sched.h revision 7d8e23df69820e6be42bcc41d441f4860e8c76f7
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#include <linux/cred.h> 91 92#include <asm/processor.h> 93 94struct mem_cgroup; 95struct exec_domain; 96struct futex_pi_state; 97struct robust_list_head; 98struct bio; 99struct bts_tracer; 100 101/* 102 * List of flags we want to share for kernel threads, 103 * if only because they are not used by them anyway. 104 */ 105#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND) 106 107/* 108 * These are the constant used to fake the fixed-point load-average 109 * counting. Some notes: 110 * - 11 bit fractions expand to 22 bits by the multiplies: this gives 111 * a load-average precision of 10 bits integer + 11 bits fractional 112 * - if you want to count load-averages more often, you need more 113 * precision, or rounding will get you. With 2-second counting freq, 114 * the EXP_n values would be 1981, 2034 and 2043 if still using only 115 * 11 bit fractions. 116 */ 117extern unsigned long avenrun[]; /* Load averages */ 118 119#define FSHIFT 11 /* nr of bits of precision */ 120#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */ 121#define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */ 122#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */ 123#define EXP_5 2014 /* 1/exp(5sec/5min) */ 124#define EXP_15 2037 /* 1/exp(5sec/15min) */ 125 126#define CALC_LOAD(load,exp,n) \ 127 load *= exp; \ 128 load += n*(FIXED_1-exp); \ 129 load >>= FSHIFT; 130 131extern unsigned long total_forks; 132extern int nr_threads; 133DECLARE_PER_CPU(unsigned long, process_counts); 134extern int nr_processes(void); 135extern unsigned long nr_running(void); 136extern unsigned long nr_uninterruptible(void); 137extern unsigned long nr_active(void); 138extern unsigned long nr_iowait(void); 139 140struct seq_file; 141struct cfs_rq; 142struct task_group; 143#ifdef CONFIG_SCHED_DEBUG 144extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m); 145extern void proc_sched_set_task(struct task_struct *p); 146extern void 147print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq); 148#else 149static inline void 150proc_sched_show_task(struct task_struct *p, struct seq_file *m) 151{ 152} 153static inline void proc_sched_set_task(struct task_struct *p) 154{ 155} 156static inline void 157print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) 158{ 159} 160#endif 161 162extern unsigned long long time_sync_thresh; 163 164/* 165 * Task state bitmask. NOTE! These bits are also 166 * encoded in fs/proc/array.c: get_task_state(). 167 * 168 * We have two separate sets of flags: task->state 169 * is about runnability, while task->exit_state are 170 * about the task exiting. Confusing, but this way 171 * modifying one set can't modify the other one by 172 * mistake. 173 */ 174#define TASK_RUNNING 0 175#define TASK_INTERRUPTIBLE 1 176#define TASK_UNINTERRUPTIBLE 2 177#define __TASK_STOPPED 4 178#define __TASK_TRACED 8 179/* in tsk->exit_state */ 180#define EXIT_ZOMBIE 16 181#define EXIT_DEAD 32 182/* in tsk->state again */ 183#define TASK_DEAD 64 184#define TASK_WAKEKILL 128 185 186/* Convenience macros for the sake of set_task_state */ 187#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) 188#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) 189#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED) 190 191/* Convenience macros for the sake of wake_up */ 192#define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) 193#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED) 194 195/* get_task_state() */ 196#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ 197 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ 198 __TASK_TRACED) 199 200#define task_is_traced(task) ((task->state & __TASK_TRACED) != 0) 201#define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0) 202#define task_is_stopped_or_traced(task) \ 203 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0) 204#define task_contributes_to_load(task) \ 205 ((task->state & TASK_UNINTERRUPTIBLE) != 0) 206 207#define __set_task_state(tsk, state_value) \ 208 do { (tsk)->state = (state_value); } while (0) 209#define set_task_state(tsk, state_value) \ 210 set_mb((tsk)->state, (state_value)) 211 212/* 213 * set_current_state() includes a barrier so that the write of current->state 214 * is correctly serialised wrt the caller's subsequent test of whether to 215 * actually sleep: 216 * 217 * set_current_state(TASK_UNINTERRUPTIBLE); 218 * if (do_i_need_to_sleep()) 219 * schedule(); 220 * 221 * If the caller does not need such serialisation then use __set_current_state() 222 */ 223#define __set_current_state(state_value) \ 224 do { current->state = (state_value); } while (0) 225#define set_current_state(state_value) \ 226 set_mb(current->state, (state_value)) 227 228/* Task command name length */ 229#define TASK_COMM_LEN 16 230 231#include <linux/spinlock.h> 232 233/* 234 * This serializes "schedule()" and also protects 235 * the run-queue from deletions/modifications (but 236 * _adding_ to the beginning of the run-queue has 237 * a separate lock). 238 */ 239extern rwlock_t tasklist_lock; 240extern spinlock_t mmlist_lock; 241 242struct task_struct; 243 244extern void sched_init(void); 245extern void sched_init_smp(void); 246extern asmlinkage void schedule_tail(struct task_struct *prev); 247extern void init_idle(struct task_struct *idle, int cpu); 248extern void init_idle_bootup_task(struct task_struct *idle); 249 250extern int runqueue_is_locked(void); 251extern void task_rq_unlock_wait(struct task_struct *p); 252 253extern cpumask_var_t nohz_cpu_mask; 254#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ) 255extern int select_nohz_load_balancer(int cpu); 256#else 257static inline int select_nohz_load_balancer(int cpu) 258{ 259 return 0; 260} 261#endif 262 263/* 264 * Only dump TASK_* tasks. (0 for all tasks) 265 */ 266extern void show_state_filter(unsigned long state_filter); 267 268static inline void show_state(void) 269{ 270 show_state_filter(0); 271} 272 273extern void show_regs(struct pt_regs *); 274 275/* 276 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current 277 * task), SP is the stack pointer of the first frame that should be shown in the back 278 * trace (or NULL if the entire call-chain of the task should be shown). 279 */ 280extern void show_stack(struct task_struct *task, unsigned long *sp); 281 282void io_schedule(void); 283long io_schedule_timeout(long timeout); 284 285extern void cpu_init (void); 286extern void trap_init(void); 287extern void update_process_times(int user); 288extern void scheduler_tick(void); 289 290extern void sched_show_task(struct task_struct *p); 291 292#ifdef CONFIG_DETECT_SOFTLOCKUP 293extern void softlockup_tick(void); 294extern void touch_softlockup_watchdog(void); 295extern void touch_all_softlockup_watchdogs(void); 296extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write, 297 struct file *filp, void __user *buffer, 298 size_t *lenp, loff_t *ppos); 299extern unsigned int softlockup_panic; 300extern unsigned long sysctl_hung_task_check_count; 301extern unsigned long sysctl_hung_task_timeout_secs; 302extern unsigned long sysctl_hung_task_warnings; 303extern int softlockup_thresh; 304#else 305static inline void softlockup_tick(void) 306{ 307} 308static inline void spawn_softlockup_task(void) 309{ 310} 311static inline void touch_softlockup_watchdog(void) 312{ 313} 314static inline void touch_all_softlockup_watchdogs(void) 315{ 316} 317#endif 318 319 320/* Attach to any functions which should be ignored in wchan output. */ 321#define __sched __attribute__((__section__(".sched.text"))) 322 323/* Linker adds these: start and end of __sched functions */ 324extern char __sched_text_start[], __sched_text_end[]; 325 326/* Is this address in the __sched functions? */ 327extern int in_sched_functions(unsigned long addr); 328 329#define MAX_SCHEDULE_TIMEOUT LONG_MAX 330extern signed long schedule_timeout(signed long timeout); 331extern signed long schedule_timeout_interruptible(signed long timeout); 332extern signed long schedule_timeout_killable(signed long timeout); 333extern signed long schedule_timeout_uninterruptible(signed long timeout); 334asmlinkage void schedule(void); 335 336struct nsproxy; 337struct user_namespace; 338 339/* Maximum number of active map areas.. This is a random (large) number */ 340#define DEFAULT_MAX_MAP_COUNT 65536 341 342extern int sysctl_max_map_count; 343 344#include <linux/aio.h> 345 346extern unsigned long 347arch_get_unmapped_area(struct file *, unsigned long, unsigned long, 348 unsigned long, unsigned long); 349extern unsigned long 350arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr, 351 unsigned long len, unsigned long pgoff, 352 unsigned long flags); 353extern void arch_unmap_area(struct mm_struct *, unsigned long); 354extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long); 355 356#if USE_SPLIT_PTLOCKS 357/* 358 * The mm counters are not protected by its page_table_lock, 359 * so must be incremented atomically. 360 */ 361#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value) 362#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member)) 363#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member) 364#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member) 365#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member) 366 367#else /* !USE_SPLIT_PTLOCKS */ 368/* 369 * The mm counters are protected by its page_table_lock, 370 * so can be incremented directly. 371 */ 372#define set_mm_counter(mm, member, value) (mm)->_##member = (value) 373#define get_mm_counter(mm, member) ((mm)->_##member) 374#define add_mm_counter(mm, member, value) (mm)->_##member += (value) 375#define inc_mm_counter(mm, member) (mm)->_##member++ 376#define dec_mm_counter(mm, member) (mm)->_##member-- 377 378#endif /* !USE_SPLIT_PTLOCKS */ 379 380#define get_mm_rss(mm) \ 381 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss)) 382#define update_hiwater_rss(mm) do { \ 383 unsigned long _rss = get_mm_rss(mm); \ 384 if ((mm)->hiwater_rss < _rss) \ 385 (mm)->hiwater_rss = _rss; \ 386} while (0) 387#define update_hiwater_vm(mm) do { \ 388 if ((mm)->hiwater_vm < (mm)->total_vm) \ 389 (mm)->hiwater_vm = (mm)->total_vm; \ 390} while (0) 391 392#define get_mm_hiwater_rss(mm) max((mm)->hiwater_rss, get_mm_rss(mm)) 393#define get_mm_hiwater_vm(mm) max((mm)->hiwater_vm, (mm)->total_vm) 394 395extern void set_dumpable(struct mm_struct *mm, int value); 396extern int get_dumpable(struct mm_struct *mm); 397 398/* mm flags */ 399/* dumpable bits */ 400#define MMF_DUMPABLE 0 /* core dump is permitted */ 401#define MMF_DUMP_SECURELY 1 /* core file is readable only by root */ 402#define MMF_DUMPABLE_BITS 2 403 404/* coredump filter bits */ 405#define MMF_DUMP_ANON_PRIVATE 2 406#define MMF_DUMP_ANON_SHARED 3 407#define MMF_DUMP_MAPPED_PRIVATE 4 408#define MMF_DUMP_MAPPED_SHARED 5 409#define MMF_DUMP_ELF_HEADERS 6 410#define MMF_DUMP_HUGETLB_PRIVATE 7 411#define MMF_DUMP_HUGETLB_SHARED 8 412#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS 413#define MMF_DUMP_FILTER_BITS 7 414#define MMF_DUMP_FILTER_MASK \ 415 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT) 416#define MMF_DUMP_FILTER_DEFAULT \ 417 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\ 418 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF) 419 420#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS 421# define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS) 422#else 423# define MMF_DUMP_MASK_DEFAULT_ELF 0 424#endif 425 426struct sighand_struct { 427 atomic_t count; 428 struct k_sigaction action[_NSIG]; 429 spinlock_t siglock; 430 wait_queue_head_t signalfd_wqh; 431}; 432 433struct pacct_struct { 434 int ac_flag; 435 long ac_exitcode; 436 unsigned long ac_mem; 437 cputime_t ac_utime, ac_stime; 438 unsigned long ac_minflt, ac_majflt; 439}; 440 441/** 442 * struct task_cputime - collected CPU time counts 443 * @utime: time spent in user mode, in &cputime_t units 444 * @stime: time spent in kernel mode, in &cputime_t units 445 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds 446 * 447 * This structure groups together three kinds of CPU time that are 448 * tracked for threads and thread groups. Most things considering 449 * CPU time want to group these counts together and treat all three 450 * of them in parallel. 451 */ 452struct task_cputime { 453 cputime_t utime; 454 cputime_t stime; 455 unsigned long long sum_exec_runtime; 456}; 457/* Alternate field names when used to cache expirations. */ 458#define prof_exp stime 459#define virt_exp utime 460#define sched_exp sum_exec_runtime 461 462#define INIT_CPUTIME \ 463 (struct task_cputime) { \ 464 .utime = cputime_zero, \ 465 .stime = cputime_zero, \ 466 .sum_exec_runtime = 0, \ 467 } 468 469/** 470 * struct thread_group_cputimer - thread group interval timer counts 471 * @cputime: thread group interval timers. 472 * @running: non-zero when there are timers running and 473 * @cputime receives updates. 474 * @lock: lock for fields in this struct. 475 * 476 * This structure contains the version of task_cputime, above, that is 477 * used for thread group CPU timer calculations. 478 */ 479struct thread_group_cputimer { 480 struct task_cputime cputime; 481 int running; 482 spinlock_t lock; 483}; 484 485/* 486 * NOTE! "signal_struct" does not have it's own 487 * locking, because a shared signal_struct always 488 * implies a shared sighand_struct, so locking 489 * sighand_struct is always a proper superset of 490 * the locking of signal_struct. 491 */ 492struct signal_struct { 493 atomic_t count; 494 atomic_t live; 495 496 wait_queue_head_t wait_chldexit; /* for wait4() */ 497 498 /* current thread group signal load-balancing target: */ 499 struct task_struct *curr_target; 500 501 /* shared signal handling: */ 502 struct sigpending shared_pending; 503 504 /* thread group exit support */ 505 int group_exit_code; 506 /* overloaded: 507 * - notify group_exit_task when ->count is equal to notify_count 508 * - everyone except group_exit_task is stopped during signal delivery 509 * of fatal signals, group_exit_task processes the signal. 510 */ 511 int notify_count; 512 struct task_struct *group_exit_task; 513 514 /* thread group stop support, overloads group_exit_code too */ 515 int group_stop_count; 516 unsigned int flags; /* see SIGNAL_* flags below */ 517 518 /* POSIX.1b Interval Timers */ 519 struct list_head posix_timers; 520 521 /* ITIMER_REAL timer for the process */ 522 struct hrtimer real_timer; 523 struct pid *leader_pid; 524 ktime_t it_real_incr; 525 526 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */ 527 cputime_t it_prof_expires, it_virt_expires; 528 cputime_t it_prof_incr, it_virt_incr; 529 530 /* 531 * Thread group totals for process CPU timers. 532 * See thread_group_cputimer(), et al, for details. 533 */ 534 struct thread_group_cputimer cputimer; 535 536 /* Earliest-expiration cache. */ 537 struct task_cputime cputime_expires; 538 539 struct list_head cpu_timers[3]; 540 541 /* job control IDs */ 542 543 /* 544 * pgrp and session fields are deprecated. 545 * use the task_session_Xnr and task_pgrp_Xnr routines below 546 */ 547 548 union { 549 pid_t pgrp __deprecated; 550 pid_t __pgrp; 551 }; 552 553 struct pid *tty_old_pgrp; 554 555 union { 556 pid_t session __deprecated; 557 pid_t __session; 558 }; 559 560 /* boolean value for session group leader */ 561 int leader; 562 563 struct tty_struct *tty; /* NULL if no tty */ 564 565 /* 566 * Cumulative resource counters for dead threads in the group, 567 * and for reaped dead child processes forked by this group. 568 * Live threads maintain their own counters and add to these 569 * in __exit_signal, except for the group leader. 570 */ 571 cputime_t utime, stime, cutime, cstime; 572 cputime_t gtime; 573 cputime_t cgtime; 574 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw; 575 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt; 576 unsigned long inblock, oublock, cinblock, coublock; 577 struct task_io_accounting ioac; 578 579 /* 580 * Cumulative ns of schedule CPU time fo dead threads in the 581 * group, not including a zombie group leader, (This only differs 582 * from jiffies_to_ns(utime + stime) if sched_clock uses something 583 * other than jiffies.) 584 */ 585 unsigned long long sum_sched_runtime; 586 587 /* 588 * We don't bother to synchronize most readers of this at all, 589 * because there is no reader checking a limit that actually needs 590 * to get both rlim_cur and rlim_max atomically, and either one 591 * alone is a single word that can safely be read normally. 592 * getrlimit/setrlimit use task_lock(current->group_leader) to 593 * protect this instead of the siglock, because they really 594 * have no need to disable irqs. 595 */ 596 struct rlimit rlim[RLIM_NLIMITS]; 597 598#ifdef CONFIG_BSD_PROCESS_ACCT 599 struct pacct_struct pacct; /* per-process accounting information */ 600#endif 601#ifdef CONFIG_TASKSTATS 602 struct taskstats *stats; 603#endif 604#ifdef CONFIG_AUDIT 605 unsigned audit_tty; 606 struct tty_audit_buf *tty_audit_buf; 607#endif 608}; 609 610/* Context switch must be unlocked if interrupts are to be enabled */ 611#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW 612# define __ARCH_WANT_UNLOCKED_CTXSW 613#endif 614 615/* 616 * Bits in flags field of signal_struct. 617 */ 618#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */ 619#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */ 620#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */ 621#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */ 622/* 623 * Pending notifications to parent. 624 */ 625#define SIGNAL_CLD_STOPPED 0x00000010 626#define SIGNAL_CLD_CONTINUED 0x00000020 627#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED) 628 629#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */ 630 631/* If true, all threads except ->group_exit_task have pending SIGKILL */ 632static inline int signal_group_exit(const struct signal_struct *sig) 633{ 634 return (sig->flags & SIGNAL_GROUP_EXIT) || 635 (sig->group_exit_task != NULL); 636} 637 638/* 639 * Some day this will be a full-fledged user tracking system.. 640 */ 641struct user_struct { 642 atomic_t __count; /* reference count */ 643 atomic_t processes; /* How many processes does this user have? */ 644 atomic_t files; /* How many open files does this user have? */ 645 atomic_t sigpending; /* How many pending signals does this user have? */ 646#ifdef CONFIG_INOTIFY_USER 647 atomic_t inotify_watches; /* How many inotify watches does this user have? */ 648 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */ 649#endif 650#ifdef CONFIG_EPOLL 651 atomic_t epoll_watches; /* The number of file descriptors currently watched */ 652#endif 653#ifdef CONFIG_POSIX_MQUEUE 654 /* protected by mq_lock */ 655 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */ 656#endif 657 unsigned long locked_shm; /* How many pages of mlocked shm ? */ 658 659#ifdef CONFIG_KEYS 660 struct key *uid_keyring; /* UID specific keyring */ 661 struct key *session_keyring; /* UID's default session keyring */ 662#endif 663 664 /* Hash table maintenance information */ 665 struct hlist_node uidhash_node; 666 uid_t uid; 667 struct user_namespace *user_ns; 668 669#ifdef CONFIG_USER_SCHED 670 struct task_group *tg; 671#ifdef CONFIG_SYSFS 672 struct kobject kobj; 673 struct work_struct work; 674#endif 675#endif 676}; 677 678extern int uids_sysfs_init(void); 679 680extern struct user_struct *find_user(uid_t); 681 682extern struct user_struct root_user; 683#define INIT_USER (&root_user) 684 685 686struct backing_dev_info; 687struct reclaim_state; 688 689#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 690struct sched_info { 691 /* cumulative counters */ 692 unsigned long pcount; /* # of times run on this cpu */ 693 unsigned long long run_delay; /* time spent waiting on a runqueue */ 694 695 /* timestamps */ 696 unsigned long long last_arrival,/* when we last ran on a cpu */ 697 last_queued; /* when we were last queued to run */ 698#ifdef CONFIG_SCHEDSTATS 699 /* BKL stats */ 700 unsigned int bkl_count; 701#endif 702}; 703#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */ 704 705#ifdef CONFIG_TASK_DELAY_ACCT 706struct task_delay_info { 707 spinlock_t lock; 708 unsigned int flags; /* Private per-task flags */ 709 710 /* For each stat XXX, add following, aligned appropriately 711 * 712 * struct timespec XXX_start, XXX_end; 713 * u64 XXX_delay; 714 * u32 XXX_count; 715 * 716 * Atomicity of updates to XXX_delay, XXX_count protected by 717 * single lock above (split into XXX_lock if contention is an issue). 718 */ 719 720 /* 721 * XXX_count is incremented on every XXX operation, the delay 722 * associated with the operation is added to XXX_delay. 723 * XXX_delay contains the accumulated delay time in nanoseconds. 724 */ 725 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */ 726 u64 blkio_delay; /* wait for sync block io completion */ 727 u64 swapin_delay; /* wait for swapin block io completion */ 728 u32 blkio_count; /* total count of the number of sync block */ 729 /* io operations performed */ 730 u32 swapin_count; /* total count of the number of swapin block */ 731 /* io operations performed */ 732 733 struct timespec freepages_start, freepages_end; 734 u64 freepages_delay; /* wait for memory reclaim */ 735 u32 freepages_count; /* total count of memory reclaim */ 736}; 737#endif /* CONFIG_TASK_DELAY_ACCT */ 738 739static inline int sched_info_on(void) 740{ 741#ifdef CONFIG_SCHEDSTATS 742 return 1; 743#elif defined(CONFIG_TASK_DELAY_ACCT) 744 extern int delayacct_on; 745 return delayacct_on; 746#else 747 return 0; 748#endif 749} 750 751enum cpu_idle_type { 752 CPU_IDLE, 753 CPU_NOT_IDLE, 754 CPU_NEWLY_IDLE, 755 CPU_MAX_IDLE_TYPES 756}; 757 758/* 759 * sched-domains (multiprocessor balancing) declarations: 760 */ 761 762/* 763 * Increase resolution of nice-level calculations: 764 */ 765#define SCHED_LOAD_SHIFT 10 766#define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT) 767 768#define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE 769 770#ifdef CONFIG_SMP 771#define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */ 772#define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */ 773#define SD_BALANCE_EXEC 4 /* Balance on exec */ 774#define SD_BALANCE_FORK 8 /* Balance on fork, clone */ 775#define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */ 776#define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */ 777#define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */ 778#define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */ 779#define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */ 780#define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */ 781#define SD_SERIALIZE 1024 /* Only a single load balancing instance */ 782#define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */ 783 784enum powersavings_balance_level { 785 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */ 786 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package 787 * first for long running threads 788 */ 789 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle 790 * cpu package for power savings 791 */ 792 MAX_POWERSAVINGS_BALANCE_LEVELS 793}; 794 795extern int sched_mc_power_savings, sched_smt_power_savings; 796 797static inline int sd_balance_for_mc_power(void) 798{ 799 if (sched_smt_power_savings) 800 return SD_POWERSAVINGS_BALANCE; 801 802 return 0; 803} 804 805static inline int sd_balance_for_package_power(void) 806{ 807 if (sched_mc_power_savings | sched_smt_power_savings) 808 return SD_POWERSAVINGS_BALANCE; 809 810 return 0; 811} 812 813/* 814 * Optimise SD flags for power savings: 815 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings. 816 * Keep default SD flags if sched_{smt,mc}_power_saving=0 817 */ 818 819static inline int sd_power_saving_flags(void) 820{ 821 if (sched_mc_power_savings | sched_smt_power_savings) 822 return SD_BALANCE_NEWIDLE; 823 824 return 0; 825} 826 827struct sched_group { 828 struct sched_group *next; /* Must be a circular list */ 829 830 /* 831 * CPU power of this group, SCHED_LOAD_SCALE being max power for a 832 * single CPU. This is read only (except for setup, hotplug CPU). 833 * Note : Never change cpu_power without recompute its reciprocal 834 */ 835 unsigned int __cpu_power; 836 /* 837 * reciprocal value of cpu_power to avoid expensive divides 838 * (see include/linux/reciprocal_div.h) 839 */ 840 u32 reciprocal_cpu_power; 841 842 unsigned long cpumask[]; 843}; 844 845static inline struct cpumask *sched_group_cpus(struct sched_group *sg) 846{ 847 return to_cpumask(sg->cpumask); 848} 849 850enum sched_domain_level { 851 SD_LV_NONE = 0, 852 SD_LV_SIBLING, 853 SD_LV_MC, 854 SD_LV_CPU, 855 SD_LV_NODE, 856 SD_LV_ALLNODES, 857 SD_LV_MAX 858}; 859 860struct sched_domain_attr { 861 int relax_domain_level; 862}; 863 864#define SD_ATTR_INIT (struct sched_domain_attr) { \ 865 .relax_domain_level = -1, \ 866} 867 868struct sched_domain { 869 /* These fields must be setup */ 870 struct sched_domain *parent; /* top domain must be null terminated */ 871 struct sched_domain *child; /* bottom domain must be null terminated */ 872 struct sched_group *groups; /* the balancing groups of the domain */ 873 unsigned long min_interval; /* Minimum balance interval ms */ 874 unsigned long max_interval; /* Maximum balance interval ms */ 875 unsigned int busy_factor; /* less balancing by factor if busy */ 876 unsigned int imbalance_pct; /* No balance until over watermark */ 877 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */ 878 unsigned int busy_idx; 879 unsigned int idle_idx; 880 unsigned int newidle_idx; 881 unsigned int wake_idx; 882 unsigned int forkexec_idx; 883 int flags; /* See SD_* */ 884 enum sched_domain_level level; 885 886 /* Runtime fields. */ 887 unsigned long last_balance; /* init to jiffies. units in jiffies */ 888 unsigned int balance_interval; /* initialise to 1. units in ms. */ 889 unsigned int nr_balance_failed; /* initialise to 0 */ 890 891 u64 last_update; 892 893#ifdef CONFIG_SCHEDSTATS 894 /* load_balance() stats */ 895 unsigned int lb_count[CPU_MAX_IDLE_TYPES]; 896 unsigned int lb_failed[CPU_MAX_IDLE_TYPES]; 897 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES]; 898 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES]; 899 unsigned int lb_gained[CPU_MAX_IDLE_TYPES]; 900 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES]; 901 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES]; 902 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES]; 903 904 /* Active load balancing */ 905 unsigned int alb_count; 906 unsigned int alb_failed; 907 unsigned int alb_pushed; 908 909 /* SD_BALANCE_EXEC stats */ 910 unsigned int sbe_count; 911 unsigned int sbe_balanced; 912 unsigned int sbe_pushed; 913 914 /* SD_BALANCE_FORK stats */ 915 unsigned int sbf_count; 916 unsigned int sbf_balanced; 917 unsigned int sbf_pushed; 918 919 /* try_to_wake_up() stats */ 920 unsigned int ttwu_wake_remote; 921 unsigned int ttwu_move_affine; 922 unsigned int ttwu_move_balance; 923#endif 924#ifdef CONFIG_SCHED_DEBUG 925 char *name; 926#endif 927 928 /* span of all CPUs in this domain */ 929 unsigned long span[]; 930}; 931 932static inline struct cpumask *sched_domain_span(struct sched_domain *sd) 933{ 934 return to_cpumask(sd->span); 935} 936 937extern void partition_sched_domains(int ndoms_new, struct cpumask *doms_new, 938 struct sched_domain_attr *dattr_new); 939 940/* Test a flag in parent sched domain */ 941static inline int test_sd_parent(struct sched_domain *sd, int flag) 942{ 943 if (sd->parent && (sd->parent->flags & flag)) 944 return 1; 945 946 return 0; 947} 948 949#else /* CONFIG_SMP */ 950 951struct sched_domain_attr; 952 953static inline void 954partition_sched_domains(int ndoms_new, struct cpumask *doms_new, 955 struct sched_domain_attr *dattr_new) 956{ 957} 958#endif /* !CONFIG_SMP */ 959 960struct io_context; /* See blkdev.h */ 961 962 963#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK 964extern void prefetch_stack(struct task_struct *t); 965#else 966static inline void prefetch_stack(struct task_struct *t) { } 967#endif 968 969struct audit_context; /* See audit.c */ 970struct mempolicy; 971struct pipe_inode_info; 972struct uts_namespace; 973 974struct rq; 975struct sched_domain; 976 977struct sched_class { 978 const struct sched_class *next; 979 980 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup); 981 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep); 982 void (*yield_task) (struct rq *rq); 983 984 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int sync); 985 986 struct task_struct * (*pick_next_task) (struct rq *rq); 987 void (*put_prev_task) (struct rq *rq, struct task_struct *p); 988 989#ifdef CONFIG_SMP 990 int (*select_task_rq)(struct task_struct *p, int sync); 991 992 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu, 993 struct rq *busiest, unsigned long max_load_move, 994 struct sched_domain *sd, enum cpu_idle_type idle, 995 int *all_pinned, int *this_best_prio); 996 997 int (*move_one_task) (struct rq *this_rq, int this_cpu, 998 struct rq *busiest, struct sched_domain *sd, 999 enum cpu_idle_type idle); 1000 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task); 1001 void (*post_schedule) (struct rq *this_rq); 1002 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task); 1003 1004 void (*set_cpus_allowed)(struct task_struct *p, 1005 const struct cpumask *newmask); 1006 1007 void (*rq_online)(struct rq *rq); 1008 void (*rq_offline)(struct rq *rq); 1009#endif 1010 1011 void (*set_curr_task) (struct rq *rq); 1012 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued); 1013 void (*task_new) (struct rq *rq, struct task_struct *p); 1014 1015 void (*switched_from) (struct rq *this_rq, struct task_struct *task, 1016 int running); 1017 void (*switched_to) (struct rq *this_rq, struct task_struct *task, 1018 int running); 1019 void (*prio_changed) (struct rq *this_rq, struct task_struct *task, 1020 int oldprio, int running); 1021 1022#ifdef CONFIG_FAIR_GROUP_SCHED 1023 void (*moved_group) (struct task_struct *p); 1024#endif 1025}; 1026 1027struct load_weight { 1028 unsigned long weight, inv_weight; 1029}; 1030 1031/* 1032 * CFS stats for a schedulable entity (task, task-group etc) 1033 * 1034 * Current field usage histogram: 1035 * 1036 * 4 se->block_start 1037 * 4 se->run_node 1038 * 4 se->sleep_start 1039 * 6 se->load.weight 1040 */ 1041struct sched_entity { 1042 struct load_weight load; /* for load-balancing */ 1043 struct rb_node run_node; 1044 struct list_head group_node; 1045 unsigned int on_rq; 1046 1047 u64 exec_start; 1048 u64 sum_exec_runtime; 1049 u64 vruntime; 1050 u64 prev_sum_exec_runtime; 1051 1052 u64 last_wakeup; 1053 u64 avg_overlap; 1054 1055#ifdef CONFIG_SCHEDSTATS 1056 u64 wait_start; 1057 u64 wait_max; 1058 u64 wait_count; 1059 u64 wait_sum; 1060 1061 u64 sleep_start; 1062 u64 sleep_max; 1063 s64 sum_sleep_runtime; 1064 1065 u64 block_start; 1066 u64 block_max; 1067 u64 exec_max; 1068 u64 slice_max; 1069 1070 u64 nr_migrations; 1071 u64 nr_migrations_cold; 1072 u64 nr_failed_migrations_affine; 1073 u64 nr_failed_migrations_running; 1074 u64 nr_failed_migrations_hot; 1075 u64 nr_forced_migrations; 1076 u64 nr_forced2_migrations; 1077 1078 u64 nr_wakeups; 1079 u64 nr_wakeups_sync; 1080 u64 nr_wakeups_migrate; 1081 u64 nr_wakeups_local; 1082 u64 nr_wakeups_remote; 1083 u64 nr_wakeups_affine; 1084 u64 nr_wakeups_affine_attempts; 1085 u64 nr_wakeups_passive; 1086 u64 nr_wakeups_idle; 1087#endif 1088 1089#ifdef CONFIG_FAIR_GROUP_SCHED 1090 struct sched_entity *parent; 1091 /* rq on which this entity is (to be) queued: */ 1092 struct cfs_rq *cfs_rq; 1093 /* rq "owned" by this entity/group: */ 1094 struct cfs_rq *my_q; 1095#endif 1096}; 1097 1098struct sched_rt_entity { 1099 struct list_head run_list; 1100 unsigned long timeout; 1101 unsigned int time_slice; 1102 int nr_cpus_allowed; 1103 1104 struct sched_rt_entity *back; 1105#ifdef CONFIG_RT_GROUP_SCHED 1106 struct sched_rt_entity *parent; 1107 /* rq on which this entity is (to be) queued: */ 1108 struct rt_rq *rt_rq; 1109 /* rq "owned" by this entity/group: */ 1110 struct rt_rq *my_q; 1111#endif 1112}; 1113 1114struct task_struct { 1115 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */ 1116 void *stack; 1117 atomic_t usage; 1118 unsigned int flags; /* per process flags, defined below */ 1119 unsigned int ptrace; 1120 1121 int lock_depth; /* BKL lock depth */ 1122 1123#ifdef CONFIG_SMP 1124#ifdef __ARCH_WANT_UNLOCKED_CTXSW 1125 int oncpu; 1126#endif 1127#endif 1128 1129 int prio, static_prio, normal_prio; 1130 unsigned int rt_priority; 1131 const struct sched_class *sched_class; 1132 struct sched_entity se; 1133 struct sched_rt_entity rt; 1134 1135#ifdef CONFIG_PREEMPT_NOTIFIERS 1136 /* list of struct preempt_notifier: */ 1137 struct hlist_head preempt_notifiers; 1138#endif 1139 1140 /* 1141 * fpu_counter contains the number of consecutive context switches 1142 * that the FPU is used. If this is over a threshold, the lazy fpu 1143 * saving becomes unlazy to save the trap. This is an unsigned char 1144 * so that after 256 times the counter wraps and the behavior turns 1145 * lazy again; this to deal with bursty apps that only use FPU for 1146 * a short time 1147 */ 1148 unsigned char fpu_counter; 1149 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */ 1150#ifdef CONFIG_BLK_DEV_IO_TRACE 1151 unsigned int btrace_seq; 1152#endif 1153 1154 unsigned int policy; 1155 cpumask_t cpus_allowed; 1156 1157#ifdef CONFIG_PREEMPT_RCU 1158 int rcu_read_lock_nesting; 1159 int rcu_flipctr_idx; 1160#endif /* #ifdef CONFIG_PREEMPT_RCU */ 1161 1162#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 1163 struct sched_info sched_info; 1164#endif 1165 1166 struct list_head tasks; 1167 1168 struct mm_struct *mm, *active_mm; 1169 1170/* task state */ 1171 struct linux_binfmt *binfmt; 1172 int exit_state; 1173 int exit_code, exit_signal; 1174 int pdeath_signal; /* The signal sent when the parent dies */ 1175 /* ??? */ 1176 unsigned int personality; 1177 unsigned did_exec:1; 1178 pid_t pid; 1179 pid_t tgid; 1180 1181#ifdef CONFIG_CC_STACKPROTECTOR 1182 /* Canary value for the -fstack-protector gcc feature */ 1183 unsigned long stack_canary; 1184#endif 1185 /* 1186 * pointers to (original) parent process, youngest child, younger sibling, 1187 * older sibling, respectively. (p->father can be replaced with 1188 * p->real_parent->pid) 1189 */ 1190 struct task_struct *real_parent; /* real parent process */ 1191 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */ 1192 /* 1193 * children/sibling forms the list of my natural children 1194 */ 1195 struct list_head children; /* list of my children */ 1196 struct list_head sibling; /* linkage in my parent's children list */ 1197 struct task_struct *group_leader; /* threadgroup leader */ 1198 1199 /* 1200 * ptraced is the list of tasks this task is using ptrace on. 1201 * This includes both natural children and PTRACE_ATTACH targets. 1202 * p->ptrace_entry is p's link on the p->parent->ptraced list. 1203 */ 1204 struct list_head ptraced; 1205 struct list_head ptrace_entry; 1206 1207#ifdef CONFIG_X86_PTRACE_BTS 1208 /* 1209 * This is the tracer handle for the ptrace BTS extension. 1210 * This field actually belongs to the ptracer task. 1211 */ 1212 struct bts_tracer *bts; 1213 /* 1214 * The buffer to hold the BTS data. 1215 */ 1216 void *bts_buffer; 1217 size_t bts_size; 1218#endif /* CONFIG_X86_PTRACE_BTS */ 1219 1220 /* PID/PID hash table linkage. */ 1221 struct pid_link pids[PIDTYPE_MAX]; 1222 struct list_head thread_group; 1223 1224 struct completion *vfork_done; /* for vfork() */ 1225 int __user *set_child_tid; /* CLONE_CHILD_SETTID */ 1226 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ 1227 1228 cputime_t utime, stime, utimescaled, stimescaled; 1229 cputime_t gtime; 1230 cputime_t prev_utime, prev_stime; 1231 unsigned long nvcsw, nivcsw; /* context switch counts */ 1232 struct timespec start_time; /* monotonic time */ 1233 struct timespec real_start_time; /* boot based time */ 1234/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ 1235 unsigned long min_flt, maj_flt; 1236 1237 struct task_cputime cputime_expires; 1238 struct list_head cpu_timers[3]; 1239 1240/* process credentials */ 1241 const struct cred *real_cred; /* objective and real subjective task 1242 * credentials (COW) */ 1243 const struct cred *cred; /* effective (overridable) subjective task 1244 * credentials (COW) */ 1245 struct mutex cred_exec_mutex; /* execve vs ptrace cred calculation mutex */ 1246 1247 char comm[TASK_COMM_LEN]; /* executable name excluding path 1248 - access with [gs]et_task_comm (which lock 1249 it with task_lock()) 1250 - initialized normally by flush_old_exec */ 1251/* file system info */ 1252 int link_count, total_link_count; 1253#ifdef CONFIG_SYSVIPC 1254/* ipc stuff */ 1255 struct sysv_sem sysvsem; 1256#endif 1257#ifdef CONFIG_DETECT_SOFTLOCKUP 1258/* hung task detection */ 1259 unsigned long last_switch_timestamp; 1260 unsigned long last_switch_count; 1261#endif 1262/* CPU-specific state of this task */ 1263 struct thread_struct thread; 1264/* filesystem information */ 1265 struct fs_struct *fs; 1266/* open file information */ 1267 struct files_struct *files; 1268/* namespaces */ 1269 struct nsproxy *nsproxy; 1270/* signal handlers */ 1271 struct signal_struct *signal; 1272 struct sighand_struct *sighand; 1273 1274 sigset_t blocked, real_blocked; 1275 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */ 1276 struct sigpending pending; 1277 1278 unsigned long sas_ss_sp; 1279 size_t sas_ss_size; 1280 int (*notifier)(void *priv); 1281 void *notifier_data; 1282 sigset_t *notifier_mask; 1283 struct audit_context *audit_context; 1284#ifdef CONFIG_AUDITSYSCALL 1285 uid_t loginuid; 1286 unsigned int sessionid; 1287#endif 1288 seccomp_t seccomp; 1289 1290/* Thread group tracking */ 1291 u32 parent_exec_id; 1292 u32 self_exec_id; 1293/* Protection of (de-)allocation: mm, files, fs, tty, keyrings */ 1294 spinlock_t alloc_lock; 1295 1296 /* Protection of the PI data structures: */ 1297 spinlock_t pi_lock; 1298 1299#ifdef CONFIG_RT_MUTEXES 1300 /* PI waiters blocked on a rt_mutex held by this task */ 1301 struct plist_head pi_waiters; 1302 /* Deadlock detection and priority inheritance handling */ 1303 struct rt_mutex_waiter *pi_blocked_on; 1304#endif 1305 1306#ifdef CONFIG_DEBUG_MUTEXES 1307 /* mutex deadlock detection */ 1308 struct mutex_waiter *blocked_on; 1309#endif 1310#ifdef CONFIG_TRACE_IRQFLAGS 1311 unsigned int irq_events; 1312 int hardirqs_enabled; 1313 unsigned long hardirq_enable_ip; 1314 unsigned int hardirq_enable_event; 1315 unsigned long hardirq_disable_ip; 1316 unsigned int hardirq_disable_event; 1317 int softirqs_enabled; 1318 unsigned long softirq_disable_ip; 1319 unsigned int softirq_disable_event; 1320 unsigned long softirq_enable_ip; 1321 unsigned int softirq_enable_event; 1322 int hardirq_context; 1323 int softirq_context; 1324#endif 1325#ifdef CONFIG_LOCKDEP 1326# define MAX_LOCK_DEPTH 48UL 1327 u64 curr_chain_key; 1328 int lockdep_depth; 1329 unsigned int lockdep_recursion; 1330 struct held_lock held_locks[MAX_LOCK_DEPTH]; 1331#endif 1332 1333/* journalling filesystem info */ 1334 void *journal_info; 1335 1336/* stacked block device info */ 1337 struct bio *bio_list, **bio_tail; 1338 1339/* VM state */ 1340 struct reclaim_state *reclaim_state; 1341 1342 struct backing_dev_info *backing_dev_info; 1343 1344 struct io_context *io_context; 1345 1346 unsigned long ptrace_message; 1347 siginfo_t *last_siginfo; /* For ptrace use. */ 1348 struct task_io_accounting ioac; 1349#if defined(CONFIG_TASK_XACCT) 1350 u64 acct_rss_mem1; /* accumulated rss usage */ 1351 u64 acct_vm_mem1; /* accumulated virtual memory usage */ 1352 cputime_t acct_timexpd; /* stime + utime since last update */ 1353#endif 1354#ifdef CONFIG_CPUSETS 1355 nodemask_t mems_allowed; 1356 int cpuset_mems_generation; 1357 int cpuset_mem_spread_rotor; 1358#endif 1359#ifdef CONFIG_CGROUPS 1360 /* Control Group info protected by css_set_lock */ 1361 struct css_set *cgroups; 1362 /* cg_list protected by css_set_lock and tsk->alloc_lock */ 1363 struct list_head cg_list; 1364#endif 1365#ifdef CONFIG_FUTEX 1366 struct robust_list_head __user *robust_list; 1367#ifdef CONFIG_COMPAT 1368 struct compat_robust_list_head __user *compat_robust_list; 1369#endif 1370 struct list_head pi_state_list; 1371 struct futex_pi_state *pi_state_cache; 1372#endif 1373#ifdef CONFIG_NUMA 1374 struct mempolicy *mempolicy; 1375 short il_next; 1376#endif 1377 atomic_t fs_excl; /* holding fs exclusive resources */ 1378 struct rcu_head rcu; 1379 1380 /* 1381 * cache last used pipe for splice 1382 */ 1383 struct pipe_inode_info *splice_pipe; 1384#ifdef CONFIG_TASK_DELAY_ACCT 1385 struct task_delay_info *delays; 1386#endif 1387#ifdef CONFIG_FAULT_INJECTION 1388 int make_it_fail; 1389#endif 1390 struct prop_local_single dirties; 1391#ifdef CONFIG_LATENCYTOP 1392 int latency_record_count; 1393 struct latency_record latency_record[LT_SAVECOUNT]; 1394#endif 1395 /* 1396 * time slack values; these are used to round up poll() and 1397 * select() etc timeout values. These are in nanoseconds. 1398 */ 1399 unsigned long timer_slack_ns; 1400 unsigned long default_timer_slack_ns; 1401 1402 struct list_head *scm_work_list; 1403#ifdef CONFIG_FUNCTION_GRAPH_TRACER 1404 /* Index of current stored adress in ret_stack */ 1405 int curr_ret_stack; 1406 /* Stack of return addresses for return function tracing */ 1407 struct ftrace_ret_stack *ret_stack; 1408 /* 1409 * Number of functions that haven't been traced 1410 * because of depth overrun. 1411 */ 1412 atomic_t trace_overrun; 1413 /* Pause for the tracing */ 1414 atomic_t tracing_graph_pause; 1415#endif 1416#ifdef CONFIG_TRACING 1417 /* state flags for use by tracers */ 1418 unsigned long trace; 1419#endif 1420}; 1421 1422/* 1423 * Priority of a process goes from 0..MAX_PRIO-1, valid RT 1424 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH 1425 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority 1426 * values are inverted: lower p->prio value means higher priority. 1427 * 1428 * The MAX_USER_RT_PRIO value allows the actual maximum 1429 * RT priority to be separate from the value exported to 1430 * user-space. This allows kernel threads to set their 1431 * priority to a value higher than any user task. Note: 1432 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO. 1433 */ 1434 1435#define MAX_USER_RT_PRIO 100 1436#define MAX_RT_PRIO MAX_USER_RT_PRIO 1437 1438#define MAX_PRIO (MAX_RT_PRIO + 40) 1439#define DEFAULT_PRIO (MAX_RT_PRIO + 20) 1440 1441static inline int rt_prio(int prio) 1442{ 1443 if (unlikely(prio < MAX_RT_PRIO)) 1444 return 1; 1445 return 0; 1446} 1447 1448static inline int rt_task(struct task_struct *p) 1449{ 1450 return rt_prio(p->prio); 1451} 1452 1453static inline void set_task_session(struct task_struct *tsk, pid_t session) 1454{ 1455 tsk->signal->__session = session; 1456} 1457 1458static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp) 1459{ 1460 tsk->signal->__pgrp = pgrp; 1461} 1462 1463static inline struct pid *task_pid(struct task_struct *task) 1464{ 1465 return task->pids[PIDTYPE_PID].pid; 1466} 1467 1468static inline struct pid *task_tgid(struct task_struct *task) 1469{ 1470 return task->group_leader->pids[PIDTYPE_PID].pid; 1471} 1472 1473static inline struct pid *task_pgrp(struct task_struct *task) 1474{ 1475 return task->group_leader->pids[PIDTYPE_PGID].pid; 1476} 1477 1478static inline struct pid *task_session(struct task_struct *task) 1479{ 1480 return task->group_leader->pids[PIDTYPE_SID].pid; 1481} 1482 1483struct pid_namespace; 1484 1485/* 1486 * the helpers to get the task's different pids as they are seen 1487 * from various namespaces 1488 * 1489 * task_xid_nr() : global id, i.e. the id seen from the init namespace; 1490 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of 1491 * current. 1492 * task_xid_nr_ns() : id seen from the ns specified; 1493 * 1494 * set_task_vxid() : assigns a virtual id to a task; 1495 * 1496 * see also pid_nr() etc in include/linux/pid.h 1497 */ 1498 1499static inline pid_t task_pid_nr(struct task_struct *tsk) 1500{ 1501 return tsk->pid; 1502} 1503 1504pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1505 1506static inline pid_t task_pid_vnr(struct task_struct *tsk) 1507{ 1508 return pid_vnr(task_pid(tsk)); 1509} 1510 1511 1512static inline pid_t task_tgid_nr(struct task_struct *tsk) 1513{ 1514 return tsk->tgid; 1515} 1516 1517pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1518 1519static inline pid_t task_tgid_vnr(struct task_struct *tsk) 1520{ 1521 return pid_vnr(task_tgid(tsk)); 1522} 1523 1524 1525static inline pid_t task_pgrp_nr(struct task_struct *tsk) 1526{ 1527 return tsk->signal->__pgrp; 1528} 1529 1530pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1531 1532static inline pid_t task_pgrp_vnr(struct task_struct *tsk) 1533{ 1534 return pid_vnr(task_pgrp(tsk)); 1535} 1536 1537 1538static inline pid_t task_session_nr(struct task_struct *tsk) 1539{ 1540 return tsk->signal->__session; 1541} 1542 1543pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1544 1545static inline pid_t task_session_vnr(struct task_struct *tsk) 1546{ 1547 return pid_vnr(task_session(tsk)); 1548} 1549 1550 1551/** 1552 * pid_alive - check that a task structure is not stale 1553 * @p: Task structure to be checked. 1554 * 1555 * Test if a process is not yet dead (at most zombie state) 1556 * If pid_alive fails, then pointers within the task structure 1557 * can be stale and must not be dereferenced. 1558 */ 1559static inline int pid_alive(struct task_struct *p) 1560{ 1561 return p->pids[PIDTYPE_PID].pid != NULL; 1562} 1563 1564/** 1565 * is_global_init - check if a task structure is init 1566 * @tsk: Task structure to be checked. 1567 * 1568 * Check if a task structure is the first user space task the kernel created. 1569 */ 1570static inline int is_global_init(struct task_struct *tsk) 1571{ 1572 return tsk->pid == 1; 1573} 1574 1575/* 1576 * is_container_init: 1577 * check whether in the task is init in its own pid namespace. 1578 */ 1579extern int is_container_init(struct task_struct *tsk); 1580 1581extern struct pid *cad_pid; 1582 1583extern void free_task(struct task_struct *tsk); 1584#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0) 1585 1586extern void __put_task_struct(struct task_struct *t); 1587 1588static inline void put_task_struct(struct task_struct *t) 1589{ 1590 if (atomic_dec_and_test(&t->usage)) 1591 __put_task_struct(t); 1592} 1593 1594extern cputime_t task_utime(struct task_struct *p); 1595extern cputime_t task_stime(struct task_struct *p); 1596extern cputime_t task_gtime(struct task_struct *p); 1597 1598/* 1599 * Per process flags 1600 */ 1601#define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */ 1602 /* Not implemented yet, only for 486*/ 1603#define PF_STARTING 0x00000002 /* being created */ 1604#define PF_EXITING 0x00000004 /* getting shut down */ 1605#define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */ 1606#define PF_VCPU 0x00000010 /* I'm a virtual CPU */ 1607#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ 1608#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ 1609#define PF_DUMPCORE 0x00000200 /* dumped core */ 1610#define PF_SIGNALED 0x00000400 /* killed by a signal */ 1611#define PF_MEMALLOC 0x00000800 /* Allocating memory */ 1612#define PF_FLUSHER 0x00001000 /* responsible for disk writeback */ 1613#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */ 1614#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */ 1615#define PF_FROZEN 0x00010000 /* frozen for system suspend */ 1616#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ 1617#define PF_KSWAPD 0x00040000 /* I am kswapd */ 1618#define PF_SWAPOFF 0x00080000 /* I am in swapoff */ 1619#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ 1620#define PF_KTHREAD 0x00200000 /* I am a kernel thread */ 1621#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */ 1622#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ 1623#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */ 1624#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */ 1625#define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */ 1626#define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */ 1627#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ 1628#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */ 1629#define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */ 1630 1631/* 1632 * Only the _current_ task can read/write to tsk->flags, but other 1633 * tasks can access tsk->flags in readonly mode for example 1634 * with tsk_used_math (like during threaded core dumping). 1635 * There is however an exception to this rule during ptrace 1636 * or during fork: the ptracer task is allowed to write to the 1637 * child->flags of its traced child (same goes for fork, the parent 1638 * can write to the child->flags), because we're guaranteed the 1639 * child is not running and in turn not changing child->flags 1640 * at the same time the parent does it. 1641 */ 1642#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) 1643#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) 1644#define clear_used_math() clear_stopped_child_used_math(current) 1645#define set_used_math() set_stopped_child_used_math(current) 1646#define conditional_stopped_child_used_math(condition, child) \ 1647 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) 1648#define conditional_used_math(condition) \ 1649 conditional_stopped_child_used_math(condition, current) 1650#define copy_to_stopped_child_used_math(child) \ 1651 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) 1652/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ 1653#define tsk_used_math(p) ((p)->flags & PF_USED_MATH) 1654#define used_math() tsk_used_math(current) 1655 1656#ifdef CONFIG_SMP 1657extern int set_cpus_allowed_ptr(struct task_struct *p, 1658 const struct cpumask *new_mask); 1659#else 1660static inline int set_cpus_allowed_ptr(struct task_struct *p, 1661 const struct cpumask *new_mask) 1662{ 1663 if (!cpumask_test_cpu(0, new_mask)) 1664 return -EINVAL; 1665 return 0; 1666} 1667#endif 1668static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask) 1669{ 1670 return set_cpus_allowed_ptr(p, &new_mask); 1671} 1672 1673extern unsigned long long sched_clock(void); 1674 1675extern void sched_clock_init(void); 1676extern u64 sched_clock_cpu(int cpu); 1677 1678#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK 1679static inline void sched_clock_tick(void) 1680{ 1681} 1682 1683static inline void sched_clock_idle_sleep_event(void) 1684{ 1685} 1686 1687static inline void sched_clock_idle_wakeup_event(u64 delta_ns) 1688{ 1689} 1690#else 1691extern void sched_clock_tick(void); 1692extern void sched_clock_idle_sleep_event(void); 1693extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1694#endif 1695 1696/* 1697 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu 1698 * clock constructed from sched_clock(): 1699 */ 1700extern unsigned long long cpu_clock(int cpu); 1701 1702extern unsigned long long 1703task_sched_runtime(struct task_struct *task); 1704extern unsigned long long thread_group_sched_runtime(struct task_struct *task); 1705 1706/* sched_exec is called by processes performing an exec */ 1707#ifdef CONFIG_SMP 1708extern void sched_exec(void); 1709#else 1710#define sched_exec() {} 1711#endif 1712 1713extern void sched_clock_idle_sleep_event(void); 1714extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1715 1716#ifdef CONFIG_HOTPLUG_CPU 1717extern void idle_task_exit(void); 1718#else 1719static inline void idle_task_exit(void) {} 1720#endif 1721 1722extern void sched_idle_next(void); 1723 1724#if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP) 1725extern void wake_up_idle_cpu(int cpu); 1726#else 1727static inline void wake_up_idle_cpu(int cpu) { } 1728#endif 1729 1730extern unsigned int sysctl_sched_latency; 1731extern unsigned int sysctl_sched_min_granularity; 1732extern unsigned int sysctl_sched_wakeup_granularity; 1733extern unsigned int sysctl_sched_shares_ratelimit; 1734extern unsigned int sysctl_sched_shares_thresh; 1735#ifdef CONFIG_SCHED_DEBUG 1736extern unsigned int sysctl_sched_child_runs_first; 1737extern unsigned int sysctl_sched_features; 1738extern unsigned int sysctl_sched_migration_cost; 1739extern unsigned int sysctl_sched_nr_migrate; 1740 1741int sched_nr_latency_handler(struct ctl_table *table, int write, 1742 struct file *file, void __user *buffer, size_t *length, 1743 loff_t *ppos); 1744#endif 1745extern unsigned int sysctl_sched_rt_period; 1746extern int sysctl_sched_rt_runtime; 1747 1748int sched_rt_handler(struct ctl_table *table, int write, 1749 struct file *filp, void __user *buffer, size_t *lenp, 1750 loff_t *ppos); 1751 1752extern unsigned int sysctl_sched_compat_yield; 1753 1754#ifdef CONFIG_RT_MUTEXES 1755extern int rt_mutex_getprio(struct task_struct *p); 1756extern void rt_mutex_setprio(struct task_struct *p, int prio); 1757extern void rt_mutex_adjust_pi(struct task_struct *p); 1758#else 1759static inline int rt_mutex_getprio(struct task_struct *p) 1760{ 1761 return p->normal_prio; 1762} 1763# define rt_mutex_adjust_pi(p) do { } while (0) 1764#endif 1765 1766extern void set_user_nice(struct task_struct *p, long nice); 1767extern int task_prio(const struct task_struct *p); 1768extern int task_nice(const struct task_struct *p); 1769extern int can_nice(const struct task_struct *p, const int nice); 1770extern int task_curr(const struct task_struct *p); 1771extern int idle_cpu(int cpu); 1772extern int sched_setscheduler(struct task_struct *, int, struct sched_param *); 1773extern int sched_setscheduler_nocheck(struct task_struct *, int, 1774 struct sched_param *); 1775extern struct task_struct *idle_task(int cpu); 1776extern struct task_struct *curr_task(int cpu); 1777extern void set_curr_task(int cpu, struct task_struct *p); 1778 1779void yield(void); 1780 1781/* 1782 * The default (Linux) execution domain. 1783 */ 1784extern struct exec_domain default_exec_domain; 1785 1786union thread_union { 1787 struct thread_info thread_info; 1788 unsigned long stack[THREAD_SIZE/sizeof(long)]; 1789}; 1790 1791#ifndef __HAVE_ARCH_KSTACK_END 1792static inline int kstack_end(void *addr) 1793{ 1794 /* Reliable end of stack detection: 1795 * Some APM bios versions misalign the stack 1796 */ 1797 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*))); 1798} 1799#endif 1800 1801extern union thread_union init_thread_union; 1802extern struct task_struct init_task; 1803 1804extern struct mm_struct init_mm; 1805 1806extern struct pid_namespace init_pid_ns; 1807 1808/* 1809 * find a task by one of its numerical ids 1810 * 1811 * find_task_by_pid_type_ns(): 1812 * it is the most generic call - it finds a task by all id, 1813 * type and namespace specified 1814 * find_task_by_pid_ns(): 1815 * finds a task by its pid in the specified namespace 1816 * find_task_by_vpid(): 1817 * finds a task by its virtual pid 1818 * 1819 * see also find_vpid() etc in include/linux/pid.h 1820 */ 1821 1822extern struct task_struct *find_task_by_pid_type_ns(int type, int pid, 1823 struct pid_namespace *ns); 1824 1825extern struct task_struct *find_task_by_vpid(pid_t nr); 1826extern struct task_struct *find_task_by_pid_ns(pid_t nr, 1827 struct pid_namespace *ns); 1828 1829extern void __set_special_pids(struct pid *pid); 1830 1831/* per-UID process charging. */ 1832extern struct user_struct * alloc_uid(struct user_namespace *, uid_t); 1833static inline struct user_struct *get_uid(struct user_struct *u) 1834{ 1835 atomic_inc(&u->__count); 1836 return u; 1837} 1838extern void free_uid(struct user_struct *); 1839extern void release_uids(struct user_namespace *ns); 1840 1841#include <asm/current.h> 1842 1843extern void do_timer(unsigned long ticks); 1844 1845extern int wake_up_state(struct task_struct *tsk, unsigned int state); 1846extern int wake_up_process(struct task_struct *tsk); 1847extern void wake_up_new_task(struct task_struct *tsk, 1848 unsigned long clone_flags); 1849#ifdef CONFIG_SMP 1850 extern void kick_process(struct task_struct *tsk); 1851#else 1852 static inline void kick_process(struct task_struct *tsk) { } 1853#endif 1854extern void sched_fork(struct task_struct *p, int clone_flags); 1855extern void sched_dead(struct task_struct *p); 1856 1857extern void proc_caches_init(void); 1858extern void flush_signals(struct task_struct *); 1859extern void ignore_signals(struct task_struct *); 1860extern void flush_signal_handlers(struct task_struct *, int force_default); 1861extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info); 1862 1863static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info) 1864{ 1865 unsigned long flags; 1866 int ret; 1867 1868 spin_lock_irqsave(&tsk->sighand->siglock, flags); 1869 ret = dequeue_signal(tsk, mask, info); 1870 spin_unlock_irqrestore(&tsk->sighand->siglock, flags); 1871 1872 return ret; 1873} 1874 1875extern void block_all_signals(int (*notifier)(void *priv), void *priv, 1876 sigset_t *mask); 1877extern void unblock_all_signals(void); 1878extern void release_task(struct task_struct * p); 1879extern int send_sig_info(int, struct siginfo *, struct task_struct *); 1880extern int force_sigsegv(int, struct task_struct *); 1881extern int force_sig_info(int, struct siginfo *, struct task_struct *); 1882extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); 1883extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid); 1884extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32); 1885extern int kill_pgrp(struct pid *pid, int sig, int priv); 1886extern int kill_pid(struct pid *pid, int sig, int priv); 1887extern int kill_proc_info(int, struct siginfo *, pid_t); 1888extern int do_notify_parent(struct task_struct *, int); 1889extern void force_sig(int, struct task_struct *); 1890extern void force_sig_specific(int, struct task_struct *); 1891extern int send_sig(int, struct task_struct *, int); 1892extern void zap_other_threads(struct task_struct *p); 1893extern struct sigqueue *sigqueue_alloc(void); 1894extern void sigqueue_free(struct sigqueue *); 1895extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group); 1896extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *); 1897extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long); 1898 1899static inline int kill_cad_pid(int sig, int priv) 1900{ 1901 return kill_pid(cad_pid, sig, priv); 1902} 1903 1904/* These can be the second arg to send_sig_info/send_group_sig_info. */ 1905#define SEND_SIG_NOINFO ((struct siginfo *) 0) 1906#define SEND_SIG_PRIV ((struct siginfo *) 1) 1907#define SEND_SIG_FORCED ((struct siginfo *) 2) 1908 1909static inline int is_si_special(const struct siginfo *info) 1910{ 1911 return info <= SEND_SIG_FORCED; 1912} 1913 1914/* True if we are on the alternate signal stack. */ 1915 1916static inline int on_sig_stack(unsigned long sp) 1917{ 1918 return (sp - current->sas_ss_sp < current->sas_ss_size); 1919} 1920 1921static inline int sas_ss_flags(unsigned long sp) 1922{ 1923 return (current->sas_ss_size == 0 ? SS_DISABLE 1924 : on_sig_stack(sp) ? SS_ONSTACK : 0); 1925} 1926 1927/* 1928 * Routines for handling mm_structs 1929 */ 1930extern struct mm_struct * mm_alloc(void); 1931 1932/* mmdrop drops the mm and the page tables */ 1933extern void __mmdrop(struct mm_struct *); 1934static inline void mmdrop(struct mm_struct * mm) 1935{ 1936 if (unlikely(atomic_dec_and_test(&mm->mm_count))) 1937 __mmdrop(mm); 1938} 1939 1940/* mmput gets rid of the mappings and all user-space */ 1941extern void mmput(struct mm_struct *); 1942/* Grab a reference to a task's mm, if it is not already going away */ 1943extern struct mm_struct *get_task_mm(struct task_struct *task); 1944/* Remove the current tasks stale references to the old mm_struct */ 1945extern void mm_release(struct task_struct *, struct mm_struct *); 1946/* Allocate a new mm structure and copy contents from tsk->mm */ 1947extern struct mm_struct *dup_mm(struct task_struct *tsk); 1948 1949extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *); 1950extern void flush_thread(void); 1951extern void exit_thread(void); 1952 1953extern void exit_files(struct task_struct *); 1954extern void __cleanup_signal(struct signal_struct *); 1955extern void __cleanup_sighand(struct sighand_struct *); 1956 1957extern void exit_itimers(struct signal_struct *); 1958extern void flush_itimer_signals(void); 1959 1960extern NORET_TYPE void do_group_exit(int); 1961 1962extern void daemonize(const char *, ...); 1963extern int allow_signal(int); 1964extern int disallow_signal(int); 1965 1966extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *); 1967extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *); 1968struct task_struct *fork_idle(int); 1969 1970extern void set_task_comm(struct task_struct *tsk, char *from); 1971extern char *get_task_comm(char *to, struct task_struct *tsk); 1972 1973#ifdef CONFIG_SMP 1974extern unsigned long wait_task_inactive(struct task_struct *, long match_state); 1975#else 1976static inline unsigned long wait_task_inactive(struct task_struct *p, 1977 long match_state) 1978{ 1979 return 1; 1980} 1981#endif 1982 1983#define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks) 1984 1985#define for_each_process(p) \ 1986 for (p = &init_task ; (p = next_task(p)) != &init_task ; ) 1987 1988extern bool is_single_threaded(struct task_struct *); 1989 1990/* 1991 * Careful: do_each_thread/while_each_thread is a double loop so 1992 * 'break' will not work as expected - use goto instead. 1993 */ 1994#define do_each_thread(g, t) \ 1995 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do 1996 1997#define while_each_thread(g, t) \ 1998 while ((t = next_thread(t)) != g) 1999 2000/* de_thread depends on thread_group_leader not being a pid based check */ 2001#define thread_group_leader(p) (p == p->group_leader) 2002 2003/* Do to the insanities of de_thread it is possible for a process 2004 * to have the pid of the thread group leader without actually being 2005 * the thread group leader. For iteration through the pids in proc 2006 * all we care about is that we have a task with the appropriate 2007 * pid, we don't actually care if we have the right task. 2008 */ 2009static inline int has_group_leader_pid(struct task_struct *p) 2010{ 2011 return p->pid == p->tgid; 2012} 2013 2014static inline 2015int same_thread_group(struct task_struct *p1, struct task_struct *p2) 2016{ 2017 return p1->tgid == p2->tgid; 2018} 2019 2020static inline struct task_struct *next_thread(const struct task_struct *p) 2021{ 2022 return list_entry(rcu_dereference(p->thread_group.next), 2023 struct task_struct, thread_group); 2024} 2025 2026static inline int thread_group_empty(struct task_struct *p) 2027{ 2028 return list_empty(&p->thread_group); 2029} 2030 2031#define delay_group_leader(p) \ 2032 (thread_group_leader(p) && !thread_group_empty(p)) 2033 2034/* 2035 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring 2036 * subscriptions and synchronises with wait4(). Also used in procfs. Also 2037 * pins the final release of task.io_context. Also protects ->cpuset and 2038 * ->cgroup.subsys[]. 2039 * 2040 * Nests both inside and outside of read_lock(&tasklist_lock). 2041 * It must not be nested with write_lock_irq(&tasklist_lock), 2042 * neither inside nor outside. 2043 */ 2044static inline void task_lock(struct task_struct *p) 2045{ 2046 spin_lock(&p->alloc_lock); 2047} 2048 2049static inline void task_unlock(struct task_struct *p) 2050{ 2051 spin_unlock(&p->alloc_lock); 2052} 2053 2054extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk, 2055 unsigned long *flags); 2056 2057static inline void unlock_task_sighand(struct task_struct *tsk, 2058 unsigned long *flags) 2059{ 2060 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags); 2061} 2062 2063#ifndef __HAVE_THREAD_FUNCTIONS 2064 2065#define task_thread_info(task) ((struct thread_info *)(task)->stack) 2066#define task_stack_page(task) ((task)->stack) 2067 2068static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org) 2069{ 2070 *task_thread_info(p) = *task_thread_info(org); 2071 task_thread_info(p)->task = p; 2072} 2073 2074static inline unsigned long *end_of_stack(struct task_struct *p) 2075{ 2076 return (unsigned long *)(task_thread_info(p) + 1); 2077} 2078 2079#endif 2080 2081static inline int object_is_on_stack(void *obj) 2082{ 2083 void *stack = task_stack_page(current); 2084 2085 return (obj >= stack) && (obj < (stack + THREAD_SIZE)); 2086} 2087 2088extern void thread_info_cache_init(void); 2089 2090/* set thread flags in other task's structures 2091 * - see asm/thread_info.h for TIF_xxxx flags available 2092 */ 2093static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) 2094{ 2095 set_ti_thread_flag(task_thread_info(tsk), flag); 2096} 2097 2098static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2099{ 2100 clear_ti_thread_flag(task_thread_info(tsk), flag); 2101} 2102 2103static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) 2104{ 2105 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); 2106} 2107 2108static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2109{ 2110 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); 2111} 2112 2113static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) 2114{ 2115 return test_ti_thread_flag(task_thread_info(tsk), flag); 2116} 2117 2118static inline void set_tsk_need_resched(struct task_struct *tsk) 2119{ 2120 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2121} 2122 2123static inline void clear_tsk_need_resched(struct task_struct *tsk) 2124{ 2125 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2126} 2127 2128static inline int test_tsk_need_resched(struct task_struct *tsk) 2129{ 2130 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); 2131} 2132 2133static inline int signal_pending(struct task_struct *p) 2134{ 2135 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING)); 2136} 2137 2138extern int __fatal_signal_pending(struct task_struct *p); 2139 2140static inline int fatal_signal_pending(struct task_struct *p) 2141{ 2142 return signal_pending(p) && __fatal_signal_pending(p); 2143} 2144 2145static inline int signal_pending_state(long state, struct task_struct *p) 2146{ 2147 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL))) 2148 return 0; 2149 if (!signal_pending(p)) 2150 return 0; 2151 2152 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p); 2153} 2154 2155static inline int need_resched(void) 2156{ 2157 return unlikely(test_thread_flag(TIF_NEED_RESCHED)); 2158} 2159 2160/* 2161 * cond_resched() and cond_resched_lock(): latency reduction via 2162 * explicit rescheduling in places that are safe. The return 2163 * value indicates whether a reschedule was done in fact. 2164 * cond_resched_lock() will drop the spinlock before scheduling, 2165 * cond_resched_softirq() will enable bhs before scheduling. 2166 */ 2167extern int _cond_resched(void); 2168#ifdef CONFIG_PREEMPT_BKL 2169static inline int cond_resched(void) 2170{ 2171 return 0; 2172} 2173#else 2174static inline int cond_resched(void) 2175{ 2176 return _cond_resched(); 2177} 2178#endif 2179extern int cond_resched_lock(spinlock_t * lock); 2180extern int cond_resched_softirq(void); 2181static inline int cond_resched_bkl(void) 2182{ 2183 return _cond_resched(); 2184} 2185 2186/* 2187 * Does a critical section need to be broken due to another 2188 * task waiting?: (technically does not depend on CONFIG_PREEMPT, 2189 * but a general need for low latency) 2190 */ 2191static inline int spin_needbreak(spinlock_t *lock) 2192{ 2193#ifdef CONFIG_PREEMPT 2194 return spin_is_contended(lock); 2195#else 2196 return 0; 2197#endif 2198} 2199 2200/* 2201 * Thread group CPU time accounting. 2202 */ 2203void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times); 2204 2205static inline 2206void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times) 2207{ 2208 struct thread_group_cputimer *cputimer = &tsk->signal->cputimer; 2209 unsigned long flags; 2210 2211 spin_lock_irqsave(&cputimer->lock, flags); 2212 *times = cputimer->cputime; 2213 spin_unlock_irqrestore(&cputimer->lock, flags); 2214} 2215 2216static inline void thread_group_cputime_init(struct signal_struct *sig) 2217{ 2218 sig->cputimer.cputime = INIT_CPUTIME; 2219 spin_lock_init(&sig->cputimer.lock); 2220 sig->cputimer.running = 0; 2221} 2222 2223static inline void thread_group_cputime_free(struct signal_struct *sig) 2224{ 2225} 2226 2227/* 2228 * Reevaluate whether the task has signals pending delivery. 2229 * Wake the task if so. 2230 * This is required every time the blocked sigset_t changes. 2231 * callers must hold sighand->siglock. 2232 */ 2233extern void recalc_sigpending_and_wake(struct task_struct *t); 2234extern void recalc_sigpending(void); 2235 2236extern void signal_wake_up(struct task_struct *t, int resume_stopped); 2237 2238/* 2239 * Wrappers for p->thread_info->cpu access. No-op on UP. 2240 */ 2241#ifdef CONFIG_SMP 2242 2243static inline unsigned int task_cpu(const struct task_struct *p) 2244{ 2245 return task_thread_info(p)->cpu; 2246} 2247 2248extern void set_task_cpu(struct task_struct *p, unsigned int cpu); 2249 2250#else 2251 2252static inline unsigned int task_cpu(const struct task_struct *p) 2253{ 2254 return 0; 2255} 2256 2257static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 2258{ 2259} 2260 2261#endif /* CONFIG_SMP */ 2262 2263extern void arch_pick_mmap_layout(struct mm_struct *mm); 2264 2265#ifdef CONFIG_TRACING 2266extern void 2267__trace_special(void *__tr, void *__data, 2268 unsigned long arg1, unsigned long arg2, unsigned long arg3); 2269#else 2270static inline void 2271__trace_special(void *__tr, void *__data, 2272 unsigned long arg1, unsigned long arg2, unsigned long arg3) 2273{ 2274} 2275#endif 2276 2277extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask); 2278extern long sched_getaffinity(pid_t pid, struct cpumask *mask); 2279 2280extern void normalize_rt_tasks(void); 2281 2282#ifdef CONFIG_GROUP_SCHED 2283 2284extern struct task_group init_task_group; 2285#ifdef CONFIG_USER_SCHED 2286extern struct task_group root_task_group; 2287extern void set_tg_uid(struct user_struct *user); 2288#endif 2289 2290extern struct task_group *sched_create_group(struct task_group *parent); 2291extern void sched_destroy_group(struct task_group *tg); 2292extern void sched_move_task(struct task_struct *tsk); 2293#ifdef CONFIG_FAIR_GROUP_SCHED 2294extern int sched_group_set_shares(struct task_group *tg, unsigned long shares); 2295extern unsigned long sched_group_shares(struct task_group *tg); 2296#endif 2297#ifdef CONFIG_RT_GROUP_SCHED 2298extern int sched_group_set_rt_runtime(struct task_group *tg, 2299 long rt_runtime_us); 2300extern long sched_group_rt_runtime(struct task_group *tg); 2301extern int sched_group_set_rt_period(struct task_group *tg, 2302 long rt_period_us); 2303extern long sched_group_rt_period(struct task_group *tg); 2304#endif 2305#endif 2306 2307#ifdef CONFIG_TASK_XACCT 2308static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2309{ 2310 tsk->ioac.rchar += amt; 2311} 2312 2313static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2314{ 2315 tsk->ioac.wchar += amt; 2316} 2317 2318static inline void inc_syscr(struct task_struct *tsk) 2319{ 2320 tsk->ioac.syscr++; 2321} 2322 2323static inline void inc_syscw(struct task_struct *tsk) 2324{ 2325 tsk->ioac.syscw++; 2326} 2327#else 2328static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2329{ 2330} 2331 2332static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2333{ 2334} 2335 2336static inline void inc_syscr(struct task_struct *tsk) 2337{ 2338} 2339 2340static inline void inc_syscw(struct task_struct *tsk) 2341{ 2342} 2343#endif 2344 2345#ifndef TASK_SIZE_OF 2346#define TASK_SIZE_OF(tsk) TASK_SIZE 2347#endif 2348 2349#ifdef CONFIG_MM_OWNER 2350extern void mm_update_next_owner(struct mm_struct *mm); 2351extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p); 2352#else 2353static inline void mm_update_next_owner(struct mm_struct *mm) 2354{ 2355} 2356 2357static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p) 2358{ 2359} 2360#endif /* CONFIG_MM_OWNER */ 2361 2362#define TASK_STATE_TO_CHAR_STR "RSDTtZX" 2363 2364#endif /* __KERNEL__ */ 2365 2366#endif 2367