ifdef CONFIG_DEBUG_MUTEXES 1211 /* mutex deadlock detection */ 1212 struct mutex_waiter *blocked_on; 1213#endif 1214#ifdef CONFIG_TRACE_IRQFLAGS 1215 unsigned int irq_events; 1216 int hardirqs_enabled; 1217 unsigned long hardirq_enable_ip; 1218 unsigned int hardirq_enable_event; 1219 unsigned long hardirq_disable_ip; 1220 unsigned int hardirq_disable_event; 1221 int softirqs_enabled; 1222 unsigned long softirq_disable_ip; 1223 unsigned int softirq_disable_event; 1224 unsigned long softirq_enable_ip; 1225 unsigned int softirq_enable_event; 1226 int hardirq_context; 1227 int softirq_context; 1228#endif 1229#ifdef CONFIG_LOCKDEP 1230# define MAX_LOCK_DEPTH 48UL 1231 u64 curr_chain_key; 1232 int lockdep_depth; 1233 unsigned int lockdep_recursion; 1234 struct held_lock held_locks[MAX_LOCK_DEPTH]; 1235#endif 1236 1237/* journalling filesystem info */ 1238 void *journal_info; 1239 1240/* stacked block device info */ 1241 struct bio *bio_list, **bio_tail; 1242 1243/* VM state */ 1244 struct reclaim_state *reclaim_state; 1245 1246 struct backing_dev_info *backing_dev_info; 1247 1248 struct io_context *io_context; 1249 1250 unsigned long ptrace_message; 1251 siginfo_t *last_siginfo; /* For ptrace use. */ 1252#ifdef CONFIG_TASK_XACCT 1253/* i/o counters(bytes read/written, #syscalls */ 1254 u64 rchar, wchar, syscr, syscw; 1255#endif 1256 struct task_io_accounting ioac; 1257#if defined(CONFIG_TASK_XACCT) 1258 u64 acct_rss_mem1; /* accumulated rss usage */ 1259 u64 acct_vm_mem1; /* accumulated virtual memory usage */ 1260 cputime_t acct_stimexpd;/* stime since last update */ 1261#endif 1262#ifdef CONFIG_CPUSETS 1263 nodemask_t mems_allowed; 1264 int cpuset_mems_generation; 1265 int cpuset_mem_spread_rotor; 1266#endif 1267#ifdef CONFIG_CGROUPS 1268 /* Control Group info protected by css_set_lock */ 1269 struct css_set *cgroups; 1270 /* cg_list protected by css_set_lock and tsk->alloc_lock */ 1271 struct list_head cg_list; 1272#endif 1273#ifdef CONFIG_FUTEX 1274 struct robust_list_head __user *robust_list; 1275#ifdef CONFIG_COMPAT 1276 struct compat_robust_list_head __user *compat_robust_list; 1277#endif 1278 struct list_head pi_state_list; 1279 struct futex_pi_state *pi_state_cache; 1280#endif 1281#ifdef CONFIG_NUMA 1282 struct mempolicy *mempolicy; 1283 short il_next; 1284#endif 1285 atomic_t fs_excl; /* holding fs exclusive resources */ 1286 struct rcu_head rcu; 1287 1288 /* 1289 * cache last used pipe for splice 1290 */ 1291 struct pipe_inode_info *splice_pipe; 1292#ifdef CONFIG_TASK_DELAY_ACCT 1293 struct task_delay_info *delays; 1294#endif 1295#ifdef CONFIG_FAULT_INJECTION 1296 int make_it_fail; 1297#endif 1298 struct prop_local_single dirties; 1299#ifdef CONFIG_LATENCYTOP 1300 int latency_record_count; 1301 struct latency_record latency_record[LT_SAVECOUNT]; 1302#endif 1303}; 1304 1305/* 1306 * Priority of a process goes from 0..MAX_PRIO-1, valid RT 1307 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH 1308 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority 1309 * values are inverted: lower p->prio value means higher priority. 1310 * 1311 * The MAX_USER_RT_PRIO value allows the actual maximum 1312 * RT priority to be separate from the value exported to 1313 * user-space. This allows kernel threads to set their 1314 * priority to a value higher than any user task. Note: 1315 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO. 1316 */ 1317 1318#define MAX_USER_RT_PRIO 100 1319#define MAX_RT_PRIO MAX_USER_RT_PRIO 1320 1321#define MAX_PRIO (MAX_RT_PRIO + 40) 1322#define DEFAULT_PRIO (MAX_RT_PRIO + 20) 1323 1324static inline int rt_prio(int prio) 1325{ 1326 if (unlikely(prio < MAX_RT_PRIO)) 1327 return 1; 1328 return 0; 1329} 1330 1331static inline int rt_task(struct task_struct *p) 1332{ 1333 return rt_prio(p->prio); 1334} 1335 1336static inline void set_task_session(struct task_struct *tsk, pid_t session) 1337{ 1338 tsk->signal->__session = session; 1339} 1340 1341static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp) 1342{ 1343 tsk->signal->__pgrp = pgrp; 1344} 1345 1346static inline struct pid *task_pid(struct task_struct *task) 1347{ 1348 return task->pids[PIDTYPE_PID].pid; 1349} 1350 1351static inline struct pid *task_tgid(struct task_struct *task) 1352{ 1353 return task->group_leader->pids[PIDTYPE_PID].pid; 1354} 1355 1356static inline struct pid *task_pgrp(struct task_struct *task) 1357{ 1358 return task->group_leader->pids[PIDTYPE_PGID].pid; 1359} 1360 1361static inline struct pid *task_session(struct task_struct *task) 1362{ 1363 return task->group_leader->pids[PIDTYPE_SID].pid; 1364} 1365 1366struct pid_namespace; 1367 1368/* 1369 * the helpers to get the task's different pids as they are seen 1370 * from various namespaces 1371 * 1372 * task_xid_nr() : global id, i.e. the id seen from the init namespace; 1373 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of 1374 * current. 1375 * task_xid_nr_ns() : id seen from the ns specified; 1376 * 1377 * set_task_vxid() : assigns a virtual id to a task; 1378 * 1379 * see also pid_nr() etc in include/linux/pid.h 1380 */ 1381 1382static inline pid_t task_pid_nr(struct task_struct *tsk) 1383{ 1384 return tsk->pid; 1385} 1386 1387pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1388 1389static inline pid_t task_pid_vnr(struct task_struct *tsk) 1390{ 1391 return pid_vnr(task_pid(tsk)); 1392} 1393 1394 1395static inline pid_t task_tgid_nr(struct task_struct *tsk) 1396{ 1397 return tsk->tgid; 1398} 1399 1400pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1401 1402static inline pid_t task_tgid_vnr(struct task_struct *tsk) 1403{ 1404 return pid_vnr(task_tgid(tsk)); 1405} 1406 1407 1408static inline pid_t task_pgrp_nr(struct task_struct *tsk) 1409{ 1410 return tsk->signal->__pgrp; 1411} 1412 1413pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1414 1415static inline pid_t task_pgrp_vnr(struct task_struct *tsk) 1416{ 1417 return pid_vnr(task_pgrp(tsk)); 1418} 1419 1420 1421static inline pid_t task_session_nr(struct task_struct *tsk) 1422{ 1423 return tsk->signal->__session; 1424} 1425 1426pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1427 1428static inline pid_t task_session_vnr(struct task_struct *tsk) 1429{ 1430 return pid_vnr(task_session(tsk)); 1431} 1432 1433 1434/** 1435 * pid_alive - check that a task structure is not stale 1436 * @p: Task structure to be checked. 1437 * 1438 * Test if a process is not yet dead (at most zombie state) 1439 * If pid_alive fails, then pointers within the task structure 1440 * can be stale and must not be dereferenced. 1441 */ 1442static inline int pid_alive(struct task_struct *p) 1443{ 1444 return p->pids[PIDTYPE_PID].pid != NULL; 1445} 1446 1447/** 1448 * is_global_init - check if a task structure is init 1449 * @tsk: Task structure to be checked. 1450 * 1451 * Check if a task structure is the first user space task the kernel created. 1452 */ 1453static inline int is_global_init(struct task_struct *tsk) 1454{ 1455 return tsk->pid == 1; 1456} 1457 1458/* 1459 * is_container_init: 1460 * check whether in the task is init in its own pid namespace. 1461 */ 1462extern int is_container_init(struct task_struct *tsk); 1463 1464extern struct pid *cad_pid; 1465 1466extern void free_task(struct task_struct *tsk); 1467#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0) 1468 1469extern void __put_task_struct(struct task_struct *t); 1470 1471static inline void put_task_struct(struct task_struct *t) 1472{ 1473 if (atomic_dec_and_test(&t->usage)) 1474 __put_task_struct(t); 1475} 1476 1477/* 1478 * Per process flags 1479 */ 1480#define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */ 1481 /* Not implemented yet, only for 486*/ 1482#define PF_STARTING 0x00000002 /* being created */ 1483#define PF_EXITING 0x00000004 /* getting shut down */ 1484#define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */ 1485#define PF_VCPU 0x00000010 /* I'm a virtual CPU */ 1486#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ 1487#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ 1488#define PF_DUMPCORE 0x00000200 /* dumped core */ 1489#define PF_SIGNALED 0x00000400 /* killed by a signal */ 1490#define PF_MEMALLOC 0x00000800 /* Allocating memory */ 1491#define PF_FLUSHER 0x00001000 /* responsible for disk writeback */ 1492#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */ 1493#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */ 1494#define PF_FROZEN 0x00010000 /* frozen for system suspend */ 1495#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ 1496#define PF_KSWAPD 0x00040000 /* I am kswapd */ 1497#define PF_SWAPOFF 0x00080000 /* I am in swapoff */ 1498#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ 1499#define PF_KTHREAD 0x00200000 /* I am a kernel thread */ 1500#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */ 1501#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ 1502#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */ 1503#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */ 1504#define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */ 1505#define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */ 1506#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ 1507#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */ 1508#define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */ 1509 1510/* 1511 * Only the _current_ task can read/write to tsk->flags, but other 1512 * tasks can access tsk->flags in readonly mode for example 1513 * with tsk_used_math (like during threaded core dumping). 1514 * There is however an exception to this rule during ptrace 1515 * or during fork: the ptracer task is allowed to write to the 1516 * child->flags of its traced child (same goes for fork, the parent 1517 * can write to the child->flags), because we're guaranteed the 1518 * child is not running and in turn not changing child->flags 1519 * at the same time the parent does it. 1520 */ 1521#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) 1522#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) 1523#define clear_used_math() clear_stopped_child_used_math(current) 1524#define set_used_math() set_stopped_child_used_math(current) 1525#define conditional_stopped_child_used_math(condition, child) \ 1526 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) 1527#define conditional_used_math(condition) \ 1528 conditional_stopped_child_used_math(condition, current) 1529#define copy_to_stopped_child_used_math(child) \ 1530 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) 1531/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ 1532#define tsk_used_math(p) ((p)->flags & PF_USED_MATH) 1533#define used_math() tsk_used_math(current) 1534 1535#ifdef CONFIG_SMP 1536extern int set_cpus_allowed_ptr(struct task_struct *p, 1537 const cpumask_t *new_mask); 1538#else 1539static inline int set_cpus_allowed_ptr(struct task_struct *p, 1540 const cpumask_t *new_mask) 1541{ 1542 if (!cpu_isset(0, *new_mask)) 1543 return -EINVAL; 1544 return 0; 1545} 1546#endif 1547static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask) 1548{ 1549 return set_cpus_allowed_ptr(p, &new_mask); 1550} 1551 1552extern unsigned long long sched_clock(void); 1553 1554#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK 1555static inline void sched_clock_init(void) 1556{ 1557} 1558 1559static inline u64 sched_clock_cpu(int cpu) 1560{ 1561 return sched_clock(); 1562} 1563 1564static inline void sched_clock_tick(void) 1565{ 1566} 1567 1568static inline void sched_clock_idle_sleep_event(void) 1569{ 1570} 1571 1572static inline void sched_clock_idle_wakeup_event(u64 delta_ns) 1573{ 1574} 1575 1576#ifdef CONFIG_NO_HZ 1577static inline void sched_clock_tick_stop(int cpu) 1578{ 1579} 1580 1581static inline void sched_clock_tick_start(int cpu) 1582{ 1583} 1584#endif 1585 1586#else /* CONFIG_HAVE_UNSTABLE_SCHED_CLOCK */ 1587extern void sched_clock_init(void); 1588extern u64 sched_clock_cpu(int cpu); 1589extern void sched_clock_tick(void); 1590extern void sched_clock_idle_sleep_event(void); 1591extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1592#ifdef CONFIG_NO_HZ 1593extern void sched_clock_tick_stop(int cpu); 1594extern void sched_clock_tick_start(int cpu); 1595#endif 1596#endif /* CONFIG_HAVE_UNSTABLE_SCHED_CLOCK */ 1597 1598/* 1599 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu 1600 * clock constructed from sched_clock(): 1601 */ 1602extern unsigned long long cpu_clock(int cpu); 1603 1604extern unsigned long long 1605task_sched_runtime(struct task_struct *task); 1606 1607/* sched_exec is called by processes performing an exec */ 1608#ifdef CONFIG_SMP 1609extern void sched_exec(void); 1610#else 1611#define sched_exec() {} 1612#endif 1613 1614extern void sched_clock_idle_sleep_event(void); 1615extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1616 1617#ifdef CONFIG_HOTPLUG_CPU 1618extern void idle_task_exit(void); 1619#else 1620static inline void idle_task_exit(void) {} 1621#endif 1622 1623extern void sched_idle_next(void); 1624 1625#if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP) 1626extern void wake_up_idle_cpu(int cpu); 1627#else 1628static inline void wake_up_idle_cpu(int cpu) { } 1629#endif 1630 1631#ifdef CONFIG_SCHED_DEBUG 1632extern unsigned int sysctl_sched_latency; 1633extern unsigned int sysctl_sched_min_granularity; 1634extern unsigned int sysctl_sched_wakeup_granularity; 1635extern unsigned int sysctl_sched_child_runs_first; 1636extern unsigned int sysctl_sched_features; 1637extern unsigned int sysctl_sched_migration_cost; 1638extern unsigned int sysctl_sched_nr_migrate; 1639extern unsigned int sysctl_sched_shares_ratelimit; 1640 1641int sched_nr_latency_handler(struct ctl_table *table, int write, 1642 struct file *file, void __user *buffer, size_t *length, 1643 loff_t *ppos); 1644#endif 1645extern unsigned int sysctl_sched_rt_period; 1646extern int sysctl_sched_rt_runtime; 1647 1648int sched_rt_handler(struct ctl_table *table, int write, 1649 struct file *filp, void __user *buffer, size_t *lenp, 1650 loff_t *ppos); 1651 1652extern unsigned int sysctl_sched_compat_yield; 1653 1654#ifdef CONFIG_RT_MUTEXES 1655extern int rt_mutex_getprio(struct task_struct *p); 1656extern void rt_mutex_setprio(struct task_struct *p, int prio); 1657extern void rt_mutex_adjust_pi(struct task_struct *p); 1658#else 1659static inline int rt_mutex_getprio(struct task_struct *p) 1660{ 1661 return p->normal_prio; 1662} 1663# define rt_mutex_adjust_pi(p) do { } while (0) 1664#endif 1665 1666extern void set_user_nice(struct task_struct *p, long nice); 1667extern int task_prio(const struct task_struct *p); 1668extern int task_nice(const struct task_struct *p); 1669extern int can_nice(const struct task_struct *p, const int nice); 1670extern int task_curr(const struct task_struct *p); 1671extern int idle_cpu(int cpu); 1672extern int sched_setscheduler(struct task_struct *, int, struct sched_param *); 1673extern int sched_setscheduler_nocheck(struct task_struct *, int, 1674 struct sched_param *); 1675extern struct task_struct *idle_task(int cpu); 1676extern struct task_struct *curr_task(int cpu); 1677extern void set_curr_task(int cpu, struct task_struct *p); 1678 1679void yield(void); 1680 1681/* 1682 * The default (Linux) execution domain. 1683 */ 1684extern struct exec_domain default_exec_domain; 1685 1686union thread_union { 1687 struct thread_info thread_info; 1688 unsigned long stack[THREAD_SIZE/sizeof(long)]; 1689}; 1690 1691#ifndef __HAVE_ARCH_KSTACK_END 1692static inline int kstack_end(void *addr) 1693{ 1694 /* Reliable end of stack detection: 1695 * Some APM bios versions misalign the stack 1696 */ 1697 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*))); 1698} 1699#endif 1700 1701extern union thread_union init_thread_union; 1702extern struct task_struct init_task; 1703 1704extern struct mm_struct init_mm; 1705 1706extern struct pid_namespace init_pid_ns; 1707 1708/* 1709 * find a task by one of its numerical ids 1710 * 1711 * find_task_by_pid_type_ns(): 1712 * it is the most generic call - it finds a task by all id, 1713 * type and namespace specified 1714 * find_task_by_pid_ns(): 1715 * finds a task by its pid in the specified namespace 1716 * find_task_by_vpid(): 1717 * finds a task by its virtual pid 1718 * find_task_by_pid(): 1719 * finds a task by its global pid 1720 * 1721 * see also find_pid() etc in include/linux/pid.h 1722 */ 1723 1724extern struct task_struct *find_task_by_pid_type_ns(int type, int pid, 1725 struct pid_namespace *ns); 1726 1727static inline struct task_struct *__deprecated find_task_by_pid(pid_t nr) 1728{ 1729 return find_task_by_pid_type_ns(PIDTYPE_PID, nr, &init_pid_ns); 1730} 1731extern struct task_struct *find_task_by_vpid(pid_t nr); 1732extern struct task_struct *find_task_by_pid_ns(pid_t nr, 1733 struct pid_namespace *ns); 1734 1735extern void __set_special_pids(struct pid *pid); 1736 1737/* per-UID process charging. */ 1738extern struct user_struct * alloc_uid(struct user_namespace *, uid_t); 1739static inline struct user_struct *get_uid(struct user_struct *u) 1740{ 1741 atomic_inc(&u->__count); 1742 return u; 1743} 1744extern void free_uid(struct user_struct *); 1745extern void switch_uid(struct user_struct *); 1746extern void release_uids(struct user_namespace *ns); 1747 1748#include <asm/current.h> 1749 1750extern void do_timer(unsigned long ticks); 1751 1752extern int wake_up_state(struct task_struct *tsk, unsigned int state); 1753extern int wake_up_process(struct task_struct *tsk); 1754extern void wake_up_new_task(struct task_struct *tsk, 1755 unsigned long clone_flags); 1756#ifdef CONFIG_SMP 1757 extern void kick_process(struct task_struct *tsk); 1758#else 1759 static inline void kick_process(struct task_struct *tsk) { } 1760#endif 1761extern void sched_fork(struct task_struct *p, int clone_flags); 1762extern void sched_dead(struct task_struct *p); 1763 1764extern int in_group_p(gid_t); 1765extern int in_egroup_p(gid_t); 1766 1767extern void proc_caches_init(void); 1768extern void flush_signals(struct task_struct *); 1769extern void ignore_signals(struct task_struct *); 1770extern void flush_signal_handlers(struct task_struct *, int force_default); 1771extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info); 1772 1773static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info) 1774{ 1775 unsigned long flags; 1776 int ret; 1777 1778 spin_lock_irqsave(&tsk->sighand->siglock, flags); 1779 ret = dequeue_signal(tsk, mask, info); 1780 spin_unlock_irqrestore(&tsk->sighand->siglock, flags); 1781 1782 return ret; 1783} 1784 1785extern void block_all_signals(int (*notifier)(void *priv), void *priv, 1786 sigset_t *mask); 1787extern void unblock_all_signals(void); 1788extern void release_task(struct task_struct * p); 1789extern int send_sig_info(int, struct siginfo *, struct task_struct *); 1790extern int force_sigsegv(int, struct task_struct *); 1791extern int force_sig_info(int, struct siginfo *, struct task_struct *); 1792extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); 1793extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid); 1794extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32); 1795extern int kill_pgrp(struct pid *pid, int sig, int priv); 1796extern int kill_pid(struct pid *pid, int sig, int priv); 1797extern int kill_proc_info(int, struct siginfo *, pid_t); 1798extern void do_notify_parent(struct task_struct *, int); 1799extern void force_sig(int, struct task_struct *); 1800extern void force_sig_specific(int, struct task_struct *); 1801extern int send_sig(int, struct task_struct *, int); 1802extern void zap_other_threads(struct task_struct *p); 1803extern int kill_proc(pid_t, int, int); 1804extern struct sigqueue *sigqueue_alloc(void); 1805extern void sigqueue_free(struct sigqueue *); 1806extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group); 1807extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *); 1808extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long); 1809 1810static inline int kill_cad_pid(int sig, int priv) 1811{ 1812 return kill_pid(cad_pid, sig, priv); 1813} 1814 1815/* These can be the second arg to send_sig_info/send_group_sig_info. */ 1816#define SEND_SIG_NOINFO ((struct siginfo *) 0) 1817#define SEND_SIG_PRIV ((struct siginfo *) 1) 1818#define SEND_SIG_FORCED ((struct siginfo *) 2) 1819 1820static inline int is_si_special(const struct siginfo *info) 1821{ 1822 return info <= SEND_SIG_FORCED; 1823} 1824 1825/* True if we are on the alternate signal stack. */ 1826 1827static inline int on_sig_stack(unsigned long sp) 1828{ 1829 return (sp - current->sas_ss_sp < current->sas_ss_size); 1830} 1831 1832static inline int sas_ss_flags(unsigned long sp) 1833{ 1834 return (current->sas_ss_size == 0 ? SS_DISABLE 1835 : on_sig_stack(sp) ? SS_ONSTACK : 0); 1836} 1837 1838/* 1839 * Routines for handling mm_structs 1840 */ 1841extern struct mm_struct * mm_alloc(void); 1842 1843/* mmdrop drops the mm and the page tables */ 1844extern void __mmdrop(struct mm_struct *); 1845static inline void mmdrop(struct mm_struct * mm) 1846{ 1847 if (unlikely(atomic_dec_and_test(&mm->mm_count))) 1848 __mmdrop(mm); 1849} 1850 1851/* mmput gets rid of the mappings and all user-space */ 1852extern void mmput(struct mm_struct *); 1853/* Grab a reference to a task's mm, if it is not already going away */ 1854extern struct mm_struct *get_task_mm(struct task_struct *task); 1855/* Remove the current tasks stale references to the old mm_struct */ 1856extern void mm_release(struct task_struct *, struct mm_struct *); 1857/* Allocate a new mm structure and copy contents from tsk->mm */ 1858extern struct mm_struct *dup_mm(struct task_struct *tsk); 1859 1860extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *); 1861extern void flush_thread(void); 1862extern void exit_thread(void); 1863 1864extern void exit_files(struct task_struct *); 1865extern void __cleanup_signal(struct signal_struct *); 1866extern void __cleanup_sighand(struct sighand_struct *); 1867 1868extern void exit_itimers(struct signal_struct *); 1869extern void flush_itimer_signals(void); 1870 1871extern NORET_TYPE void do_group_exit(int); 1872 1873extern void daemonize(const char *, ...); 1874extern int allow_signal(int); 1875extern int disallow_signal(int); 1876 1877extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *); 1878extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *); 1879struct task_struct *fork_idle(int); 1880 1881extern void set_task_comm(struct task_struct *tsk, char *from); 1882extern char *get_task_comm(char *to, struct task_struct *tsk); 1883 1884#ifdef CONFIG_SMP 1885extern void wait_task_inactive(struct task_struct * p); 1886#else 1887#define wait_task_inactive(p) do { } while (0) 1888#endif 1889 1890#define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks) 1891 1892#define for_each_process(p) \ 1893 for (p = &init_task ; (p = next_task(p)) != &init_task ; ) 1894 1895/* 1896 * Careful: do_each_thread/while_each_thread is a double loop so 1897 * 'break' will not work as expected - use goto instead. 1898 */ 1899#define do_each_thread(g, t) \ 1900 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do 1901 1902#define while_each_thread(g, t) \ 1903 while ((t = next_thread(t)) != g) 1904 1905/* de_thread depends on thread_group_leader not being a pid based check */ 1906#define thread_group_leader(p) (p == p->group_leader) 1907 1908/* Do to the insanities of de_thread it is possible for a process 1909 * to have the pid of the thread group leader without actually being 1910 * the thread group leader. For iteration through the pids in proc 1911 * all we care about is that we have a task with the appropriate 1912 * pid, we don't actually care if we have the right task. 1913 */ 1914static inline int has_group_leader_pid(struct task_struct *p) 1915{ 1916 return p->pid == p->tgid; 1917} 1918 1919static inline 1920int same_thread_group(struct task_struct *p1, struct task_struct *p2) 1921{ 1922 return p1->tgid == p2->tgid; 1923} 1924 1925static inline struct task_struct *next_thread(const struct task_struct *p) 1926{ 1927 return list_entry(rcu_dereference(p->thread_group.next), 1928 struct task_struct, thread_group); 1929} 1930 1931static inline int thread_group_empty(struct task_struct *p) 1932{ 1933 return list_empty(&p->thread_group); 1934} 1935 1936#define delay_group_leader(p) \ 1937 (thread_group_leader(p) && !thread_group_empty(p)) 1938 1939/* 1940 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring 1941 * subscriptions and synchronises with wait4(). Also used in procfs. Also 1942 * pins the final release of task.io_context. Also protects ->cpuset and 1943 * ->cgroup.subsys[]. 1944 * 1945 * Nests both inside and outside of read_lock(&tasklist_lock). 1946 * It must not be nested with write_lock_irq(&tasklist_lock), 1947 * neither inside nor outside. 1948 */ 1949static inline void task_lock(struct task_struct *p) 1950{ 1951 spin_lock(&p->alloc_lock); 1952} 1953 1954static inline void task_unlock(struct task_struct *p) 1955{ 1956 spin_unlock(&p->alloc_lock); 1957} 1958 1959extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk, 1960 unsigned long *flags); 1961 1962static inline void unlock_task_sighand(struct task_struct *tsk, 1963 unsigned long *flags) 1964{ 1965 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags); 1966} 1967 1968#ifndef __HAVE_THREAD_FUNCTIONS 1969 1970#define task_thread_info(task) ((struct thread_info *)(task)->stack) 1971#define task_stack_page(task) ((task)->stack) 1972 1973static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org) 1974{ 1975 *task_thread_info(p) = *task_thread_info(org); 1976 task_thread_info(p)->task = p; 1977} 1978 1979static inline unsigned long *end_of_stack(struct task_struct *p) 1980{ 1981 return (unsigned long *)(task_thread_info(p) + 1); 1982} 1983 1984#endif 1985 1986static inline int object_is_on_stack(void *obj) 1987{ 1988 void *stack = task_stack_page(current); 1989 1990 return (obj >= stack) && (obj < (stack + THREAD_SIZE)); 1991} 1992 1993extern void thread_info_cache_init(void); 1994 1995/* set thread flags in other task's structures 1996 * - see asm/thread_info.h for TIF_xxxx flags available 1997 */ 1998static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) 1999{ 2000 set_ti_thread_flag(task_thread_info(tsk), flag); 2001} 2002 2003static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2004{ 2005 clear_ti_thread_flag(task_thread_info(tsk), flag); 2006} 2007 2008static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) 2009{ 2010 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); 2011} 2012 2013static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2014{ 2015 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); 2016} 2017 2018static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) 2019{ 2020 return test_ti_thread_flag(task_thread_info(tsk), flag); 2021} 2022 2023static inline void set_tsk_need_resched(struct task_struct *tsk) 2024{ 2025 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2026} 2027 2028static inline void clear_tsk_need_resched(struct task_struct *tsk) 2029{ 2030 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2031} 2032 2033static inline int test_tsk_need_resched(struct task_struct *tsk) 2034{ 2035 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); 2036} 2037 2038static inline int signal_pending(struct task_struct *p) 2039{ 2040 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING)); 2041} 2042 2043extern int __fatal_signal_pending(struct task_struct *p); 2044 2045static inline int fatal_signal_pending(struct task_struct *p) 2046{ 2047 return signal_pending(p) && __fatal_signal_pending(p); 2048} 2049 2050static inline int signal_pending_state(long state, struct task_struct *p) 2051{ 2052 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL))) 2053 return 0; 2054 if (!signal_pending(p)) 2055 return 0; 2056 2057 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p); 2058} 2059 2060static inline int need_resched(void) 2061{ 2062 return unlikely(test_thread_flag(TIF_NEED_RESCHED)); 2063} 2064 2065/* 2066 * cond_resched() and cond_resched_lock(): latency reduction via 2067 * explicit rescheduling in places that are safe. The return 2068 * value indicates whether a reschedule was done in fact. 2069 * cond_resched_lock() will drop the spinlock before scheduling, 2070 * cond_resched_softirq() will enable bhs before scheduling. 2071 */ 2072extern int _cond_resched(void); 2073#ifdef CONFIG_PREEMPT_BKL 2074static inline int cond_resched(void) 2075{ 2076 return 0; 2077} 2078#else 2079static inline int cond_resched(void) 2080{ 2081 return _cond_resched(); 2082} 2083#endif 2084extern int cond_resched_lock(spinlock_t * lock); 2085extern int cond_resched_softirq(void); 2086static inline int cond_resched_bkl(void) 2087{ 2088 return _cond_resched(); 2089} 2090 2091/* 2092 * Does a critical section need to be broken due to another 2093 * task waiting?: (technically does not depend on CONFIG_PREEMPT, 2094 * but a general need for low latency) 2095 */ 2096static inline int spin_needbreak(spinlock_t *lock) 2097{ 2098#ifdef CONFIG_PREEMPT 2099 return spin_is_contended(lock); 2100#else 2101 return 0; 2102#endif 2103} 2104 2105/* 2106 * Reevaluate whether the task has signals pending delivery. 2107 * Wake the task if so. 2108 * This is required every time the blocked sigset_t changes. 2109 * callers must hold sighand->siglock. 2110 */ 2111extern void recalc_sigpending_and_wake(struct task_struct *t); 2112extern void recalc_sigpending(void); 2113 2114extern void signal_wake_up(struct task_struct *t, int resume_stopped); 2115 2116/* 2117 * Wrappers for p->thread_info->cpu access. No-op on UP. 2118 */ 2119#ifdef CONFIG_SMP 2120 2121static inline unsigned int task_cpu(const struct task_struct *p) 2122{ 2123 return task_thread_info(p)->cpu; 2124} 2125 2126extern void set_task_cpu(struct task_struct *p, unsigned int cpu); 2127 2128#else 2129 2130static inline unsigned int task_cpu(const struct task_struct *p) 2131{ 2132 return 0; 2133} 2134 2135static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 2136{ 2137} 2138 2139#endif /* CONFIG_SMP */ 2140 2141#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 2142extern void arch_pick_mmap_layout(struct mm_struct *mm); 2143#else 2144static inline void arch_pick_mmap_layout(struct mm_struct *mm) 2145{ 2146 mm->mmap_base = TASK_UNMAPPED_BASE; 2147 mm->get_unmapped_area = arch_get_unmapped_area; 2148 mm->unmap_area = arch_unmap_area; 2149} 2150#endif 2151 2152#ifdef CONFIG_TRACING 2153extern void 2154__trace_special(void *__tr, void *__data, 2155 unsigned long arg1, unsigned long arg2, unsigned long arg3); 2156#else 2157static inline void 2158__trace_special(void *__tr, void *__data, 2159 unsigned long arg1, unsigned long arg2, unsigned long arg3) 2160{ 2161} 2162#endif 2163 2164extern long sched_setaffinity(pid_t pid, const cpumask_t *new_mask); 2165extern long sched_getaffinity(pid_t pid, cpumask_t *mask); 2166 2167extern int sched_mc_power_savings, sched_smt_power_savings; 2168 2169extern void normalize_rt_tasks(void); 2170 2171#ifdef CONFIG_GROUP_SCHED 2172 2173extern struct task_group init_task_group; 2174#ifdef CONFIG_USER_SCHED 2175extern struct task_group root_task_group; 2176#endif 2177 2178extern struct task_group *sched_create_group(struct task_group *parent); 2179extern void sched_destroy_group(struct task_group *tg); 2180extern void sched_move_task(struct task_struct *tsk); 2181#ifdef CONFIG_FAIR_GROUP_SCHED 2182extern int sched_group_set_shares(struct task_group *tg, unsigned long shares); 2183extern unsigned long sched_group_shares(struct task_group *tg); 2184#endif 2185#ifdef CONFIG_RT_GROUP_SCHED 2186extern int sched_group_set_rt_runtime(struct task_group *tg, 2187 long rt_runtime_us); 2188extern long sched_group_rt_runtime(struct task_group *tg); 2189extern int sched_group_set_rt_period(struct task_group *tg, 2190 long rt_period_us); 2191extern long sched_group_rt_period(struct task_group *tg); 2192#endif 2193#endif 2194 2195#ifdef CONFIG_TASK_XACCT 2196static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2197{ 2198 tsk->rchar += amt; 2199} 2200 2201static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2202{ 2203 tsk->wchar += amt; 2204} 2205 2206static inline void inc_syscr(struct task_struct *tsk) 2207{ 2208 tsk->syscr++; 2209} 2210 2211static inline void inc_syscw(struct task_struct *tsk) 2212{ 2213 tsk->syscw++; 2214} 2215#else 2216static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2217{ 2218} 2219 2220static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2221{ 2222} 2223 2224static inline void inc_syscr(struct task_struct *tsk) 2225{ 2226} 2227 2228static inline void inc_syscw(struct task_struct *tsk) 2229{ 2230} 2231#endif 2232 2233#ifdef CONFIG_SMP 2234void migration_init(void); 2235#else 2236static inline void migration_init(void) 2237{ 2238} 2239#endif 2240 2241#ifndef TASK_SIZE_OF 2242#define TASK_SIZE_OF(tsk) TASK_SIZE 2243#endif 2244 2245#ifdef CONFIG_MM_OWNER 2246extern void mm_update_next_owner(struct mm_struct *mm); 2247extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p); 2248#else 2249static inline void mm_update_next_owner(struct mm_struct *mm) 2250{ 2251} 2252 2253static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p) 2254{ 2255} 2256#endif /* CONFIG_MM_OWNER */ 2257 2258#define TASK_STATE_TO_CHAR_STR "RSDTtZX" 2259 2260#endif /* __KERNEL__ */ 2261 2262#endif 2263