sched.h revision c714a534d85576af21b06be605ca55cb2fb887ee
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); 137extern unsigned long weighted_cpuload(const int cpu); 138 139struct seq_file; 140struct cfs_rq; 141struct task_group; 142#ifdef CONFIG_SCHED_DEBUG 143extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m); 144extern void proc_sched_set_task(struct task_struct *p); 145extern void 146print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq); 147#else 148static inline void 149proc_sched_show_task(struct task_struct *p, struct seq_file *m) 150{ 151} 152static inline void proc_sched_set_task(struct task_struct *p) 153{ 154} 155static inline void 156print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) 157{ 158} 159#endif 160 161extern unsigned long long time_sync_thresh; 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 * Pending notifications to parent. 561 */ 562#define SIGNAL_CLD_STOPPED 0x00000010 563#define SIGNAL_CLD_CONTINUED 0x00000020 564#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED) 565 566#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */ 567 568/* If true, all threads except ->group_exit_task have pending SIGKILL */ 569static inline int signal_group_exit(const struct signal_struct *sig) 570{ 571 return (sig->flags & SIGNAL_GROUP_EXIT) || 572 (sig->group_exit_task != NULL); 573} 574 575/* 576 * Some day this will be a full-fledged user tracking system.. 577 */ 578struct user_struct { 579 atomic_t __count; /* reference count */ 580 atomic_t processes; /* How many processes does this user have? */ 581 atomic_t files; /* How many open files does this user have? */ 582 atomic_t sigpending; /* How many pending signals does this user have? */ 583#ifdef CONFIG_INOTIFY_USER 584 atomic_t inotify_watches; /* How many inotify watches does this user have? */ 585 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */ 586#endif 587#ifdef CONFIG_POSIX_MQUEUE 588 /* protected by mq_lock */ 589 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */ 590#endif 591 unsigned long locked_shm; /* How many pages of mlocked shm ? */ 592 593#ifdef CONFIG_KEYS 594 struct key *uid_keyring; /* UID specific keyring */ 595 struct key *session_keyring; /* UID's default session keyring */ 596#endif 597 598 /* Hash table maintenance information */ 599 struct hlist_node uidhash_node; 600 uid_t uid; 601 602#ifdef CONFIG_USER_SCHED 603 struct task_group *tg; 604#ifdef CONFIG_SYSFS 605 struct kobject kobj; 606 struct work_struct work; 607#endif 608#endif 609}; 610 611extern int uids_sysfs_init(void); 612 613extern struct user_struct *find_user(uid_t); 614 615extern struct user_struct root_user; 616#define INIT_USER (&root_user) 617 618struct backing_dev_info; 619struct reclaim_state; 620 621#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 622struct sched_info { 623 /* cumulative counters */ 624 unsigned long pcount; /* # of times run on this cpu */ 625 unsigned long long cpu_time, /* time spent on the cpu */ 626 run_delay; /* time spent waiting on a runqueue */ 627 628 /* timestamps */ 629 unsigned long long last_arrival,/* when we last ran on a cpu */ 630 last_queued; /* when we were last queued to run */ 631#ifdef CONFIG_SCHEDSTATS 632 /* BKL stats */ 633 unsigned int bkl_count; 634#endif 635}; 636#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */ 637 638#ifdef CONFIG_SCHEDSTATS 639extern const struct file_operations proc_schedstat_operations; 640#endif /* CONFIG_SCHEDSTATS */ 641 642#ifdef CONFIG_TASK_DELAY_ACCT 643struct task_delay_info { 644 spinlock_t lock; 645 unsigned int flags; /* Private per-task flags */ 646 647 /* For each stat XXX, add following, aligned appropriately 648 * 649 * struct timespec XXX_start, XXX_end; 650 * u64 XXX_delay; 651 * u32 XXX_count; 652 * 653 * Atomicity of updates to XXX_delay, XXX_count protected by 654 * single lock above (split into XXX_lock if contention is an issue). 655 */ 656 657 /* 658 * XXX_count is incremented on every XXX operation, the delay 659 * associated with the operation is added to XXX_delay. 660 * XXX_delay contains the accumulated delay time in nanoseconds. 661 */ 662 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */ 663 u64 blkio_delay; /* wait for sync block io completion */ 664 u64 swapin_delay; /* wait for swapin block io completion */ 665 u32 blkio_count; /* total count of the number of sync block */ 666 /* io operations performed */ 667 u32 swapin_count; /* total count of the number of swapin block */ 668 /* io operations performed */ 669}; 670#endif /* CONFIG_TASK_DELAY_ACCT */ 671 672static inline int sched_info_on(void) 673{ 674#ifdef CONFIG_SCHEDSTATS 675 return 1; 676#elif defined(CONFIG_TASK_DELAY_ACCT) 677 extern int delayacct_on; 678 return delayacct_on; 679#else 680 return 0; 681#endif 682} 683 684enum cpu_idle_type { 685 CPU_IDLE, 686 CPU_NOT_IDLE, 687 CPU_NEWLY_IDLE, 688 CPU_MAX_IDLE_TYPES 689}; 690 691/* 692 * sched-domains (multiprocessor balancing) declarations: 693 */ 694 695/* 696 * Increase resolution of nice-level calculations: 697 */ 698#define SCHED_LOAD_SHIFT 10 699#define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT) 700 701#define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE 702 703#ifdef CONFIG_SMP 704#define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */ 705#define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */ 706#define SD_BALANCE_EXEC 4 /* Balance on exec */ 707#define SD_BALANCE_FORK 8 /* Balance on fork, clone */ 708#define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */ 709#define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */ 710#define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */ 711#define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */ 712#define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */ 713#define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */ 714#define SD_SERIALIZE 1024 /* Only a single load balancing instance */ 715#define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */ 716 717#define BALANCE_FOR_MC_POWER \ 718 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0) 719 720#define BALANCE_FOR_PKG_POWER \ 721 ((sched_mc_power_savings || sched_smt_power_savings) ? \ 722 SD_POWERSAVINGS_BALANCE : 0) 723 724#define test_sd_parent(sd, flag) ((sd->parent && \ 725 (sd->parent->flags & flag)) ? 1 : 0) 726 727 728struct sched_group { 729 struct sched_group *next; /* Must be a circular list */ 730 cpumask_t cpumask; 731 732 /* 733 * CPU power of this group, SCHED_LOAD_SCALE being max power for a 734 * single CPU. This is read only (except for setup, hotplug CPU). 735 * Note : Never change cpu_power without recompute its reciprocal 736 */ 737 unsigned int __cpu_power; 738 /* 739 * reciprocal value of cpu_power to avoid expensive divides 740 * (see include/linux/reciprocal_div.h) 741 */ 742 u32 reciprocal_cpu_power; 743}; 744 745enum sched_domain_level { 746 SD_LV_NONE = 0, 747 SD_LV_SIBLING, 748 SD_LV_MC, 749 SD_LV_CPU, 750 SD_LV_NODE, 751 SD_LV_ALLNODES, 752 SD_LV_MAX 753}; 754 755struct sched_domain_attr { 756 int relax_domain_level; 757}; 758 759#define SD_ATTR_INIT (struct sched_domain_attr) { \ 760 .relax_domain_level = -1, \ 761} 762 763struct sched_domain { 764 /* These fields must be setup */ 765 struct sched_domain *parent; /* top domain must be null terminated */ 766 struct sched_domain *child; /* bottom domain must be null terminated */ 767 struct sched_group *groups; /* the balancing groups of the domain */ 768 cpumask_t span; /* span of all CPUs in this domain */ 769 int first_cpu; /* cache of the first cpu in this domain */ 770 unsigned long min_interval; /* Minimum balance interval ms */ 771 unsigned long max_interval; /* Maximum balance interval ms */ 772 unsigned int busy_factor; /* less balancing by factor if busy */ 773 unsigned int imbalance_pct; /* No balance until over watermark */ 774 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */ 775 unsigned int busy_idx; 776 unsigned int idle_idx; 777 unsigned int newidle_idx; 778 unsigned int wake_idx; 779 unsigned int forkexec_idx; 780 int flags; /* See SD_* */ 781 enum sched_domain_level level; 782 783 /* Runtime fields. */ 784 unsigned long last_balance; /* init to jiffies. units in jiffies */ 785 unsigned int balance_interval; /* initialise to 1. units in ms. */ 786 unsigned int nr_balance_failed; /* initialise to 0 */ 787 788#ifdef CONFIG_SCHEDSTATS 789 /* load_balance() stats */ 790 unsigned int lb_count[CPU_MAX_IDLE_TYPES]; 791 unsigned int lb_failed[CPU_MAX_IDLE_TYPES]; 792 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES]; 793 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES]; 794 unsigned int lb_gained[CPU_MAX_IDLE_TYPES]; 795 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES]; 796 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES]; 797 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES]; 798 799 /* Active load balancing */ 800 unsigned int alb_count; 801 unsigned int alb_failed; 802 unsigned int alb_pushed; 803 804 /* SD_BALANCE_EXEC stats */ 805 unsigned int sbe_count; 806 unsigned int sbe_balanced; 807 unsigned int sbe_pushed; 808 809 /* SD_BALANCE_FORK stats */ 810 unsigned int sbf_count; 811 unsigned int sbf_balanced; 812 unsigned int sbf_pushed; 813 814 /* try_to_wake_up() stats */ 815 unsigned int ttwu_wake_remote; 816 unsigned int ttwu_move_affine; 817 unsigned int ttwu_move_balance; 818#endif 819}; 820 821extern void partition_sched_domains(int ndoms_new, cpumask_t *doms_new, 822 struct sched_domain_attr *dattr_new); 823extern int arch_reinit_sched_domains(void); 824 825#endif /* CONFIG_SMP */ 826 827/* 828 * A runqueue laden with a single nice 0 task scores a weighted_cpuload of 829 * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a 830 * task of nice 0 or enough lower priority tasks to bring up the 831 * weighted_cpuload 832 */ 833static inline int above_background_load(void) 834{ 835 unsigned long cpu; 836 837 for_each_online_cpu(cpu) { 838 if (weighted_cpuload(cpu) >= SCHED_LOAD_SCALE) 839 return 1; 840 } 841 return 0; 842} 843 844struct io_context; /* See blkdev.h */ 845#define NGROUPS_SMALL 32 846#define NGROUPS_PER_BLOCK ((unsigned int)(PAGE_SIZE / sizeof(gid_t))) 847struct group_info { 848 int ngroups; 849 atomic_t usage; 850 gid_t small_block[NGROUPS_SMALL]; 851 int nblocks; 852 gid_t *blocks[0]; 853}; 854 855/* 856 * get_group_info() must be called with the owning task locked (via task_lock()) 857 * when task != current. The reason being that the vast majority of callers are 858 * looking at current->group_info, which can not be changed except by the 859 * current task. Changing current->group_info requires the task lock, too. 860 */ 861#define get_group_info(group_info) do { \ 862 atomic_inc(&(group_info)->usage); \ 863} while (0) 864 865#define put_group_info(group_info) do { \ 866 if (atomic_dec_and_test(&(group_info)->usage)) \ 867 groups_free(group_info); \ 868} while (0) 869 870extern struct group_info *groups_alloc(int gidsetsize); 871extern void groups_free(struct group_info *group_info); 872extern int set_current_groups(struct group_info *group_info); 873extern int groups_search(struct group_info *group_info, gid_t grp); 874/* access the groups "array" with this macro */ 875#define GROUP_AT(gi, i) \ 876 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK]) 877 878#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK 879extern void prefetch_stack(struct task_struct *t); 880#else 881static inline void prefetch_stack(struct task_struct *t) { } 882#endif 883 884struct audit_context; /* See audit.c */ 885struct mempolicy; 886struct pipe_inode_info; 887struct uts_namespace; 888 889struct rq; 890struct sched_domain; 891 892struct sched_class { 893 const struct sched_class *next; 894 895 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup); 896 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep); 897 void (*yield_task) (struct rq *rq); 898 int (*select_task_rq)(struct task_struct *p, int sync); 899 900 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p); 901 902 struct task_struct * (*pick_next_task) (struct rq *rq); 903 void (*put_prev_task) (struct rq *rq, struct task_struct *p); 904 905#ifdef CONFIG_SMP 906 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu, 907 struct rq *busiest, unsigned long max_load_move, 908 struct sched_domain *sd, enum cpu_idle_type idle, 909 int *all_pinned, int *this_best_prio); 910 911 int (*move_one_task) (struct rq *this_rq, int this_cpu, 912 struct rq *busiest, struct sched_domain *sd, 913 enum cpu_idle_type idle); 914 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task); 915 void (*post_schedule) (struct rq *this_rq); 916 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task); 917#endif 918 919 void (*set_curr_task) (struct rq *rq); 920 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued); 921 void (*task_new) (struct rq *rq, struct task_struct *p); 922 void (*set_cpus_allowed)(struct task_struct *p, 923 const cpumask_t *newmask); 924 925 void (*join_domain)(struct rq *rq); 926 void (*leave_domain)(struct rq *rq); 927 928 void (*switched_from) (struct rq *this_rq, struct task_struct *task, 929 int running); 930 void (*switched_to) (struct rq *this_rq, struct task_struct *task, 931 int running); 932 void (*prio_changed) (struct rq *this_rq, struct task_struct *task, 933 int oldprio, int running); 934 935#ifdef CONFIG_FAIR_GROUP_SCHED 936 void (*moved_group) (struct task_struct *p); 937#endif 938}; 939 940struct load_weight { 941 unsigned long weight, inv_weight; 942}; 943 944/* 945 * CFS stats for a schedulable entity (task, task-group etc) 946 * 947 * Current field usage histogram: 948 * 949 * 4 se->block_start 950 * 4 se->run_node 951 * 4 se->sleep_start 952 * 6 se->load.weight 953 */ 954struct sched_entity { 955 struct load_weight load; /* for load-balancing */ 956 struct rb_node run_node; 957 struct list_head group_node; 958 unsigned int on_rq; 959 960 u64 exec_start; 961 u64 sum_exec_runtime; 962 u64 vruntime; 963 u64 prev_sum_exec_runtime; 964 965 u64 last_wakeup; 966 u64 avg_overlap; 967 968#ifdef CONFIG_SCHEDSTATS 969 u64 wait_start; 970 u64 wait_max; 971 u64 wait_count; 972 u64 wait_sum; 973 974 u64 sleep_start; 975 u64 sleep_max; 976 s64 sum_sleep_runtime; 977 978 u64 block_start; 979 u64 block_max; 980 u64 exec_max; 981 u64 slice_max; 982 983 u64 nr_migrations; 984 u64 nr_migrations_cold; 985 u64 nr_failed_migrations_affine; 986 u64 nr_failed_migrations_running; 987 u64 nr_failed_migrations_hot; 988 u64 nr_forced_migrations; 989 u64 nr_forced2_migrations; 990 991 u64 nr_wakeups; 992 u64 nr_wakeups_sync; 993 u64 nr_wakeups_migrate; 994 u64 nr_wakeups_local; 995 u64 nr_wakeups_remote; 996 u64 nr_wakeups_affine; 997 u64 nr_wakeups_affine_attempts; 998 u64 nr_wakeups_passive; 999 u64 nr_wakeups_idle; 1000#endif 1001 1002#ifdef CONFIG_FAIR_GROUP_SCHED 1003 struct sched_entity *parent; 1004 /* rq on which this entity is (to be) queued: */ 1005 struct cfs_rq *cfs_rq; 1006 /* rq "owned" by this entity/group: */ 1007 struct cfs_rq *my_q; 1008#endif 1009}; 1010 1011struct sched_rt_entity { 1012 struct list_head run_list; 1013 unsigned int time_slice; 1014 unsigned long timeout; 1015 int nr_cpus_allowed; 1016 1017 struct sched_rt_entity *back; 1018#ifdef CONFIG_RT_GROUP_SCHED 1019 struct sched_rt_entity *parent; 1020 /* rq on which this entity is (to be) queued: */ 1021 struct rt_rq *rt_rq; 1022 /* rq "owned" by this entity/group: */ 1023 struct rt_rq *my_q; 1024#endif 1025}; 1026 1027struct task_struct { 1028 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */ 1029 void *stack; 1030 atomic_t usage; 1031 unsigned int flags; /* per process flags, defined below */ 1032 unsigned int ptrace; 1033 1034 int lock_depth; /* BKL lock depth */ 1035 1036#ifdef CONFIG_SMP 1037#ifdef __ARCH_WANT_UNLOCKED_CTXSW 1038 int oncpu; 1039#endif 1040#endif 1041 1042 int prio, static_prio, normal_prio; 1043 const struct sched_class *sched_class; 1044 struct sched_entity se; 1045 struct sched_rt_entity rt; 1046 1047#ifdef CONFIG_PREEMPT_NOTIFIERS 1048 /* list of struct preempt_notifier: */ 1049 struct hlist_head preempt_notifiers; 1050#endif 1051 1052 /* 1053 * fpu_counter contains the number of consecutive context switches 1054 * that the FPU is used. If this is over a threshold, the lazy fpu 1055 * saving becomes unlazy to save the trap. This is an unsigned char 1056 * so that after 256 times the counter wraps and the behavior turns 1057 * lazy again; this to deal with bursty apps that only use FPU for 1058 * a short time 1059 */ 1060 unsigned char fpu_counter; 1061 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */ 1062#ifdef CONFIG_BLK_DEV_IO_TRACE 1063 unsigned int btrace_seq; 1064#endif 1065 1066 unsigned int policy; 1067 cpumask_t cpus_allowed; 1068 1069#ifdef CONFIG_PREEMPT_RCU 1070 int rcu_read_lock_nesting; 1071 int rcu_flipctr_idx; 1072#endif /* #ifdef CONFIG_PREEMPT_RCU */ 1073 1074#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 1075 struct sched_info sched_info; 1076#endif 1077 1078 struct list_head tasks; 1079 /* 1080 * ptrace_list/ptrace_children forms the list of my children 1081 * that were stolen by a ptracer. 1082 */ 1083 struct list_head ptrace_children; 1084 struct list_head ptrace_list; 1085 1086 struct mm_struct *mm, *active_mm; 1087 1088/* task state */ 1089 struct linux_binfmt *binfmt; 1090 int exit_state; 1091 int exit_code, exit_signal; 1092 int pdeath_signal; /* The signal sent when the parent dies */ 1093 /* ??? */ 1094 unsigned int personality; 1095 unsigned did_exec:1; 1096 pid_t pid; 1097 pid_t tgid; 1098 1099#ifdef CONFIG_CC_STACKPROTECTOR 1100 /* Canary value for the -fstack-protector gcc feature */ 1101 unsigned long stack_canary; 1102#endif 1103 /* 1104 * pointers to (original) parent process, youngest child, younger sibling, 1105 * older sibling, respectively. (p->father can be replaced with 1106 * p->parent->pid) 1107 */ 1108 struct task_struct *real_parent; /* real parent process (when being debugged) */ 1109 struct task_struct *parent; /* parent process */ 1110 /* 1111 * children/sibling forms the list of my children plus the 1112 * tasks I'm ptracing. 1113 */ 1114 struct list_head children; /* list of my children */ 1115 struct list_head sibling; /* linkage in my parent's children list */ 1116 struct task_struct *group_leader; /* threadgroup leader */ 1117 1118 /* PID/PID hash table linkage. */ 1119 struct pid_link pids[PIDTYPE_MAX]; 1120 struct list_head thread_group; 1121 1122 struct completion *vfork_done; /* for vfork() */ 1123 int __user *set_child_tid; /* CLONE_CHILD_SETTID */ 1124 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ 1125 1126 unsigned int rt_priority; 1127 cputime_t utime, stime, utimescaled, stimescaled; 1128 cputime_t gtime; 1129 cputime_t prev_utime, prev_stime; 1130 unsigned long nvcsw, nivcsw; /* context switch counts */ 1131 struct timespec start_time; /* monotonic time */ 1132 struct timespec real_start_time; /* boot based time */ 1133/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ 1134 unsigned long min_flt, maj_flt; 1135 1136 cputime_t it_prof_expires, it_virt_expires; 1137 unsigned long long it_sched_expires; 1138 struct list_head cpu_timers[3]; 1139 1140/* process credentials */ 1141 uid_t uid,euid,suid,fsuid; 1142 gid_t gid,egid,sgid,fsgid; 1143 struct group_info *group_info; 1144 kernel_cap_t cap_effective, cap_inheritable, cap_permitted, cap_bset; 1145 unsigned securebits; 1146 struct user_struct *user; 1147#ifdef CONFIG_KEYS 1148 struct key *request_key_auth; /* assumed request_key authority */ 1149 struct key *thread_keyring; /* keyring private to this thread */ 1150 unsigned char jit_keyring; /* default keyring to attach requested keys to */ 1151#endif 1152 char comm[TASK_COMM_LEN]; /* executable name excluding path 1153 - access with [gs]et_task_comm (which lock 1154 it with task_lock()) 1155 - initialized normally by flush_old_exec */ 1156/* file system info */ 1157 int link_count, total_link_count; 1158#ifdef CONFIG_SYSVIPC 1159/* ipc stuff */ 1160 struct sysv_sem sysvsem; 1161#endif 1162#ifdef CONFIG_DETECT_SOFTLOCKUP 1163/* hung task detection */ 1164 unsigned long last_switch_timestamp; 1165 unsigned long last_switch_count; 1166#endif 1167/* CPU-specific state of this task */ 1168 struct thread_struct thread; 1169/* filesystem information */ 1170 struct fs_struct *fs; 1171/* open file information */ 1172 struct files_struct *files; 1173/* namespaces */ 1174 struct nsproxy *nsproxy; 1175/* signal handlers */ 1176 struct signal_struct *signal; 1177 struct sighand_struct *sighand; 1178 1179 sigset_t blocked, real_blocked; 1180 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */ 1181 struct sigpending pending; 1182 1183 unsigned long sas_ss_sp; 1184 size_t sas_ss_size; 1185 int (*notifier)(void *priv); 1186 void *notifier_data; 1187 sigset_t *notifier_mask; 1188#ifdef CONFIG_SECURITY 1189 void *security; 1190#endif 1191 struct audit_context *audit_context; 1192#ifdef CONFIG_AUDITSYSCALL 1193 uid_t loginuid; 1194 unsigned int sessionid; 1195#endif 1196 seccomp_t seccomp; 1197 1198/* Thread group tracking */ 1199 u32 parent_exec_id; 1200 u32 self_exec_id; 1201/* Protection of (de-)allocation: mm, files, fs, tty, keyrings */ 1202 spinlock_t alloc_lock; 1203 1204 /* Protection of the PI data structures: */ 1205 spinlock_t pi_lock; 1206 1207#ifdef CONFIG_RT_MUTEXES 1208 /* PI waiters blocked on a rt_mutex held by this task */ 1209 struct plist_head pi_waiters; 1210 /* Deadlock detection and priority inheritance handling */ 1211 struct rt_mutex_waiter *pi_blocked_on; 1212#endif 1213 1214#ifdef CONFIG_DEBUG_MUTEXES 1215 /* mutex deadlock detection */ 1216 struct mutex_waiter *blocked_on; 1217#endif 1218#ifdef CONFIG_TRACE_IRQFLAGS 1219 unsigned int irq_events; 1220 int hardirqs_enabled; 1221 unsigned long hardirq_enable_ip; 1222 unsigned int hardirq_enable_event; 1223 unsigned long hardirq_disable_ip; 1224 unsigned int hardirq_disable_event; 1225 int softirqs_enabled; 1226 unsigned long softirq_disable_ip; 1227 unsigned int softirq_disable_event; 1228 unsigned long softirq_enable_ip; 1229 unsigned int softirq_enable_event; 1230 int hardirq_context; 1231 int softirq_context; 1232#endif 1233#ifdef CONFIG_LOCKDEP 1234# define MAX_LOCK_DEPTH 48UL 1235 u64 curr_chain_key; 1236 int lockdep_depth; 1237 struct held_lock held_locks[MAX_LOCK_DEPTH]; 1238 unsigned int lockdep_recursion; 1239#endif 1240 1241/* journalling filesystem info */ 1242 void *journal_info; 1243 1244/* stacked block device info */ 1245 struct bio *bio_list, **bio_tail; 1246 1247/* VM state */ 1248 struct reclaim_state *reclaim_state; 1249 1250 struct backing_dev_info *backing_dev_info; 1251 1252 struct io_context *io_context; 1253 1254 unsigned long ptrace_message; 1255 siginfo_t *last_siginfo; /* For ptrace use. */ 1256#ifdef CONFIG_TASK_XACCT 1257/* i/o counters(bytes read/written, #syscalls */ 1258 u64 rchar, wchar, syscr, syscw; 1259#endif 1260 struct task_io_accounting ioac; 1261#if defined(CONFIG_TASK_XACCT) 1262 u64 acct_rss_mem1; /* accumulated rss usage */ 1263 u64 acct_vm_mem1; /* accumulated virtual memory usage */ 1264 cputime_t acct_stimexpd;/* stime since last update */ 1265#endif 1266#ifdef CONFIG_NUMA 1267 struct mempolicy *mempolicy; 1268 short il_next; 1269#endif 1270#ifdef CONFIG_CPUSETS 1271 nodemask_t mems_allowed; 1272 int cpuset_mems_generation; 1273 int cpuset_mem_spread_rotor; 1274#endif 1275#ifdef CONFIG_CGROUPS 1276 /* Control Group info protected by css_set_lock */ 1277 struct css_set *cgroups; 1278 /* cg_list protected by css_set_lock and tsk->alloc_lock */ 1279 struct list_head cg_list; 1280#endif 1281#ifdef CONFIG_FUTEX 1282 struct robust_list_head __user *robust_list; 1283#ifdef CONFIG_COMPAT 1284 struct compat_robust_list_head __user *compat_robust_list; 1285#endif 1286 struct list_head pi_state_list; 1287 struct futex_pi_state *pi_state_cache; 1288#endif 1289 atomic_t fs_excl; /* holding fs exclusive resources */ 1290 struct rcu_head rcu; 1291 1292 /* 1293 * cache last used pipe for splice 1294 */ 1295 struct pipe_inode_info *splice_pipe; 1296#ifdef CONFIG_TASK_DELAY_ACCT 1297 struct task_delay_info *delays; 1298#endif 1299#ifdef CONFIG_FAULT_INJECTION 1300 int make_it_fail; 1301#endif 1302 struct prop_local_single dirties; 1303#ifdef CONFIG_LATENCYTOP 1304 int latency_record_count; 1305 struct latency_record latency_record[LT_SAVECOUNT]; 1306#endif 1307}; 1308 1309/* 1310 * Priority of a process goes from 0..MAX_PRIO-1, valid RT 1311 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH 1312 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority 1313 * values are inverted: lower p->prio value means higher priority. 1314 * 1315 * The MAX_USER_RT_PRIO value allows the actual maximum 1316 * RT priority to be separate from the value exported to 1317 * user-space. This allows kernel threads to set their 1318 * priority to a value higher than any user task. Note: 1319 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO. 1320 */ 1321 1322#define MAX_USER_RT_PRIO 100 1323#define MAX_RT_PRIO MAX_USER_RT_PRIO 1324 1325#define MAX_PRIO (MAX_RT_PRIO + 40) 1326#define DEFAULT_PRIO (MAX_RT_PRIO + 20) 1327 1328static inline int rt_prio(int prio) 1329{ 1330 if (unlikely(prio < MAX_RT_PRIO)) 1331 return 1; 1332 return 0; 1333} 1334 1335static inline int rt_task(struct task_struct *p) 1336{ 1337 return rt_prio(p->prio); 1338} 1339 1340static inline void set_task_session(struct task_struct *tsk, pid_t session) 1341{ 1342 tsk->signal->__session = session; 1343} 1344 1345static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp) 1346{ 1347 tsk->signal->__pgrp = pgrp; 1348} 1349 1350static inline struct pid *task_pid(struct task_struct *task) 1351{ 1352 return task->pids[PIDTYPE_PID].pid; 1353} 1354 1355static inline struct pid *task_tgid(struct task_struct *task) 1356{ 1357 return task->group_leader->pids[PIDTYPE_PID].pid; 1358} 1359 1360static inline struct pid *task_pgrp(struct task_struct *task) 1361{ 1362 return task->group_leader->pids[PIDTYPE_PGID].pid; 1363} 1364 1365static inline struct pid *task_session(struct task_struct *task) 1366{ 1367 return task->group_leader->pids[PIDTYPE_SID].pid; 1368} 1369 1370struct pid_namespace; 1371 1372/* 1373 * the helpers to get the task's different pids as they are seen 1374 * from various namespaces 1375 * 1376 * task_xid_nr() : global id, i.e. the id seen from the init namespace; 1377 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of 1378 * current. 1379 * task_xid_nr_ns() : id seen from the ns specified; 1380 * 1381 * set_task_vxid() : assigns a virtual id to a task; 1382 * 1383 * see also pid_nr() etc in include/linux/pid.h 1384 */ 1385 1386static inline pid_t task_pid_nr(struct task_struct *tsk) 1387{ 1388 return tsk->pid; 1389} 1390 1391pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1392 1393static inline pid_t task_pid_vnr(struct task_struct *tsk) 1394{ 1395 return pid_vnr(task_pid(tsk)); 1396} 1397 1398 1399static inline pid_t task_tgid_nr(struct task_struct *tsk) 1400{ 1401 return tsk->tgid; 1402} 1403 1404pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1405 1406static inline pid_t task_tgid_vnr(struct task_struct *tsk) 1407{ 1408 return pid_vnr(task_tgid(tsk)); 1409} 1410 1411 1412static inline pid_t task_pgrp_nr(struct task_struct *tsk) 1413{ 1414 return tsk->signal->__pgrp; 1415} 1416 1417pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1418 1419static inline pid_t task_pgrp_vnr(struct task_struct *tsk) 1420{ 1421 return pid_vnr(task_pgrp(tsk)); 1422} 1423 1424 1425static inline pid_t task_session_nr(struct task_struct *tsk) 1426{ 1427 return tsk->signal->__session; 1428} 1429 1430pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1431 1432static inline pid_t task_session_vnr(struct task_struct *tsk) 1433{ 1434 return pid_vnr(task_session(tsk)); 1435} 1436 1437 1438/** 1439 * pid_alive - check that a task structure is not stale 1440 * @p: Task structure to be checked. 1441 * 1442 * Test if a process is not yet dead (at most zombie state) 1443 * If pid_alive fails, then pointers within the task structure 1444 * can be stale and must not be dereferenced. 1445 */ 1446static inline int pid_alive(struct task_struct *p) 1447{ 1448 return p->pids[PIDTYPE_PID].pid != NULL; 1449} 1450 1451/** 1452 * is_global_init - check if a task structure is init 1453 * @tsk: Task structure to be checked. 1454 * 1455 * Check if a task structure is the first user space task the kernel created. 1456 */ 1457static inline int is_global_init(struct task_struct *tsk) 1458{ 1459 return tsk->pid == 1; 1460} 1461 1462/* 1463 * is_container_init: 1464 * check whether in the task is init in its own pid namespace. 1465 */ 1466extern int is_container_init(struct task_struct *tsk); 1467 1468extern struct pid *cad_pid; 1469 1470extern void free_task(struct task_struct *tsk); 1471#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0) 1472 1473extern void __put_task_struct(struct task_struct *t); 1474 1475static inline void put_task_struct(struct task_struct *t) 1476{ 1477 if (atomic_dec_and_test(&t->usage)) 1478 __put_task_struct(t); 1479} 1480 1481/* 1482 * Per process flags 1483 */ 1484#define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */ 1485 /* Not implemented yet, only for 486*/ 1486#define PF_STARTING 0x00000002 /* being created */ 1487#define PF_EXITING 0x00000004 /* getting shut down */ 1488#define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */ 1489#define PF_VCPU 0x00000010 /* I'm a virtual CPU */ 1490#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ 1491#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ 1492#define PF_DUMPCORE 0x00000200 /* dumped core */ 1493#define PF_SIGNALED 0x00000400 /* killed by a signal */ 1494#define PF_MEMALLOC 0x00000800 /* Allocating memory */ 1495#define PF_FLUSHER 0x00001000 /* responsible for disk writeback */ 1496#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */ 1497#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */ 1498#define PF_FROZEN 0x00010000 /* frozen for system suspend */ 1499#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ 1500#define PF_KSWAPD 0x00040000 /* I am kswapd */ 1501#define PF_SWAPOFF 0x00080000 /* I am in swapoff */ 1502#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ 1503#define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */ 1504#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */ 1505#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ 1506#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */ 1507#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */ 1508#define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */ 1509#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ 1510#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */ 1511 1512/* 1513 * Only the _current_ task can read/write to tsk->flags, but other 1514 * tasks can access tsk->flags in readonly mode for example 1515 * with tsk_used_math (like during threaded core dumping). 1516 * There is however an exception to this rule during ptrace 1517 * or during fork: the ptracer task is allowed to write to the 1518 * child->flags of its traced child (same goes for fork, the parent 1519 * can write to the child->flags), because we're guaranteed the 1520 * child is not running and in turn not changing child->flags 1521 * at the same time the parent does it. 1522 */ 1523#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) 1524#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) 1525#define clear_used_math() clear_stopped_child_used_math(current) 1526#define set_used_math() set_stopped_child_used_math(current) 1527#define conditional_stopped_child_used_math(condition, child) \ 1528 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) 1529#define conditional_used_math(condition) \ 1530 conditional_stopped_child_used_math(condition, current) 1531#define copy_to_stopped_child_used_math(child) \ 1532 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) 1533/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ 1534#define tsk_used_math(p) ((p)->flags & PF_USED_MATH) 1535#define used_math() tsk_used_math(current) 1536 1537#ifdef CONFIG_SMP 1538extern int set_cpus_allowed_ptr(struct task_struct *p, 1539 const cpumask_t *new_mask); 1540#else 1541static inline int set_cpus_allowed_ptr(struct task_struct *p, 1542 const cpumask_t *new_mask) 1543{ 1544 if (!cpu_isset(0, *new_mask)) 1545 return -EINVAL; 1546 return 0; 1547} 1548#endif 1549static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask) 1550{ 1551 return set_cpus_allowed_ptr(p, &new_mask); 1552} 1553 1554extern unsigned long long sched_clock(void); 1555 1556#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK 1557static inline void sched_clock_init(void) 1558{ 1559} 1560 1561static inline u64 sched_clock_cpu(int cpu) 1562{ 1563 return sched_clock(); 1564} 1565 1566static inline void sched_clock_tick(void) 1567{ 1568} 1569 1570static inline void sched_clock_idle_sleep_event(void) 1571{ 1572} 1573 1574static inline void sched_clock_idle_wakeup_event(u64 delta_ns) 1575{ 1576} 1577#else 1578extern void sched_clock_init(void); 1579extern u64 sched_clock_cpu(int cpu); 1580extern void sched_clock_tick(void); 1581extern void sched_clock_idle_sleep_event(void); 1582extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1583#endif 1584 1585/* 1586 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu 1587 * clock constructed from sched_clock(): 1588 */ 1589extern unsigned long long cpu_clock(int cpu); 1590 1591extern unsigned long long 1592task_sched_runtime(struct task_struct *task); 1593 1594/* sched_exec is called by processes performing an exec */ 1595#ifdef CONFIG_SMP 1596extern void sched_exec(void); 1597#else 1598#define sched_exec() {} 1599#endif 1600 1601extern void sched_clock_idle_sleep_event(void); 1602extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1603 1604#ifdef CONFIG_HOTPLUG_CPU 1605extern void idle_task_exit(void); 1606#else 1607static inline void idle_task_exit(void) {} 1608#endif 1609 1610extern void sched_idle_next(void); 1611 1612#if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP) 1613extern void wake_up_idle_cpu(int cpu); 1614#else 1615static inline void wake_up_idle_cpu(int cpu) { } 1616#endif 1617 1618#ifdef CONFIG_SCHED_DEBUG 1619extern unsigned int sysctl_sched_latency; 1620extern unsigned int sysctl_sched_min_granularity; 1621extern unsigned int sysctl_sched_wakeup_granularity; 1622extern unsigned int sysctl_sched_child_runs_first; 1623extern unsigned int sysctl_sched_features; 1624extern unsigned int sysctl_sched_migration_cost; 1625extern unsigned int sysctl_sched_nr_migrate; 1626 1627int sched_nr_latency_handler(struct ctl_table *table, int write, 1628 struct file *file, void __user *buffer, size_t *length, 1629 loff_t *ppos); 1630#endif 1631extern unsigned int sysctl_sched_rt_period; 1632extern int sysctl_sched_rt_runtime; 1633 1634int sched_rt_handler(struct ctl_table *table, int write, 1635 struct file *filp, void __user *buffer, size_t *lenp, 1636 loff_t *ppos); 1637 1638extern unsigned int sysctl_sched_compat_yield; 1639 1640#ifdef CONFIG_RT_MUTEXES 1641extern int rt_mutex_getprio(struct task_struct *p); 1642extern void rt_mutex_setprio(struct task_struct *p, int prio); 1643extern void rt_mutex_adjust_pi(struct task_struct *p); 1644#else 1645static inline int rt_mutex_getprio(struct task_struct *p) 1646{ 1647 return p->normal_prio; 1648} 1649# define rt_mutex_adjust_pi(p) do { } while (0) 1650#endif 1651 1652extern void set_user_nice(struct task_struct *p, long nice); 1653extern int task_prio(const struct task_struct *p); 1654extern int task_nice(const struct task_struct *p); 1655extern int can_nice(const struct task_struct *p, const int nice); 1656extern int task_curr(const struct task_struct *p); 1657extern int idle_cpu(int cpu); 1658extern int sched_setscheduler(struct task_struct *, int, struct sched_param *); 1659extern struct task_struct *idle_task(int cpu); 1660extern struct task_struct *curr_task(int cpu); 1661extern void set_curr_task(int cpu, struct task_struct *p); 1662 1663void yield(void); 1664 1665/* 1666 * The default (Linux) execution domain. 1667 */ 1668extern struct exec_domain default_exec_domain; 1669 1670union thread_union { 1671 struct thread_info thread_info; 1672 unsigned long stack[THREAD_SIZE/sizeof(long)]; 1673}; 1674 1675#ifndef __HAVE_ARCH_KSTACK_END 1676static inline int kstack_end(void *addr) 1677{ 1678 /* Reliable end of stack detection: 1679 * Some APM bios versions misalign the stack 1680 */ 1681 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*))); 1682} 1683#endif 1684 1685extern union thread_union init_thread_union; 1686extern struct task_struct init_task; 1687 1688extern struct mm_struct init_mm; 1689 1690extern struct pid_namespace init_pid_ns; 1691 1692/* 1693 * find a task by one of its numerical ids 1694 * 1695 * find_task_by_pid_type_ns(): 1696 * it is the most generic call - it finds a task by all id, 1697 * type and namespace specified 1698 * find_task_by_pid_ns(): 1699 * finds a task by its pid in the specified namespace 1700 * find_task_by_vpid(): 1701 * finds a task by its virtual pid 1702 * find_task_by_pid(): 1703 * finds a task by its global pid 1704 * 1705 * see also find_pid() etc in include/linux/pid.h 1706 */ 1707 1708extern struct task_struct *find_task_by_pid_type_ns(int type, int pid, 1709 struct pid_namespace *ns); 1710 1711static inline struct task_struct *__deprecated find_task_by_pid(pid_t nr) 1712{ 1713 return find_task_by_pid_type_ns(PIDTYPE_PID, nr, &init_pid_ns); 1714} 1715extern struct task_struct *find_task_by_vpid(pid_t nr); 1716extern struct task_struct *find_task_by_pid_ns(pid_t nr, 1717 struct pid_namespace *ns); 1718 1719extern void __set_special_pids(struct pid *pid); 1720 1721/* per-UID process charging. */ 1722extern struct user_struct * alloc_uid(struct user_namespace *, uid_t); 1723static inline struct user_struct *get_uid(struct user_struct *u) 1724{ 1725 atomic_inc(&u->__count); 1726 return u; 1727} 1728extern void free_uid(struct user_struct *); 1729extern void switch_uid(struct user_struct *); 1730extern void release_uids(struct user_namespace *ns); 1731 1732#include <asm/current.h> 1733 1734extern void do_timer(unsigned long ticks); 1735 1736extern int wake_up_state(struct task_struct *tsk, unsigned int state); 1737extern int wake_up_process(struct task_struct *tsk); 1738extern void wake_up_new_task(struct task_struct *tsk, 1739 unsigned long clone_flags); 1740#ifdef CONFIG_SMP 1741 extern void kick_process(struct task_struct *tsk); 1742#else 1743 static inline void kick_process(struct task_struct *tsk) { } 1744#endif 1745extern void sched_fork(struct task_struct *p, int clone_flags); 1746extern void sched_dead(struct task_struct *p); 1747 1748extern int in_group_p(gid_t); 1749extern int in_egroup_p(gid_t); 1750 1751extern void proc_caches_init(void); 1752extern void flush_signals(struct task_struct *); 1753extern void ignore_signals(struct task_struct *); 1754extern void flush_signal_handlers(struct task_struct *, int force_default); 1755extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info); 1756 1757static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info) 1758{ 1759 unsigned long flags; 1760 int ret; 1761 1762 spin_lock_irqsave(&tsk->sighand->siglock, flags); 1763 ret = dequeue_signal(tsk, mask, info); 1764 spin_unlock_irqrestore(&tsk->sighand->siglock, flags); 1765 1766 return ret; 1767} 1768 1769extern void block_all_signals(int (*notifier)(void *priv), void *priv, 1770 sigset_t *mask); 1771extern void unblock_all_signals(void); 1772extern void release_task(struct task_struct * p); 1773extern int send_sig_info(int, struct siginfo *, struct task_struct *); 1774extern int force_sigsegv(int, struct task_struct *); 1775extern int force_sig_info(int, struct siginfo *, struct task_struct *); 1776extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); 1777extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid); 1778extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32); 1779extern int kill_pgrp(struct pid *pid, int sig, int priv); 1780extern int kill_pid(struct pid *pid, int sig, int priv); 1781extern int kill_proc_info(int, struct siginfo *, pid_t); 1782extern void do_notify_parent(struct task_struct *, int); 1783extern void force_sig(int, struct task_struct *); 1784extern void force_sig_specific(int, struct task_struct *); 1785extern int send_sig(int, struct task_struct *, int); 1786extern void zap_other_threads(struct task_struct *p); 1787extern int kill_proc(pid_t, int, int); 1788extern struct sigqueue *sigqueue_alloc(void); 1789extern void sigqueue_free(struct sigqueue *); 1790extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group); 1791extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *); 1792extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long); 1793 1794static inline int kill_cad_pid(int sig, int priv) 1795{ 1796 return kill_pid(cad_pid, sig, priv); 1797} 1798 1799/* These can be the second arg to send_sig_info/send_group_sig_info. */ 1800#define SEND_SIG_NOINFO ((struct siginfo *) 0) 1801#define SEND_SIG_PRIV ((struct siginfo *) 1) 1802#define SEND_SIG_FORCED ((struct siginfo *) 2) 1803 1804static inline int is_si_special(const struct siginfo *info) 1805{ 1806 return info <= SEND_SIG_FORCED; 1807} 1808 1809/* True if we are on the alternate signal stack. */ 1810 1811static inline int on_sig_stack(unsigned long sp) 1812{ 1813 return (sp - current->sas_ss_sp < current->sas_ss_size); 1814} 1815 1816static inline int sas_ss_flags(unsigned long sp) 1817{ 1818 return (current->sas_ss_size == 0 ? SS_DISABLE 1819 : on_sig_stack(sp) ? SS_ONSTACK : 0); 1820} 1821 1822/* 1823 * Routines for handling mm_structs 1824 */ 1825extern struct mm_struct * mm_alloc(void); 1826 1827/* mmdrop drops the mm and the page tables */ 1828extern void __mmdrop(struct mm_struct *); 1829static inline void mmdrop(struct mm_struct * mm) 1830{ 1831 if (unlikely(atomic_dec_and_test(&mm->mm_count))) 1832 __mmdrop(mm); 1833} 1834 1835/* mmput gets rid of the mappings and all user-space */ 1836extern void mmput(struct mm_struct *); 1837/* Grab a reference to a task's mm, if it is not already going away */ 1838extern struct mm_struct *get_task_mm(struct task_struct *task); 1839/* Remove the current tasks stale references to the old mm_struct */ 1840extern void mm_release(struct task_struct *, struct mm_struct *); 1841/* Allocate a new mm structure and copy contents from tsk->mm */ 1842extern struct mm_struct *dup_mm(struct task_struct *tsk); 1843 1844extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *); 1845extern void flush_thread(void); 1846extern void exit_thread(void); 1847 1848extern void exit_files(struct task_struct *); 1849extern void __cleanup_signal(struct signal_struct *); 1850extern void __cleanup_sighand(struct sighand_struct *); 1851extern void exit_itimers(struct signal_struct *); 1852 1853extern NORET_TYPE void do_group_exit(int); 1854 1855extern void daemonize(const char *, ...); 1856extern int allow_signal(int); 1857extern int disallow_signal(int); 1858 1859extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *); 1860extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *); 1861struct task_struct *fork_idle(int); 1862 1863extern void set_task_comm(struct task_struct *tsk, char *from); 1864extern char *get_task_comm(char *to, struct task_struct *tsk); 1865 1866#ifdef CONFIG_SMP 1867extern void wait_task_inactive(struct task_struct * p); 1868#else 1869#define wait_task_inactive(p) do { } while (0) 1870#endif 1871 1872#define remove_parent(p) list_del_init(&(p)->sibling) 1873#define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children) 1874 1875#define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks) 1876 1877#define for_each_process(p) \ 1878 for (p = &init_task ; (p = next_task(p)) != &init_task ; ) 1879 1880/* 1881 * Careful: do_each_thread/while_each_thread is a double loop so 1882 * 'break' will not work as expected - use goto instead. 1883 */ 1884#define do_each_thread(g, t) \ 1885 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do 1886 1887#define while_each_thread(g, t) \ 1888 while ((t = next_thread(t)) != g) 1889 1890/* de_thread depends on thread_group_leader not being a pid based check */ 1891#define thread_group_leader(p) (p == p->group_leader) 1892 1893/* Do to the insanities of de_thread it is possible for a process 1894 * to have the pid of the thread group leader without actually being 1895 * the thread group leader. For iteration through the pids in proc 1896 * all we care about is that we have a task with the appropriate 1897 * pid, we don't actually care if we have the right task. 1898 */ 1899static inline int has_group_leader_pid(struct task_struct *p) 1900{ 1901 return p->pid == p->tgid; 1902} 1903 1904static inline 1905int same_thread_group(struct task_struct *p1, struct task_struct *p2) 1906{ 1907 return p1->tgid == p2->tgid; 1908} 1909 1910static inline struct task_struct *next_thread(const struct task_struct *p) 1911{ 1912 return list_entry(rcu_dereference(p->thread_group.next), 1913 struct task_struct, thread_group); 1914} 1915 1916static inline int thread_group_empty(struct task_struct *p) 1917{ 1918 return list_empty(&p->thread_group); 1919} 1920 1921#define delay_group_leader(p) \ 1922 (thread_group_leader(p) && !thread_group_empty(p)) 1923 1924/* 1925 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring 1926 * subscriptions and synchronises with wait4(). Also used in procfs. Also 1927 * pins the final release of task.io_context. Also protects ->cpuset and 1928 * ->cgroup.subsys[]. 1929 * 1930 * Nests both inside and outside of read_lock(&tasklist_lock). 1931 * It must not be nested with write_lock_irq(&tasklist_lock), 1932 * neither inside nor outside. 1933 */ 1934static inline void task_lock(struct task_struct *p) 1935{ 1936 spin_lock(&p->alloc_lock); 1937} 1938 1939static inline void task_unlock(struct task_struct *p) 1940{ 1941 spin_unlock(&p->alloc_lock); 1942} 1943 1944extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk, 1945 unsigned long *flags); 1946 1947static inline void unlock_task_sighand(struct task_struct *tsk, 1948 unsigned long *flags) 1949{ 1950 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags); 1951} 1952 1953#ifndef __HAVE_THREAD_FUNCTIONS 1954 1955#define task_thread_info(task) ((struct thread_info *)(task)->stack) 1956#define task_stack_page(task) ((task)->stack) 1957 1958static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org) 1959{ 1960 *task_thread_info(p) = *task_thread_info(org); 1961 task_thread_info(p)->task = p; 1962} 1963 1964static inline unsigned long *end_of_stack(struct task_struct *p) 1965{ 1966 return (unsigned long *)(task_thread_info(p) + 1); 1967} 1968 1969#endif 1970 1971extern void thread_info_cache_init(void); 1972 1973/* set thread flags in other task's structures 1974 * - see asm/thread_info.h for TIF_xxxx flags available 1975 */ 1976static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) 1977{ 1978 set_ti_thread_flag(task_thread_info(tsk), flag); 1979} 1980 1981static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) 1982{ 1983 clear_ti_thread_flag(task_thread_info(tsk), flag); 1984} 1985 1986static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) 1987{ 1988 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); 1989} 1990 1991static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) 1992{ 1993 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); 1994} 1995 1996static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) 1997{ 1998 return test_ti_thread_flag(task_thread_info(tsk), flag); 1999} 2000 2001static inline void set_tsk_need_resched(struct task_struct *tsk) 2002{ 2003 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2004} 2005 2006static inline void clear_tsk_need_resched(struct task_struct *tsk) 2007{ 2008 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2009} 2010 2011static inline int test_tsk_need_resched(struct task_struct *tsk) 2012{ 2013 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); 2014} 2015 2016static inline int signal_pending(struct task_struct *p) 2017{ 2018 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING)); 2019} 2020 2021extern int __fatal_signal_pending(struct task_struct *p); 2022 2023static inline int fatal_signal_pending(struct task_struct *p) 2024{ 2025 return signal_pending(p) && __fatal_signal_pending(p); 2026} 2027 2028static inline int need_resched(void) 2029{ 2030 return unlikely(test_thread_flag(TIF_NEED_RESCHED)); 2031} 2032 2033/* 2034 * cond_resched() and cond_resched_lock(): latency reduction via 2035 * explicit rescheduling in places that are safe. The return 2036 * value indicates whether a reschedule was done in fact. 2037 * cond_resched_lock() will drop the spinlock before scheduling, 2038 * cond_resched_softirq() will enable bhs before scheduling. 2039 */ 2040extern int _cond_resched(void); 2041#ifdef CONFIG_PREEMPT_BKL 2042static inline int cond_resched(void) 2043{ 2044 return 0; 2045} 2046#else 2047static inline int cond_resched(void) 2048{ 2049 return _cond_resched(); 2050} 2051#endif 2052extern int cond_resched_lock(spinlock_t * lock); 2053extern int cond_resched_softirq(void); 2054static inline int cond_resched_bkl(void) 2055{ 2056 return _cond_resched(); 2057} 2058 2059/* 2060 * Does a critical section need to be broken due to another 2061 * task waiting?: (technically does not depend on CONFIG_PREEMPT, 2062 * but a general need for low latency) 2063 */ 2064static inline int spin_needbreak(spinlock_t *lock) 2065{ 2066#ifdef CONFIG_PREEMPT 2067 return spin_is_contended(lock); 2068#else 2069 return 0; 2070#endif 2071} 2072 2073/* 2074 * Reevaluate whether the task has signals pending delivery. 2075 * Wake the task if so. 2076 * This is required every time the blocked sigset_t changes. 2077 * callers must hold sighand->siglock. 2078 */ 2079extern void recalc_sigpending_and_wake(struct task_struct *t); 2080extern void recalc_sigpending(void); 2081 2082extern void signal_wake_up(struct task_struct *t, int resume_stopped); 2083 2084/* 2085 * Wrappers for p->thread_info->cpu access. No-op on UP. 2086 */ 2087#ifdef CONFIG_SMP 2088 2089static inline unsigned int task_cpu(const struct task_struct *p) 2090{ 2091 return task_thread_info(p)->cpu; 2092} 2093 2094extern void set_task_cpu(struct task_struct *p, unsigned int cpu); 2095 2096#else 2097 2098static inline unsigned int task_cpu(const struct task_struct *p) 2099{ 2100 return 0; 2101} 2102 2103static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 2104{ 2105} 2106 2107#endif /* CONFIG_SMP */ 2108 2109#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 2110extern void arch_pick_mmap_layout(struct mm_struct *mm); 2111#else 2112static inline void arch_pick_mmap_layout(struct mm_struct *mm) 2113{ 2114 mm->mmap_base = TASK_UNMAPPED_BASE; 2115 mm->get_unmapped_area = arch_get_unmapped_area; 2116 mm->unmap_area = arch_unmap_area; 2117} 2118#endif 2119 2120extern long sched_setaffinity(pid_t pid, const cpumask_t *new_mask); 2121extern long sched_getaffinity(pid_t pid, cpumask_t *mask); 2122 2123extern int sched_mc_power_savings, sched_smt_power_savings; 2124 2125extern void normalize_rt_tasks(void); 2126 2127#ifdef CONFIG_GROUP_SCHED 2128 2129extern struct task_group init_task_group; 2130#ifdef CONFIG_USER_SCHED 2131extern struct task_group root_task_group; 2132#endif 2133 2134extern struct task_group *sched_create_group(struct task_group *parent); 2135extern void sched_destroy_group(struct task_group *tg); 2136extern void sched_move_task(struct task_struct *tsk); 2137#ifdef CONFIG_FAIR_GROUP_SCHED 2138extern int sched_group_set_shares(struct task_group *tg, unsigned long shares); 2139extern unsigned long sched_group_shares(struct task_group *tg); 2140#endif 2141#ifdef CONFIG_RT_GROUP_SCHED 2142extern int sched_group_set_rt_runtime(struct task_group *tg, 2143 long rt_runtime_us); 2144extern long sched_group_rt_runtime(struct task_group *tg); 2145extern int sched_group_set_rt_period(struct task_group *tg, 2146 long rt_period_us); 2147extern long sched_group_rt_period(struct task_group *tg); 2148#endif 2149#endif 2150 2151#ifdef CONFIG_TASK_XACCT 2152static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2153{ 2154 tsk->rchar += amt; 2155} 2156 2157static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2158{ 2159 tsk->wchar += amt; 2160} 2161 2162static inline void inc_syscr(struct task_struct *tsk) 2163{ 2164 tsk->syscr++; 2165} 2166 2167static inline void inc_syscw(struct task_struct *tsk) 2168{ 2169 tsk->syscw++; 2170} 2171#else 2172static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2173{ 2174} 2175 2176static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2177{ 2178} 2179 2180static inline void inc_syscr(struct task_struct *tsk) 2181{ 2182} 2183 2184static inline void inc_syscw(struct task_struct *tsk) 2185{ 2186} 2187#endif 2188 2189#ifdef CONFIG_SMP 2190void migration_init(void); 2191#else 2192static inline void migration_init(void) 2193{ 2194} 2195#endif 2196 2197#ifndef TASK_SIZE_OF 2198#define TASK_SIZE_OF(tsk) TASK_SIZE 2199#endif 2200 2201#ifdef CONFIG_MM_OWNER 2202extern void mm_update_next_owner(struct mm_struct *mm); 2203extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p); 2204#else 2205static inline void mm_update_next_owner(struct mm_struct *mm) 2206{ 2207} 2208 2209static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p) 2210{ 2211} 2212#endif /* CONFIG_MM_OWNER */ 2213 2214#endif /* __KERNEL__ */ 2215 2216#endif 2217